Vol. 44 No. 2 October 2009 Design construction and ...

Vol. 44

No. 2

October 2009

Contents/Inhalt A. O. Akinro, A. A. Olufayo and P. G. Oguntunde

J. T. Fasinmirin

Design construction and performance evaluation of a drip irrigation system for plantain (Musa, spp. AAB) production in a tropical environment of Western Nigeria

159

Development and calibration of a digital recording system for automation of runoff measurement

177

A. Montazar

Artificial neural network modeling to predict wheat yield production and water productivity 187

O. Yahaya, A. A. Olufayo, A. O. Akinola and O. Olubbenga

Measurement of surface runoff and sediment yield at varying gradients using an automatic runoff-meter

209

A. O. Akinro, I. B. Ologunagba and O. Yahaya

Environmental implications of unhygienic operation of a city abattoir in Akure, Western Nigeria

223

Bacteriological and physio-chemical analysis of some domestic water wells in peri-urban areas of Akure, Nigeria

231

Potential to enhance the extent of paddy cultivation using domestic and municipal wastewater harvestimg - a case study from the dry zone of Sri Lanka

239

A.O, Akinro and I. O. Ologunagba

U. S. C. Udagedara and M. M. M. Najim

O. Yahaya, A. O. Akinro, O. M. Kehinde and I. B. Ologunagba

Evaluation of water poverty index in Ondo-State, Nigeria 249

Zeitschrift für Bewässerungswirtschaft, 44. Jahrg., Heft 2 /2009, ISSN 0049-8602

Seiten 159 - 176

Design, construction and performance evaluation of a drip irrigation system for plantain (Musa, spp. AAB ) production in a tropical environment of Western Nigeria A. O. Akinro, A. A. Olufayo and P. G. Oguntunde

Keywords Drip irrigation, emitter, lateral lines, emission uniformity, musa Abstract There is need for an affordable small scale irrigation system in local farmsteads in Nigeria today to help farmers increase their potential for maximum productivity. Consequently, an attempt was made to design and construct a gravitational drip irrigation system for plantain production. The system was used for two cropping seasons of 2006-2007 and 2007-2008. The system includes a main line and sub-mains made up of PVC pressure pipes with diameter 27mm and 19mm respectively. The drippers dissipate water from the laterals through the use of long path tube (6mm diameter tubing). Water from the overhead tank was supplied to the field by gravity, hence the system require no energy source for pumping to the field except that required to fill the reservoir from the shallow well. The system was tested for efficiency. The designed water requirement for plantain was computed as 7.85 mmday-1. The average discharge value of the emitter was 3.82 lh-1 which is closer to the designed discharge value of the emitters (4 lh-1). The emission uniformity coefficient of the emitters was 84.2% which is high and considered suitable for the irrigation of the crop under study and acceptable for humid climate (JAMES, 1988). For both seasons, plantain biomass yields were significantly different (p 0, then G = 0.1 Rn else G = 0.5 Rn Hence G = 0.1 x 1.78 = 0.178 mmh-1 (ii) Estimation of the vapour pressure deficit (es - ea) es • 0.61078exp(

17.269T ) 237.3 † T

(4)

es at 27.86oC is computed as • 17.269 € 27.86 Š = 3.75 kPa es • 0.61078 exp‹ ˆ Œ 237.3 † 27.86 ‰

Hence, es ‡ ea • es „1 ‡ RH … = 3.75 ( 1 – 0.647 ) = 1.32375 kPa (iii) Calculate slope of saturation vapour pressure curve at air temperature (kPaoC-1), ‚ for T = 27.86oC T • 17.269 Š• Š ‚ • es ‹ ˆ‹1 ‡ ˆ Œ 237.3 † T ‰Œ 237.3 † T ‰

(5)

27.86 Š • 17.269 Š• O ‡1 Ž ‚ • 3.75‹ ˆ • 0.0952kPa C ˆ‹1 ‡ Œ 237.3 † 27.86 ‰Œ 237.3 † 27.86 ‰

(iv) The latent heat of vaporization, L in Jkg-1 was approximated as L

= 2.5005 x 106 - 2.359 x 103 (Ta + 273.15)

Hence, L = 2.5005 x 106 - 2.359 x 103 ( 27.86 + 273.15) = 1,790,417.41 Jkg-1 Heat Capacity of air Cp in Jkg-1 was estimated thus C p • 1004.7 (

0.522ea † 1 ) P

(6)

Zeitschrift für Bewässerungswirtschaft, 44. Jahrg., Heft 2 /2009

165

ea = esRH = 3.75 (0.647) = 2.42625 kPa » C p • 1004.7•‹ 0.522 € 2.42625 † 1Šˆ Œ

P

‰

(v) Computation of P in kPa was obtained from the equation: P • 101.3(

293 ‡ 0.0065A 5.26 ) 293

(7)

The altitude A of the experimental site was 351m (measured). Thus • 293 ‡ 0.0065 € 351 Š P • 101.3‹ ˆ 293 Œ ‰

5. 26

• 97.22kPa

Hence • 0.522 € 2.42625 Š ‡1 C p • 1004.7‹ ˆ † 1 • 1017.78JkgK 97.22 Œ ‰

(vi) Hence, to calculate ETo: Substituting the above stated results into equation 3 yields: 37 • Š 0.408 € 0.0952„1.78 ‡ 0.178…0.08898‹ ˆ0.49„1.32375… 27 . 86 † 273 . 2 Œ ‰ ETo • • 3.49 €10 ‡1 mmh ‡1 0.0952 † 0.08898„1 † 0.34 € 0.49…

• 8.35mmday ‡1 (vii) Determination of the peak evapotranspiration rate The Peak evapotranspiration was thus determined from equation 2 as ETpeak • ETo €

P 85

ET peak • 8.35 €

80 • 7.85mmday ‡1 85

166

Journal of Applied Irrigation Science, Vol. 44, No. 2 /2009

3.1 Emitter selection The point source emitter operating at low pressure head (3 m) and small discharge was used for irrigation of plantain. There was one emitter per plant. Since there were 16 plots overall with 6 plants in a plot, hence 6 x 16 (i.e. 96) emitters were used on the experimental field. The capacity of tank (reservoir) supplying water to the field was 550 liters. Time required for discharging 550 liters capacity reservoir was computed as T•

Re servoir • capacity Emitter • disch arg e € No • of • emitters

T•

550 • 1.432hrs • 1hr 25 min s 4l / hr € 96

(8)

There were four treatments as shown in Table 1. It thus implied that water distribution to the treatment plots varied in the ratio 1:0, 5:0, 25:0 according to the water treatment. Hence for 550liters capacity tank, the time taken to irrigate per treatment is shown in Table 2. Table 2: Times to irrigate with 550 liters capacity tank Treatment High (T100) Moderate (T50) Low (T25 ) Control (T0 ) Total

Time taken to irrigate in minutes 49 24 12 0 85

The rate at which emitter discharged water depended on the hours of operation. Assuming the irrigation system was operated for 3hrs, then the emitter discharge per plant would be this implied that the lateral would deliver in a plot consisting of 6 plants. Since drippers discharged at approximately equal rate based on field experimental test conducted on the system (Appendix A), gate valves were installed on entry point to control the inflow into the various treatment plots. Stop watch was used to control the time of delivery. Water was always pumped from the well to refill the tank after the full load from the tank has been fully discharged to the field. The result of the experiment conducted on the drippers is shown in Appendix A.


Plate 1: Lateral with gate valve controls filter design

167

Plate 2: Gate valve under control

Atlas filter (manufactured by atlas filter, Italy) was purchased and installed in between the tank and the sub mains to remove impurities and debris which could clog the emitters and thus cause poor water distribution along the drip laterals or impede their discharges. This has been found to affect plant growth.

Plate 3: Tank installation

Plate 3: Filter installation

168


Emission uniformity design The emission uniformity of water application was computed by the equation: —‘ 1.27 ”‘ Qmin Eu • 100–1.0 ‡ Cv “ N c ‘’ Qave ‘•

(9)

(Agricultural Engineers Year Book, 1981) where Eu Nc Cv Qmin Q ave

= = = = =

design emission uniformity in % number of point source emitters per emission point the manufacturers coefficient of variation for the point source emitters minimum emitter discharge rte (lh-1) average or design emitter discharge rate (lh-1)

Hence, Eu • 100—–1.0 ‡ 1.27 „0.03…”“ 3.5 • 84.2 1

•

’ 4

3.2 Determination of pressure variation along the system The following procedure was adopted to evaluate the operating pressure, Ho for the system. Calculation of the emitter head loss, He The discharge equation for the emitter is Qe • 1.41P 0.5

(JAMES, 1988)

(10)

where Qe = designed emitter discharge (lh-1) = 4 lh-1 P = operating pressure • 4 Š hence P • H e • ‹ ˆ Œ 1.41 ‰

1

0.5

• 8.05m

The various head losses by friction for this set were also calculated as follows: Energy drop by friction at the Mains, Hm Using Williams and Hazen’s equation and taking C = 150 The total energy drop by friction at the main was computed from the relations • Q1.852 Š H m • 15.27‹‹ 4.871 ˆˆ € L ŒD ‰

(MICHAEL and OJHA, 2003)

(11)


where Hm Q D L

= = = =

169

energy drop by friction at the main (m) Total discharge at the main line pipe (l/sec) diameter of the main line (cm) = 2.7cm length of the pipe (m) = 92m

Total • disch arg e • disch arg e / emitter € no • of • emitters • installed

i.e. Total • disch arg e • 4 € 96 • 384lh ‡1 • 6.41l min ‡1 • 0.1071ls ‡1 • 0.1071.852 Š ˆ € 92 • 1.77m H m • 15.27 € ‹‹ 4.871 ˆ Œ 2.7 ‰ Energy drop by friction at the lateral The total energy drop by friction for lateral was computed from the relations Hence

• Q1.852 Š (MICHAEL and OJHA, 2003) (12) H LD • 5.35‹‹ 4.871 ˆˆ € L ŒD ‰ where HLD = energy drop by friction at the lateral (m) Q = Total discharge along the lateral line pipe (ls-1) = 3 x 4 lh-1. = 0.0033 ls-1 D = diameter of the lateral line (cm) = 1.9 cm L = length of the pipe (m) = 6 m • 0.00331.852 H LD • 5.35 € ‹‹ 4 .871 Œ 1.9

Š ˆˆ € 6 • 0.0000357m ‰

But ‚ZL = HI and HI= FHLD + Mi

(13)

where Mi = minor losses through the fittings. (it is zero since there is no fittings along the pipeline) F is a constant = 0.33 (JAMES, 1988) hence, HI • 0.33 € 0.0000357 † 0 • 1.179 €10 ‡ 6 m and ‚Z m • 0.33H m • 0.33 €1.77 • 0.584m Pressure head at the filter 0.5

Q f • 1.41Pf

(14)

170


• Qf Š ˆˆ Ž Pf • ‹‹ Œ 1.41 ‰

2

Volume of tank = 550 liters Time taken to discharge 550 liters = 3 hrs Hence, Q f • 550 / 3 • 183lh ‡1 • 0.0509ls ‡1 2

• 0.0509 Š Ž Pf • ‹ ˆ • 0.00130 m Œ 1.41 ‰

Average lateral pressure, Ha H a • H e † 1 / 4 H LD † 1 / 2HI

(15)

i.e • H a • 8.05 † 1 / 4„0.0000357… † 1 / 2„1.179 €10‡6 … • 8.05m

Lateral inlet pressure HLA H LA • H a † 3 / 4 H LD † 1 / 2 HI

(16)

„

H LA • 8.05 † 3 / 4„0.0000357… † 1 / 2 1.179 ˜ 10

‡6

… • 8.05m

Mainline or manifold pressure Hm H m • H LA † H m † ‚Z m / / 2 • 8.05 † 1.77 † 1 / 2„0.584… • 10.11m Operating pressure Ho H o • H m † Pf † H LA † H e • 10.11 † 0.00130 † 8.05 † 8.05 • 26.21m Pressure head variation, Hvar was then computed thus H var • (

H max ‡ H min ) H max

(17)

And the discharge variation as qvar • (

qmax ‡ qmin ) qmax

From Table 9 qmax = 4.16 lhr-1 and qmin = 3.50 lhr-1

(18)


171

hence, qvar • ( 4.16 ‡ 3.50 ) • 0.159 • 16% 4.16

The pressure variation and the emitter flow or lateral line flow are related and can be expressed as qvar • 1 ‡ (1 ‡ H var )0.5

(MICHAEL and OJHA, 2003)

(19)

Hvar was computed from equation 34 as (1 ‡ H var )0.5 • 1 ‡ 0.159 • 0.841 and H var • (1 ‡ 0.841) 2 • 0.293 • 29.3% but H var • (

H max ‡ H min ) H max

H max • H o • 26.21m 0.293 •

26.21 ‡ H min 26.21

hence, Hmin • 18.53m 4. Summary and discussions The drip irrigation system was designed, constructed and tested and the result of the test and other calculations are shown in Appendix A. The average time for the first round of test was 5 min 9 s while rounds 2, 3 and 4 produced 4min 19 s, 4min 44 s and 4 min 42 s respectively. The average discharges per emitter were 3.5 lh-1, 4.16 lh-1, 3.82 lh-1 and 3.81 lh-1 for test 1, 2, 3 and 4 in that order. The minimum discharge value of the emitters was 3.5 lh-1 while the maximum was 4.16 lh-1. The average discharge value of the emitter was 3.82 lh-1 which is closer to the designed discharge value of the emitters (4 lh-1). The emission uniformity coefficient of the emitters was 84.2% which is high and considered suitable for the irrigation of the crop under study and acceptable for humid climate (JAMES, 1988). The peak evapotranspiration ETpeak computed for the crop was 7.85 mmday-1. This is acceptable and within the range for plantain production (GOENAGA and IRIZARRY, 2000; ROBINSON and ALBERTS, 1989). The designed water requirement was 7.85 mmday-1. For a plot with six plants, the water requirement was 0.00785 m/day X 12 m2 (size of a plot) = 0.0942 m3day-1 and which translated to 3.93 lh-1. Analysis of pressure head variation showed that the discharge variation was 16%

172


while the pressure variation, was 29.3% According to MICHAEL and OJHA (2003), drip irrigation design criterion is usually based on an emitter flow variation less than 20% or pressure variation of less than 40%. The design satisfied both conditions. Effects of drip irrigation on biomass yield for each treatment are shown in Table 3. Estimated water consumed ranged from 900 mm to 1700 mm from planting to harvest in the order of T0, T25, T50 and T100 treatments respectively. For example, in the fully irrigated treatment (T100), crop consumptive use at 413DAP (at harvest) was 1691.5 mm while crop consumptive use was 910.7 mm at same period for treatment T0. Correspondingly, highest biomass yield was 23.2tha-1 at harvest for T100 treatment while lowest value of biomass yield was 8.3 tha-1 in T0 treatment. The trend was the same during the 2007-2008 season. This confirms that supplemental irrigation through the use of drip irrigation system had significant effect (p 0.94 in both cases. The quadratic calibration curve showed a better fit with R2 > 0.98. All the models are highly significant at P < 0.0001. Although linear fit showed high accuracy with their R2 not statistically different from that of the quadratic model, the values of standard error (0.150 mm) is higher making it less acceptable to the non-linear equation which has SE of 0.079 mm. Linear model starting from the origin is theoretically favoured since no negative runoff depth was observed. Given the foregoing, calibration curve based of the quadratic model was adjudged best and then adopted for this electronic water level sensor. The performance was further validated by measuring runoff simultaneously with the sensor and manually for the months of March and April. Runoff obtained from digital sensor conversion using the quadratic calibration equation were very similar to values from manual measurement as shown Figures 4.5a and 4.5b, for March and April, respectively. The variation between both the manual and digital datasets was quite small. MBE for the month of March and April were 0.19 mm and 0.15 mm, respectively and the correlation coefficients were very high (R > 0.97) for both datasets.

183


Table 1: Summary of calibration statistics S/N

#

#

Model

Overall Statistics

R2

Standard error

Sig. level

1

Linear

0,9448

0,150

< 0.0001

2

Quadratic

0,9869

0,079

< 0.0001

R2 is coefficient of determination; SE is standard error of estimates

45000 Runoff depth (Manual, mm)

40000 35000 30000

y = 2504.2x - 17693

25000

R2 = 0.9448

20000 15000 10000 5000 0 10

12

14

16

18

20

22

Runoff depth (Sensor, mm)

45000

26

a

y = -138.31x 2 + 7277.6x - 56594

40000 Runoff depth (Manual, mm)

24

R2 = 0.9869

35000 30000 25000 20000 15000 10000 5000 0 10

12

14

16

18

Runoff depth (Sensor, mm)

20

22

24

26

b

Figure 3: Calibration curves for digital sensor (a) Linear fitted (solid line) and (b) Quadratic fitted

184


5. Conclusion The design analysis, and construction of digital sensing devices for runoff measurement has been presented. It consist of runoff catchment’s box (basin), runoff storage tank ((RST), discharge pipe, digital devices such as ICL7106 analog to digital converter, resistor, liquid crystal display (LCD), sensor (floater), Pressure hose, 6F 2.29V battery, among others. All the materials used for the fabrication were sourced locally. The calibration is done such that the initial reading was 11.2 litres (minimum) and 96 litres (maximum) reading for accurate measurement. The use of digital ICL7106 analog to digital converter for measuring the amount of runoff at all period of time brings together an accuracy, versatility and economical method of estimating runoff. Notations RST LCD pA µA µV mW IC C R Ri Vr Vin &! KHz µF pF mV

Meanings Runoff Storage Tank Liquid Crystal Display picoAmpere(s) microAmpere(s) microVolt milliwatts Integrated Circuits Capacitance Resistance Internal Resistance Reference Voltage Input Voltage ohm kilohertz microFarad picoFarad milliVolt

Reference Arilesere, F., 1996: Erosion as the major cause of land degradation. -Ultimate water technology and environment Vol.1, issue 1 pp 5-.7 nd Bailey, R.G., 1995: Description of the Eco-regions of the United State2 Edition. - United State Department of agriculture, forest Services, Washington DC. Beasley, R. P., 1992: Erosion and sediment pollution control. - Iowa State University press, Iowa, USA. Cox, S. W. R., 1988: Farm Electronics, BSP Professional Books. - A division of Blackwell Scientific Publications Ltd. Beacon Street, Boston Massachesetts 02108, USA. pp 171-174.


185

Duley, F. L., 1986: The loss of soluble salts runoff water. - Soil Science Vol.21.pp 401-409. Foth, H. D., 1990: Fundamental of Soil Science. Eight Editions. - John Wiley & Son Inc. New York, NY. Hammer, M. J. and K. A. Machinechk, 1980: Hydrology and Quality of Water Resources. John Wiley and Sons. New York. Kowel, J., 1992: The hydrology of a small catchment’s basin of Samaru, Nigeria: Assessment of soil erosion under varied hand management and vegetation cover. Nigerian Agricultural Journal. Vol. 7. pq (134-138). Lal, R., 1986: Soil Erosion Problems on an Alfosol in Western Nigeria and their Control. - III A Monograph No. 1, pp. 4-10. Ozara, A., 1991: Management and Control of Erosion Process Institute of Erosion Studies Hand book. - Federal University of Technology Owerri.Vol.1, pp 100-101. Roger, H. T., 1984: Plant nutrient losses from a corn wheat clover rotation on dune more siltsleans soil. - Soil Science Soc. Amer. Proc. 6 pp 263-265. Sangodoyin, A. U. and E. O. Nwosu, 1995: Time variation and soil loss with soil physical properties in southeastern Nigeria. - Journal of Agricultural Engineering Technology Vol. 3, pp 66-69. Smith, N. J. H., 1981: Colonization Lessons from a Tropical Forest Science Vol. 214, pp.753761.

Address of Author: J. T. Fasinmirin Department of Agricultural Engineering Federal University of Technology Akure, Ondo State, Nigeria E-mail: [email protected]


Seiten 187 - 208

Artificial neural network modeling to predict wheat yield production and water productivity A. Montazar

Keywords Artificial neural networks, nitrogen, seasonal water use, weather data, water productivity, wheat Abstract Decision-making processes in agriculture often require reliable crop response models to assess the impact of specific water and nutrient management options and weather data. The objective of this study was to develop winter wheat yields and water productivity prediction artificial neural networks (ANNs) models with readily available weather data and seasonal water-nitrogen variables. The results of a three years field experiment in a semi-arid region in Iran were used to develop the models. The applied ANNs models had desirable accuracy in estimating crop yields and water productivity. The ANN models proved to be a superior methodology for accurately predicting these parameters in the study area for the specific soil types used to develop the models. Adjusting ANN parameters such as learning rate and number of hidden nodes affected the accuracy of the models predictions. ANN models consistently produced more accurate predictions than multiple linear regression (MLR) models. The high potential usefulness of proposed ANNs was confirmed and hence, they could be used as an accurate planning tool to optimize wheat water productivity with the consideration of the climatic data and different management practices of water and nitrogen in the study area. The data generated here suggest that maximum wheat water productivity would be achieved when 90 kg N ha”1 is combined with about 270 mm seasonal water use, which was determined 1.51 kg m-3.

1. Introduction Wheat (Triticum aestivum L.) is the most important crop in Iran grown on 6.5 million ha of the total national cultivated land; irrigated wheat farms accounting for 35% of the total wheat lands. Wheat yield varies greatly among years, depending mainly on weather but also on improvements in genetics and management. Maximum wheat grain yields are influenced by a number of factors, including N fertility, growing season conditions, water availability, and soil conditions. Yield prediction requires a combination of weather and technology factors, including air temperature and precipitation through the growing season and time trend (STEWART et al., 1998). Methods used to predict crop yield include regression, simulation, expert systems, and artificial neural networks (ANNs). Agronomic models are based on mechanistic or empirical approaches (POLUEKTOV and TOPAJ, 2001). Although these models

188


are suitable for areas outside the data range used for development, they tend to be complex and require many input parameters (BASSO et al., 2001; WANG et al., 2002). Empirical models attempt to determine functional relationships between crop yield and soil–land management factors using either an existing or a specially designed agronomic experiment. Regression or correlation analyses are generally used to characterize the statistical relationship between controlled variables and crop yield. Technologically, empirical crop growth models are relatively simple to build or develop, but these models can not take account of temporal changes in crop yields without long-term field experiments (JAME and CUTFORTH, 1996). Furthermore, the derived functional equation is locally specific, and it is thus difficult to extrapolate to other areas unless environmental conditions are similar. ANNs models are a powerful empirical modeling approach and relatively simple compared to mechanistic models. These models find relationships by observing a large number of input and output examples to develop a formula that can be used for predictions (PACHEPSKY et al., 1996). Although in early studies ANNs were mostly used to classify data, the approach has also shown a great potential for predicting continuous variables (ATKINSON and TATNALL, 1997; KIMES et al., 1998). ANNs were first used to predict regional crop yields based on weather data in the early 1990s (BAKER, 1991). Successful applications have already been reported for surface water quality assessment (GROSS et al., 1999; ZHANG et al., 2002), soil moisture estimation (CHANG and ISLAM, 2000; DEL FRATE et al., 2003), biomass estimation (JIN and LIU, 1997), and yield prediction (LIU et al., 2001; DRUMMOND et al., 2003). Agronomic ANNs applications include crop development modeling (ELIZONDO et al., 1994), pesticide and nutrient loss assessments (YANG et al., 1997), soil–water retention estimations (SCHAAP and BOUTEN, 1996), and disease prediction (BATCHELOR et al., 1997). In these researches, the potential of ANNs for the development of in season yield mapping was examined. The high potential usefulness of ANNs was confirmed. Investigations of the possibility of predicting crop response under varying soil and land management conditions were also performed by applying general linear models (GLMs), regression trees (RTs), and ANNs. GREEN et al. (2007) used spatial analysis Neural Networks (SANN) to relate wheat grain yield to topographic attributes. This study demonstrated the utility of SANN with topographic attributes that contain implicit soil and water information for estimating spatial patterns of dry-land wheat yield. SAFA et al. (2003) analyzed the capability of ANNs models for estimating winter wheat yield. The models were developed based on 11 weather parameters as input data. The applied ANN models predicted wheat yield with an accuracy of 45 to 60 kg ha-1 one month before crop harvesting. KAUL et al. (2005) investigated the corn and soybean yields prediction capabilities of ANNs models. The models proved to be a superior methodology for accurately predicting corn and soybean yields under typical Maryland climatic


189

conditions. According to BATCHELOR et al. (1997), ANNs produced better results than traditional statistical methods when predicting soybean rust. GUTIERREZ et al. (2008) determine the potential of evolutionary product unit neural networks (EPUNNs) to analyze multispectral imagery, weed and elevation data for predicting sunflower yield in early growth stage. The results obtained using different EPUNN models showed that the functional model and the hybrid algorithms proposed provide very accurate prediction compared to other statistical methodologies used to solve that regression problem. The literature indicates that agricultural use of seasonal climate prediction remains a new and developing technology. Impact of climatic variability, mainly the variations in the winter rains and abrupt changes in the temperatures during critical growth stages, are important in the yields production. Hence, there is a need to collate the information of growth response behaviour for winter wheat, so that the productivity can be enhanced either by breaking the yield barriers through evolving the adopting suitable resource and agronomic management practices. The main objective of this work was to develop simple and practical crop yield and water productivity prediction models with readily available weather data and seasonal water-nitrogen variables. The specific objectives were to: (1) assess the potential of ANNs models for the development of field-scale yield estimation systems for winter wheat; (2) compare the effectiveness of multiple linear regression models with ANN models for predicting wheat yields and water productivity.

2. Materials and methods 2.1 Study area and field experimental data Field experiments were conducted at the Aboureyhan Agricultural College Research Farm, Pakdasht, Iran (33°28´N, longitude of 50°58´E with an altitude of 1180m above mean sea level). It is located at the 20 km east of Tehran, a arid region in Iran (Figure 1). The annual mean air temperature is about 16.9 oC, with the mean maximum and minimum temperatures are 40.5 and -6.7 oC, occurring in the month of August and January, respectively. Mean annual precipitation is about 164 mm with more than 90% falling in the period from October to the next June. Winter wheat grows mainly in this period. The soil at the experiment site was silt loam, with the bulk density of about 1.39 g cm-3. Soil characteristics referring to five genetic horizons selected from the soil survey are presented in Table 1. Soil pH ranged between 7.5 and 7.8, its sodium absorption rate (SAR) between 1.4 to 1.8, and its electric conductivity (ECe) was from 1.6 to 2.2 ds m-1. Field trials were conducted in during three irrigation seasons (2001-2002, 20022003, and 2007-2008). One winter wheat cultivar, Pishtaz, widely adopted by farmers in the study area, was used. The bread wheat cultivar, adapted to the arid and semi-

190


arid environments was planted on 5 to 10 November at a rate of 200 kg seeds/ha. Wheat seeds were sown manually in rows 20 cm apart. Each elementary plot was 5m × 4m, and was separated from adjacent plots within the replicates by 50 cm in addition to 30 cm bund. The outer two rows were not harvested to eliminate border/edge effects. Thus, the effective width of separation between neighboring plots was 120 cm. The replicates were separated from each other by 160 cm blank space. The source of the irrigation water was groundwater with a good quality (pH: 7.6; EC: 1.1 dS m-1; SAR: 1.3).

Experimental field

Figure 1: Location of study area in Iran Table 1: Some physical properties of the soil used in the experiment Genetic horizon (cm) 0-25 25-35 35-57 57-82 82-100

Layer thickness (cm) 25 10 22 25 18

Bulk density (g cm-3) 1.39 1.42 1.32 1.41 1.27

Porosity (%) 43.2 41.4 46.7 42.5 50.2

Field capacity (%) 28.3 25.8 29.4 21.9 33.4

Wilting point (%) 11.5 11.1 12.5 11.2 14.2

Sand (%)

Silt (%)

Clay (%)

20.2 34.2 36.2 27.2 26.2

74.7 60.6 58.5 65.2 66.5

5.1 5.2 5.3 7.6 7.3


191

The experimental design was a randomized complete block (RCB) with four replicates. There were 4 levels of applied water and 4 levels of N application as treatments in the proposed plots (64 replications in 16 different treatments in each season). The applied water treatments included equivalent irrigation depth of winter wheat water demand, and three levels of providing 60%, 80%, and 120% of water demand depth. The N treatments were equivalent crop nitrogen demand, and three levels of N application 60%, 80%, and 120% of crop nitrogen demand with respect to soil type. In the study area, the seedling stage was completed at the end of November. December, January, and February were the long winter season. In March, winter wheat began to turn green. The jointing stage was after the beginning of April. The booting stage generally started from the middle of April and the flowering stage at the end of April and beginning of May. Grain filling followed the end of the flowering stage and ends at the beginning of June. Grain and straw yields data were collected from ten randomly selected plants from each plot. The crop was harvested manually. Daily rainfall was measured by a rain gauge in a weather station located in the experimental station. All the meteorological data needed for the calculation of the soil water content, and potential evapotranspiration were derived from measurements of air temperature, relative humidity, wind speed, sunshine hour and solar radiation at the weather station located in center of the experimental field (2001/02, and 2002/ 03) and at distance of 1.0 km from the plots (2007/08). Potential evapotranspiration was calculated by the Penman€Monteith approach. The actual crop evapotranspiration or seasonal water use (SET) was estimated by water balance equation as follows (ALLEN et al., 1998): ET=P + I • ‚ ‚

(1)

where P (mm) is the precipitation, ‚ ‚ the change in water storage (mm) in the soil profile, and I the irrigation water applied (mm). Other components of soil water balance, such as capillary rising, deep percolation, and surface runoff were ignored. Soil water content was measured gravimetrically in the depth range of 0€20 and 2040 cm and with a TDR-probe (moisture point MP-917 with a 15 cm, two-rod, singlediode probe) from 40 to 100 cm depth at an interval of 20 cm on the before and one day after water application. 2.2 Artificial Neural Network model development A feed-forward back-propagating Artificial Neural Network (ANN) structure with one-input layer, hidden layer and output layer as illustrated in Figure 2 was applied to develop winter wheat yields and water productivity prediction models using a standard neural net software. Briefly, this involved: (1) selecting training and validation subsets, (2) analyzing and transforming data, (3) selecting variables, (4)

192


network construction and training, and (5) model verification. For the criterion, all the data were divided into three sets (Coulibaly et al. 2000). The first set is the training set for determining the weights and biases of the network. The second set is the validation set for evaluating the weights and biases and for deciding when to stop training. The last data set is for validating the weights and biases to verify the effectiveness of the stopping criterion and to estimate the expected network operation on new data sets.

Hidden layer

Observed output

Predicted output

Input layer

Output layer

Data Flow

Figure 2: Layers and connections of a feed-forward back propagating ANN The various input variables and converts them into linear, binary, or one-of-N coding, were analyzed as appropriate. Then, the data were scaled and various transforms tested and applied to make the distribution of inputs and the output more similar. Next, variables that contain little or redundant information relative to the problem were eliminated. The software uses an optimization algorithm to search different combinations of inputs for the best variable set to use. Then a computer simulation of biological neuron layers was created. This involves an input layer (each neuron represents a piece of known data) with weighted connections to a hidden layer (each neuron represents an interaction between inputs) then to an output layer (each neuron represents a possible result). The ANN was trained by supplying input information where the corresponding outputs are known. The ANN adjusted itself (by varying the weighted connections) until the predicted output agrees with the known output.


193

The ANN with a three layer dynamic preceptron structure and supervised training methods were applied. Network training was done using the Levenberg–Marquardt (LM) algorithm. The LM algorithm was used with an early stopping criterion to improve the network training speed and efficiency. The basis of this method is to minimize the error function based on network parameters and their convergence speed. The nature of the activation functions used for back propagation networks makes it acceptable to use inputs between –1 and +1 or between 0 and 1. The sigmoid function was applied as activation functions in the form of f ( x) • 1 (1 † e‡ x )

(2)

respectively, where x is a node value. Based on this equation, f(x) has thresholds of 0 and 1; it can never be 0 unless t is equal to -and can never be 1 unless t is equal to infinity. Thus, the extreme values of the permissible range are unreachable; in particular, the values of the derivatives used for the learning function are 0 at these points (CAUDILL, 1990). Values are thus further restricted in many instances to the range 0.1–0.9 on the output node, to avoid such thresholds, to allow the activation function to go above or below the target during training, and to improve the performance of the weight adjustment algorithm. This requires input data to be transformed or coded to fit within these limits. Coding was done using the following equation: • x ‡x xn • 0.5‹‹ 0 ave Œ xmax ‡ xmin

Š ˆˆ † 0.5 ‰

(3)

where xn is the coded input value, x0 is an element of pre-processed data, and xmin, xmax and xave are the minimum, maximum and mean value in the pre-processed data set, respectively. To determine the optimum structure of each network, first the functions and training algorithms were chosen based on the input and output parameters and networks were designed with numbers of 1 to 30 nodes within the hidden nodes. The learning rate was adjusted between 0.77 and 0.90. Preliminary trials indicated that lower learning rates produced poorly developed models. During early trials, the training tolerance was set at 0.1. Better results were found when the training tolerance was initially set higher and decreased linearly as the network trained. However, learning rate, learning momentum, and hidden node number were varied over a range of values, so that combinations of network parameters could be chosen that minimized error.

194


2.3 Input and output variables The input variables were a compilation of climatic factors include: minimum and maximum air temperatures throughout the months of crop growth consisting of November, December, January, February, March, April, May, and partly June, the mean air temperature, air relative humidity, and solar radiation of each of the eight vital months, and management factors include: seasonal water use (sum of irrigation water and effective rainfall) and seasonal nitrogen use (sum of applied nitrogen during growth season and soil residual nitrogen). Therefore, the structure of ANN models were designed in the form of various mixtures of 43 input parameters (Table 2). The output variables were grain yield, straw yield, and water productivity at scale of the farm. Water productivity was considered as the ratio of grain yield to seasonal water use. 2.4 Performance analysis A multi-criteria approach was adopted for assessing the models developed, in which model performance was evaluated using several statistical error and goodness-of-fit measures, including the coefficient of determination (r2), root mean square error (RMSE), mean bias error (MBA), and standard error (STE): 2

n n Ÿ n œ • ( yo ( i ) ‡ (1 / n) yo (i ) )( y p ( i ) ‡ (1 / n) y p ( i ) ) š i •1 i •1 i •1 ž › 2 r • n n n n Ÿ œ Ÿ 2œ 2 • ( yo ( i ) ‡ (1 / n) yo (i ) ) š • ( y p ( i ) ‡ (1 / n) y p ( i ) ) š ž i •1 › ž i •1 › i •1 i •1

(4)

n

( y p ( i ) ‡ yo ( i ) ) i •1

RMSE •

(5)

n n

( y p ( i ) ‡ yo ( i ) ) MBE •

(6)

i •1 n

y p (i ) i •1

2 Ÿ Ÿn œ œ ( yo ( i ) ‡ yo )( y p ( i ) ‡ y p )š š •n • 1 • › š STE • ( y ‡ y p ) 2 ‡ ž i •1 n š n ‡ 2 • i •1 p (i ) 2 ( yo (i ) ‡ yo ) • š i •1 ž• ›š

(7)

where n is the number of data pairs, yp(i) and yo(i) are the predicted and observed variable values for the ith data pair, respectively, andare the mean predicted and observed variable values, respectively.


195

Table 2: Inputs used for development of wheat yields and water productivity prediction models Input number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

Description Seasonal water use Seasonal nitrogen use November mean air temperature December mean air temperature January mean air temperature February mean air temperature March mean air temperature April mean air temperature May mean air temperature 1-10 Jun mean air temperature November maximum air temperature December maximum air temperature January maximum air temperature February maximum air temperature March maximum air temperature April maximum air temperature May maximum air temperature 1-10 Jun maximum air temperature November minimum air temperature December minimum air temperature January minimum air temperature February minimum air temperature March minimum air temperature April minimum air temperature May minimum air temperature 1-10 Jun minimum air temperature November mean relative humidity December mean relative humidity January mean relative humidity February mean relative humidity March mean relative humidity April mean relative humidity May mean relative humidity 1-10 Jun mean relative humidity November mean solar radiation December mean solar radiation January mean solar radiation February mean solar radiation March mean solar radiation April mean solar radiation May mean solar radiation 1-10 Jun mean solar radiation

196


In addition, the performance of each model was evaluated by plotting the simulated values against the measured values and by testing the statistical significance of regression parameters. However, ten ANN models were developed and evaluated for each of wheat grain and straw yields and water productivity indices. 2.5 Regression models Multiple linear regression (MLR) models were developed and tested with the same data sets used for ANN development, thus making the results comparable, and are referred to as validated models. Field-specific climate data and management factors were independent variables and crop yields (grain and straw) and water productivity were the dependent variables (Table 2). The independent and dependent variables correspond to ANN input and output variables, respectively. Raw output variables were scaled to range from 0 to 1 so that values were within a similar numerical range as the other input variables. The validated models are indicative of the models capability to predict output variables, since the testing data are independent of the data used for model development. Specific comparisons were based on RMSE and r2 of the validated MLR model results and the ANNs model results.

3. Results and discussion 3.1 Yield prediction The results obtained with the different nets were evaluated in several ways. Firstly, a regression was performed between the observed and simulated yields for both training and validation data sets. The regression parameters for wheat grain and straw yields are given in Table 3. Ideally, the intercept should be 0 and the slope should be 1 for a perfect match. For all ANNs, the regression parameters are significantly close to the ideal values (Pd”0.05), which means that the nets performed very well. However, the regression parameters were closer to the ideal values for the ANN-10G (with an intercept and slope of 118 and 0.93 for validation data sets, respectively) and ANN10S (with an intercept and slope of 1150 and 0.77 for validation data sets, respectively) nets for grain and straw yields, respectively. This method of model evaluation is also known to be very rigorous, so other statistical methods of evaluation were also used. The r2, RMSE, MBA, and STE were calculated (Table 4). The lowest RMSE and highest r2 for the validated data set were obtained with ANN-10G model (r2=0.85; RMSE = 170 kg ha-1 or 8.4%) for grain yields. Also, the lowest RMSE and highest r2 were obtained with ANN-10S model (r2=0.78; RMSE = 370 kg ha-1 or 9.6%) for straw yields. The values obtained for validation by all models are comparable to each other, which implies that all models were able to learn and generalize. Also, the RMSE values should be close to zero which would indicate that, on average, there is no difference between the


197

simulated and observed values. Given the complexity of estimating crop yields based on spectral data, the above RMSE values can be considered to be fairly close to zero. In addition, there was no significant difference (Pd”0.05) between the r2-values of the models for both training and validation data sets (Table 4), which indicates that the models did a comparable job. The highest r2 was obtained with ANN-1G and ANN-10S models for the grain and straw yields, 0.86 and 0.78, respectively. The lowest standard error was also obtained with ANN-10G and ANN-10S models for the grain and straw yields, 133 and 203 kg ha-1, respectively. It demonstrates that the models can estimate the yields with a desirable accuracy. The MBA values show that the predictions of wheat grain and straw yield models are usually under and over estimate compared to observed data, respectively. In most cases, models have estimated grain yield lower than real amounts. Also, the extended models calculate the straw yield higher than the desired precision. A comparison of the statistical error measures shows that the ANN models estimate grain yields with a greater accuracy prediction than straw yields. The data generated here (from ANN-4G model) suggest that maximum wheat grain yield would be achieved when 90 kg N ha”1 is combined with about 392 mm seasonal water use, which was determined 5230 kg ha-1 in this study (Figure 3). Inputs for the models resulting in the best grain yield predictions as measured by r2 and RMSE were seasonal water and nitrogen use, the minimum air temperature and solar radiation of each of the eight vital months. For the best straw yield predictions model, in addition of the above parameters, the mean air temperature and relative humidity of each of the eight vital months have been important. It may be because of these effects on wheat straw yield. The number of hidden nodes and learning rate in the optimum networks of ANNs models are also presented in Table 3. The number of nodes in the hidden layer was determined by trial and error. The neural networks were trained with 1–30 hidden nodes and after each training runs, RMSE was calculated using only the test data set to find the optimal number of hidden nodes. Figure 4 shows the effect of changing the number of nodes in hidden layer on the networks (ANN-1G and ANN-1S nets) accuracy. It is clear that this factor had a significant bearing on the ANN model predictions accuracy. The best RMSE (192 and 428 kg ha-1 for ANN-1G and ANN1S nets, respectively) were found with 15 and 19 nodes for ANN-1G and ANN-1S nets, respectively. Hence, the network using these nodes in the hidden layer for the mentioned ANNs provided the best results. In most cases, the best results come from a moderate learning rate (0.84). The best nets presented here represent the best results out of all combinations

198


Table 3: Regression parameters for the observed and ANN simulated wheat grain and straw yields models for training and validation steps Model

Inputs

*

Optimum

Training

Validation

No. of hidden nodes

Learning rate

Intercept

Slope

Intercept

Slope

ANN -1G

1, 2

15

0.86

298

0.87

249

0.89

ANN -1S

1, 2

19

0.81

1521

0.68

1330

0.71

ANN -2G

1, 2, 28 -4 3

18

0.87

784

0.84

760

0.8 5

ANN -2S

1, 2, 28 -4 3

19

0.82

1753

0.63

1340

0.73

ANN -3G

1, 2, 3 -1 0 , 28 -4 3

24

0.8 4

462

0.87

459

0.89

ANN -3S

1, 2, 3 -1 0 , 28 -4 3

5

0.81

1320

0.71

1212

0.76

ANN -4G

1, 2, 3 -1 0 , 36 -4 3

7

0.84

790

0.81

675

0.84

ANN -4S

1, 2, 3 -1 0 , 36 -4 3

8

0.84

1247

0.77

1167

0.78

ANN -5G

1, 2, 1 0 -4 3

16

0.86

139

0.90

126

0.86

ANN -5S

1, 2, 1 0 -4 3

17

0.8 4

1405

0.73

1350

0.79

ANN -6G

1, 2, 3 -27 , 36 -4 3

12

0.88

129

0.90

121

0.92

ANN -6S

1, 2, 3 -27 , 36 -4 3

26

0.82

1455

0.71

1353

0.77

ANN -7G

1, 3 -4 3

21

0.86

505

0.83

490

0.85

ANN -7S

1, 3 -4 3

22

0.8 4

1770

0.74

1440

0.79

ANN -8G

2, 3 -4 3

20

0.8 4

613

0.78

587

0.82

ANN -8S

2, 3 -4 3

13

0.80

1850

0.69

1520

0.77

ANN -9G

1, 2, 3 -35

12

0.82

603

0.79

687

0.83

ANN -9S

1, 2, 3 -35

21

0.81

1731

0.68

1497

0.79

ANN -10G

1, 2, 3 -43

9

0.8 7

127

0.91

118

0.93

ANN -10S

1, 2, 3 -43

17

0.8 2

1305

0.70

1150

0.77

*

Inputs are defined in Table 1. Intercepts and slopes are significantly

different

from 0 and 1, resp ectively

(P < 0.05).

199


Table 4: Results of validated ANN wheat grain and straw yields prediction models with varying model inputs and ANN parameters Model

r2

RMSE (kg ha-1)

MBE (kg ha-1)

STE (kg ha-1)

ANN-1G

0.86

192

-105

127

ANN-1S

0.77

428

210

247

ANN-2G

0.81

284

-144

152

ANN-2S

0.74

508

293

280

ANN-3G

0.81

274

-134

149

ANN-3S

0.75

501

280

272

ANN-4G

0.80

357

-209

188

ANN-4S

0.72

621

367

370

ANN-5G

0.85

182

-141

143

ANN-5S

0.74

383

281

287

ANN-6G

0.84

165

-136

141

ANN-6S

0.74

361

298

282

ANN-7G

0.81

287

-250

249

ANN-7S

0.70

768

391

402

ANN-8G

0.78

417

-381

352

ANN-8S

0.61

809

656

698

ANN-9G

0.83

234

-135

131

ANN-9S

0.76

478

250

262

ANN-10G

0.85

170

-121

133

ANN-10S

0.78

370

221

230

200

Journal of Applied Irrigation Science, Vol. 44, No. 2 /2009 6

5

Grain yield (t ha -1)

4

3

2

ANN-4G

Observed data

1

0 100

150

200

250

300

350

400

450

500

550

Seasonal water use (mm)

Figure 3: Wheat grain yield predictions (ANN-4G) as a function of seasonal water use 3.2 Water productivity Table 5 shows regression parameters for the observed and ANN simulated wheat water productivity for training and validation steps. The regression parameters were closer to the ideal values for the ANN-10W net (with an intercept and slope of 0.41 and 0.80 for validation data sets, respectively). The number of hidden nodes and learning rate in the optimum networks of ANN models are also presented in Table 4. For example, the best RMSE (0.19 kg m-3) was also found with 11 nodes for ANN1W net. The results indicate that the lowest RMSE for the water productivity validated data set was obtained with ANN-10W model with a RMSE of 0.12 kg m-3 or 10.3%, (Table 6). The values obtained by all models are comparable to each other which implies that all models were able to learn and generalize. The highest r2 was obtained with ANN-10W model with r2 of 0.83 (Table 5). The MBA values show that the predictions of wheat water productivity models are usually under estimate compared to observed data. The lowest standard error of 0.10 kg m-3 was also obtained with ANN-5W model. It demonstrates that the model can estimate the yield with a desirable accuracy. A comparison of the statistical measures of the various models (Table 6) indicates that with the removal of the effective parameter(s) on water productivity from the proposed models, the values of the statistical indices undergo some changes. The

201


extreme changes in these indices with the removal of one or more factors demonstrate the sensitivity of water productivity to these parameter(s). However, the sensitivity analysis shows that the productivity of wheat water has the highest level of sensitivity to the seasonal water use. With the removal of this parameter (see ANN-7W and ANN-10W) the r2 decreased drastically and the RMSE, MBA and STE content increased enormously. Research results prove that the ranking of the sensitivity of wheat water productivity to the various factors is as follows: seasonal water use > seasonal nitrogen use > monthly mean solar radiation > monthly minimum air temperature > monthly mean air temperature > monthly mean relative humidity > monthly maximum air temperature. Table 5: Regression parameters for the observed and ANN simulated wheat water productivity models for training and validation steps Optimum

Training

Validation

Model

Input variables*

No. of hidden nodes

Learning rate

Intercept

Slope

Intercept

Slope

ANN-1W

1, 2

11

0.83

0.57

0.72

0.49

0.76

ANN-2W

1, 2, 28-43

10

0.84

0.55

0.69

0.60

0.77

ANN-3W

1, 2, 3-10, 28-43

5

0.84

0.50

0.71

0.55

0.73

ANN-4W

1, 2, 3-10, 40-43

26

0.83

0.56

0.68

0.58

0.73

ANN-5W

1, 2, 10-43

4

0.84

0.50

0.70

0.54

0.72

ANN-6W

1, 2, 3-27, 40-43

6

0.85

0.51

0.77

0.45

0.79

ANN-7W

1, 3-43

24

0.82

0.59

0.67

0.62

0.71

ANN-8W

2, 3-43

21

0.80

0.62

0.58

0.64

0.68

ANN-9W

1, 2, 3-35

7

0.84

0.53

0.71

0.50

0.73

ANN-10W

1, 2, 3-43

15

0.85

0.50

0.78

0.41

0.80

*

Input variables are defined in Table 1.

In arid and semi-arid regions where water is limited, small amounts of irrigation water can make up for the deficits in seasonal rain and produce satisfactory and sustainable yields. The findings indicate that use efficiency for water and nitrogen was greatly increased by deficit irrigation. The data generated here (from ANN-5W model) suggest that under deficit irrigation, maximum wheat water productivity would be achieved when 90 kg N ha”1 is combined with about 270 mm seasonal water use, which was determined 1.51 kg m-3 in this study (Figure 5). Consequently, when limited supplemental irrigation is combined with N fertilizer appropriate management, wheat water productivity can be substantially and consistently increased in the region.

202


Table 6: Results of tested ANN wheat water productivity prediction models with varying model inputs and ANN parameters Model

r2

RMSE (kg m-3)

MBE(kg m-3)

STE (kg m-3)

ANN-1W

0.78

0.19

-0.19

0.18

ANN-2W

0.79

0.18

-0.17

0.17

ANN-3W

0.78

0.18

-0.16

0.15

ANN-4W

0.79

0.20

-0.15

0.16

ANN-5W

0.81

0.18

-0.13

0.10

ANN-6W

0.80

0.19

-0.13

0.14

ANN-7W

0.77

0.22

-0.18

0.20

ANN-8W

0.70

0.29

-0.21

0.19

ANN-9W

0.79

0.18

0.17

0.16

ANN-10W

0.83

0.12

-0.10

0.11

R M S E (k g h a -1)

(a)

(b)

(c)

250

600

0.3

200

500

0.25

150

400

0.2

300

0.15

100

0.1

200 50

ANN-

ANN-2S

100

0

0

0 0

5

10

15

Hidden

20

25

30

35

ANN-2W

0.05

0

5

10

15

20

25

30

35

0

5

10

15

20

25

30

35

nodes

Figure 4: Relation between accuracy of neural network (RMSE) and the number of nodes for estimating grain yield (a), straw yield (b), and water productivity (c) for ANN2G, ANN-2S, and ANN-2W nets

203

Zeitschrift für Bewässerungswirtschaft, 44. Jahrg., Heft 2 /2009 1.8 1.6

-3

Water productivity (kg m )

1.4 1.2 1 0.8 0.6 0.4

ANN-5W

Observed data

0.2 0 100

150

200

250

300

350

400

450

500

550

Seasonal water use ( m m )

Figure 5: Wheat water productivity predictions (ANN-5W) as a function of seasonal water use 3.3 Regression The validated models are indicative of the capability of the models to predict parameters with new data. Discussion of the MLR models will refer to the validated models unless otherwise indicated. However, the regression models generally resulted in lower r2 and higher RMSE than ANN models (Table 7). The results indicate that the MLR models did not predict output variables with the same level of accuracy as ANN models. At the all cases, the same inputs were used for both modeling techniques, however ANN models gave better yield estimates. Also, the straw yield models had the lowest accuracy among the other ones. As seen in Figure 6, grain and straw yields and water productivity predictions from MLR model (r2 = 0.57, RMSE = 413 kg ha-1 for grain yields; r2 = 0.52, RMSE = 940 kg ha-1 for straw yields; and r2 = 0.43, RMSE = 0.27 kg ha-1 for water productivity ) were not a good fit to observed yields when compared with ANN-2G predictions for grain yield (r2 = 0.81, RMSE = 284 kg ha-1); with ANN-2S predictions for straw yields (r2 = 0.74, RMSE = 508 kg ha-1) and with ANN-2W predictions for water productivity (r2 = 0.79, RMSE = 0.18 kg m-3), which used the similar input data groups. Comparisons of observed versus predicted grain yields for MLR models resulted in a least squares linear regression line with a slope of 0.62 and y-intercept

204


of 2274, while the ANN-2G model comparisons resulted in a line with a slope of 0.80 and y-intercept 1299. The line slope and y-intercept of straw models were determined 0.43 and 5061, respectively, for the MLR model, and 0.76 and 1805, respectively, for the ANN-2S model. For the water productivity index, these parameters were estimated 0.70 and 0.472, respectively, for the MLR model, and 0.84 and 0.22, respectively, for the ANN-2W model. Hence, these findings also show that output variables predictions from MLR models were not a good fit to observed yields in comparison to ANN predictions. Table 7: Results of validated multiple regression wheat yield and water productivity prediction models with varying model inputs Input variables 1, 2

Grain yield RMSE (kg ha-1) 482

r2 0.65

Straw yield RMSE (kg ha-1) 831

r2 0.56

Water productivity RMSE (kg m-3) r2 0.38 0.59

1, 2, 28-43

413

0.57

940

0.52

0.27

0.43

1, 2, 3-10, 28-43

529

0.62

894

0.51

0.41

0.52

1, 2, 3-10, 40-43

618

0.59

987

0.55

0.33

0.55

1, 2, 10-43

596

0.54

791

0.48

0.29

0.61

1, 2, 3-27, 40-43

425

0.61

721

0.56

0.36

0.64

1, 3-43

480

0.69

1125

0.57

0.43

0.46

2, 3-43

682

0.54

1320

0.53

0.39

0.51

1, 2, 3-35

429

0.58

1510

0.53

0.31

1, 2, 3-43

471

0.53

1019

0.51

0.37

0.53 0.45

The comparison of the results for ANNs and MLR models demonstrates that the ANNs models have a better accuracy in estimating the wheat yield production and water productivity. The reason of the better prediction of ANNs models has to do with the intelligence of the process of analyzing the data, so that during the process of training the networks is able to learn about the logical relationships between inputs and outputs variables, and use them for estimating the data that hasn’t been used in training net. On the other hand, the regression analyzes is only capable of making a set of mathematical and statistical meanings between the data. The findings indicate that the applied ANNs models have a high potential for future crop prediction based on weather and environmental data in the form of neural nets, which can obtain

205


greater accuracy comparable to MLR. The framework for neural network development presented here for crop-weather modeling could also be extended to most other agricultural modeling with neural nets. (a)

(b) 8000

8000 y = 0 . 7 9 7 1x + 1 2 9 9 . 1 r2 = 0 . 8 1 2

Predicted grain yield (kg ha- 1 )

7000

y = 0 . 6 1 8x + 2 2 7 4 . 2 r2 = 0 . 5 6 7

7000

6000

6000

5000

5000

4000

4000

3000

3000

2000

2000

1000

1000 0

0 0

0

1000 2000 3000 4000 5000 6000 7000 8000

1000 2000 3000 4000 5000 6000 7000 8000

Observed grain yield (kg ha - 1 )

(a)

(b)

-1

P re d i c te d s tra w y ie l d (k g h a )

20000

20000 y = 0.76x + 1805.1 r 2 = 0.737

16000

12000

12000

8000

8000

4000

4000

0

0 0

4000

8000

12000

16000

0

20000

4000

8000

12000

16000

20000

1.6

2

Observed straw yield (kg ha-1) (b)

(a)

2

2 P re d ic te d w a te r p ro d u c tiv ity (k g m - 3 )

y = 0.435x + 5061.2 r 2 = 0.522

16000

y = 0.8432x + 0.2272 r 2 = 0.793

1.6

y = 0.7032x + 0.472 r 2 = 0.587

1.6

1.2

1.2

0.8

0.8

0.4

0.4

0

0 0

0.4

0.8

1.2

1.6

2

0

0.4

0.8

1.2

Observed water productivity (kg m -3)

Figure 6: Scatterplots displaying results of artificial neural network (a) and multiple regression (b) predicted versus observed output variables for the various ANN model from the validation data set. The line represents a least squares linear regression of predicted versus observed indices

206


4. Conclusions In the present study, the climate-sensitive decision points, and the seasonal water and nitrogen use practices on wheat productivity were investigated. A feed-forward back-propagating Artificial Neural Network structure was applied to develop wheat yields and water productivity prediction models. The results indicate that the ANN nets developed here have the potential to be useful as a component of water and nitrogen management planning within the study area given further development and validation. The ANN models show promise as a more accurate technique that could be used by semi-arid regions water management specialists to develop, revise, or update water and nitrogen management plans in the different climatic condition. ANN models consistently produced more accurate predictions than MLR models in estimating the wheat yields production and water productivity. The applied ANNs can estimate the output variables with a desirable accuracy, and can be used for field scheduling with the consideration of the regional parameter changes and different scenarios for managing water and nitrogen to optimize wheat water productivity. Using these models can provide the background for promoting the water productivity for this strategic crop in different weather condition, and give us the possibility of the logical and economical use of water sources and nitrogen and also programming the combined using of both resources. The importance of using these models can be seen by considering the limitation of water, environmental problems and the expenses of inconvenient use of nitrogen fertilizer which is forced on farmers in the study area, and also in a climate change status. The results indicate that the ranking of the sensitivity of wheat water productivity to the various factors may be showed as follows: Seasonal water use > seasonal nitrogen use > monthly mean solar radiation > monthly minimum air temperature > monthly mean air temperature > monthly mean relative humidity > monthly maximum air temperature. The production functions can assess the yield of wheat under different growing environments, as shown in the present study can thus help in a long way to study the impacts assessment in agri-production on a farm scale and also the regional scale, by further use of remote sensing and GIS. ANN modeling with additional locations will increase the variability of soil types and should broaden the usefulness, and possibly increases, the predictive capabilities of ANN-based water productivity prediction in the regional and state levels.


207

Acknowledgment Funding for this work was provided by the deputy of Research and Technology, University of Tehran through the project No. 7305002-01-P. The author sincerely appreciates due to N. Shabazian and M. Mohseni for their technical assistance in the field experiments.

References Allen, G. A., L. S. Pereira, D. Raes and M. Smith, 1998: Crop evapotranspiration – Guidelines for computing crop water requirements. - FAO Irrigation and Drainage Paper 56. FAO, Rome, Italy, 78-86. Atkinson, P. M. and A. L. Tatnall, 1997: Neural networks in remote sensing. - Int. J Remote Sens 18 (4): 699–709. Baker, W. L., 1991: An application of neural networks to forecasting corn yields. - MSc. Thesis, Department of Agricultural Economics. Purdue University, West Lafayette, IN, USA. Basso, B., J. T. Ritchie, F. J. Pierce, R. P. Braga and J. W. Jones, 2001: Spatial validation of crop models for precision agriculture. - Agri Sys 68: 97–112. Batchelor, W. D., X. B. Yang and A. T. Tshanz, 1997: Development of a neural network for soybean rust epidemics. - Trans of ASAE 40:247–252. Caudill, M., 1990: Neural Networks Primer. - Miller Freeman Publications, San Francisco, CA, USA. Chang, D.H. and S. Islam, 2000: Estimation of soil physical properties using remote sensing and artificial neural network. - Remote Sens Environ 74 (3):534–544. Coulibaly, P., F. Anctil and B. Bobee, 2000: Daily reservoir inflow forecasting using artificial neural networks with stopped training approach. - J Hydrol 230(3–4):244–257. Del Frate, F., P. Ferrazzoli and G. Schiavon, 2003: Retrieving soil moisture and agricultural variables by microwave radiometry using neural networks. - Remote Sens. Environ. 84 (2): 174–183. Drummond, S. T., K. A. Sudduth, A. Joshi, S. J. Birrell and N. R. Kitchen, 2003:. Statistical and neural methods for site-specific yield prediction. - Trans of ASAE 46 (1): 5–14. Elizondo, D. A., R. W. McClendon and G. Hoogenboom, G. 1994: Neural network models for predicting flowering and physiological maturity of soybean. - Trans of ASAE 37: 981–988. Green, T. R., J. D. Salas, A. Martinez and R. H. Erskine, 2007: Relating crop yield to topographic attributes using Spatial Analysis Neural Networks and regression. - Geoderma 139: 23– 37. Gross, L., S. Thiria, and R. Frouin, 1999: Applying artificial neural network methodology to ocean color remote sensing. - Ecol. Model. 120 (2–3): 237–246. Gutiérrez, P. A., F. López-Granadosb, J. M. Peña-Barragánb, M. Jurado-Expósito, M. T. GómezCaserob and C. Hervás-Martíneza, 2008: Mapping sunflower yield as affected by Ridolfia segetum patches and elevation by applying evolutionary product unit neural networks to remote sensed data. Compu and elect in agric 60: 122–132. Jame, Y. W. and H. W. Cutforth, 1996: Crop growth models for decision support systems. Canadian Journal of Plant Science 76: 9–19. Jin, Y. Q. and C. Liu, 1997: Biomass retrieval from high-dimensional active/passive remote sensing data by using artificial neural networks. - Int. J Remote Sens 18 (4): 971–979. Kaul, M., R. L. Hill and C. Walthall, 2005: Artificial neural networks for corn and soybean yield prediction. - Agricultural Systems 85: 1-18.

208


Kimes, D. S., R. F. Nelson, M. T. Manry and A. K. Fung, 1998: Attributes of neural networks for extracting continuous vegetation variables from optical and radar measurements. - Int. J Remote Sens 19 (14): 2639–2663. Liu, J., C. E. Goering and L. Tian, 2001: A neural network for setting target corn yields. - Trans of ASAE 44 (3): 705–713. Pachepsky, Y. A., D. Timlin and G. Varallyay, G. 1996:. Artificial neural networks to estimate soil water retention from easily measurable data. - Soil Science Society of American J 60: 727–733. Poluektov, R. A. and A. G. Topaj, 2001: Crop modeling: nostalgia about present or reminiscence about future. - Agronomy Journal 93: 653–659. Safa, B., A. Khalili, M. Teshnehlab and A. M. Liaghat, 2003: Predictions of wheat in dry-land using artificial neural networks. - Niowar J 48: 47-62 (in Persian). Schaap, M. and W. Bouten, 1996: Modeling water retention curves of sandy soils using neural networks. - Water Resources Research 32: 3033–3040. Stewart, D. W., L. M. Dwyer and L. L. Carrigan, L.L. 1998: Phenological temperature response of maize. - Agronomy Journal 90: 73–79. Wang, F., C. W. Fraisse, N. R. Kitchen and K. A. Sudduth, 2002: Site-specific evaluation of the CROPGROW-soybean model on Missouri claypan soils. - Agriculural Systems 76: 985– 1005. Yang, C. C., S. O. Prasher, S.Sreekanth, N. K. Patni and L. Masse, 1997: An artificial neural network model for simulating pesticide concentrations in soil. - Trans of ASAE 40: 1285– 1294. Zhang, Y., J. Pulliainen, S. Koponen and M. Hallikainen, 2002 Application of an empirical neural network to surface water quality estimation in the Gulf of Finland using combined optical data and microwave data. - Remote Sens Environ. 81 (2–3): 327–336.

Address of Author: A. Montazar Department of Irrigation and Drainage Engineering Agricultural College of Aboureyhan University of Tehran P.O. Box 11365-4117 Pakdasht, Iran Email: [email protected]


Seiten 223 - 230

Environmental implications of unhygienic operation of a city abattoir in akure, Western Nigeria A. O. Akinro, I. B. Ologunagba and O. Yahaya

Keywords Slaughterhouse wastewater, water quality, effects of pollutants on the environment, water borne diseases, recommendations to protect the environment Abstract Slaughterhouse wastewater has a complex composition and very harmful to the environment. Effluents of a major city abattoir in Nigeria was studied for possible pollutants and effects of such pollutants on the environment. Findings showed that the various water samples were contaminated with E. Coli and other enteric bacteria. The presence of coliform staphylococcus aures indicated the presence of microorganisms which are associated with water borne disease. Recommendations were made to ensure maintenance of good environmental condition in the city abattoirs particularly in the developing countries.

1. Introduction While the slaughtering of animals results in significant meat supplies, a good source of protein and production of useful by-products such as leather, skin and bones, the processing activities involved sometimes result in environmental pollution and other health hazards that may threaten animal and human health. ALONGE (1991) defined meat hygiene as a system of principles designed to ensure that meat and meat products are safe, wholesome and processed in a hygienic manner and are fit for human consumption. Meat quality control is a system that regulates the measure of extrinsic materials such as chemical residues, toxins, pathogenic microorganisms and putrefied tissues, which could be present in meat and are deleterious to human health (OLUGASA et al., 2000). Animals are slaughtered in abattoirs for sale to the public. An abattoir has been defined as a premise approved and registered by the controlling authority for hygienic slaughtering and inspection of animals, processing and effective preservation and storage of meat products for human consumption (ALONGE, 1991). Previous studies have shown that the characteristics of abattoir wastes and effluents vary from day to day depending on the numbers and type of stocks being processed (KEELY and QUINN, 1979; LITCHFIELD, 1980; TOVE, 1985). Abattoir operations

224


produce a characteristic highly organic waste with relatively high levels of suspended solid, liquid and fat. The solid waste includes condemned meat, undigested ingesta, bones, horns, hairs and aborted foetuses. The liquid waste is usually composed of dissolved solids, blood, gut contents, urine and water. Animal food is always microbiologically contaminated by organisms living in it naturally or entering it from the surroundings, such as those resulting from processing operations (LEWICKI, 1993). On going production quality control, washing and disinfection, are the main procedures of securing the hygiene of meat and meat products (PEZACKI, 1970, WINDYGA et. al., 1996). In the production of animal for food, more attention should be focused on the interactions between animal production and the environment, realizing environmental conditions and structures in animal production, which not only seek to produce wholesome and safe animal food but should also avoid environmental pollution and the associated human health risks. Animals slaughtered in Araromi abattoir alone accounts for about 65% of the total animal in Akure, the capital city of Ondo State, Nigeria. The waste from the slaughtering and dressing grounds in the abattoir are washed into open drainages untreated and the leachates from the series of decomposition processes of these wastes can introduce enteric pathogens and excess nutrients into the surrounding surface waters and also percolate into the underlying aquifers to contaminate the hand-dug wells which serve the dual purpose of drinking water for the butchers and others working in the abattoir, and the people in the neighbourhood. With inadequate slaughtering and disposal facilities, the abattoir has also become a source of infection and pollution, attracting domestic and wild carnivores, rodents and flies, which are vectors of diseases. The area is rampant with filth and scattered rubbish, which is left uncollected, apart from the blood draining trenches through which the filth is scattered rather than eliminated. Hygiene problems are not limited to slaughtering but are also associated with incorrect processing and marketing practices. Under tropical conditions, food of animal origin tends to deteriorate more rapidly and become an important vehicle for gastrointestinal infections, thereby endangering consumers’ health. Some of the human infections acquired from meat and poor handling of food animals are shown in Table 1. This paper presents our findings on the sanitary-hygienic conditions of a major city abattoir in southwestern Nigeria, the environmental implications is discussed.

225


Table 1: Some human infections acquired from meat and the handling of food animals Bacterial Anthrax, Q- fever, Campylobacteriosis, Ornithosis, Botulism, Staph. food poisoning Salmonellosis Brucellosis, Erysipelas, Streptococcosis, Tetanus, Yersiniosis, Clostridiosis, Listeriosis, Glanders, Leptospirosis, Tuberculosis.

Viral Rift valley fever, Newcastle disease, Vesicular stomatitis, Contagious ecthyma

Parasitic Taeniosis, Toxoplasmosis, Echinococcosis (indirect), Sarcosporidiosis, Trichinellosis, Fungal dermatophytosis

2. Materials and methods The study was conducted in Akure, the capital of Ondo State in South west Nigeria which lies at latitude 7o 16’ north and longitude 5o 13’ East at an altitude of 351 m above mean sea level. The population of Akure based on the 1999 census is 386,550. Although, there are other abattoirs in Akure, Araromi abattoir was selected for this study based on its strategic location right at the city centre and also because it is the major abattoir which supplies about 65% of the meat for the city residence. The physicochemical and microbiological properties of the effluent was investigated for a period spanning 3 months (July-September, 2008) which falls within the rainy period in the region and during which appreciable flow occurred in the waste effluent receiving streams. The effluents from Araromi abattoir were discharged into a stream which in turn discharges into Ala River, major surface water that spanned the entire city. The solid wastes from the abattoir are evacuated and constantly trucked away for land disposal. Farmers are also encouraged to collect the material free of charge for use as manure. Horns are washed and neatly packed for further processing either for breakable plates or livestock feeds. Samples of water were collected on different occasions from the vicinity of the abattoir site. All visitations and samplings were done as early as 6.30 am when slaughtering through processing to the sales of the meat were observed. All glasswares such as Petri dishes, conical flask, measuring cylinder and test tube were washed with detergent, rinsed in clean water and dried in the drying cabinet. The glasswares were then sterilized in the hot air oven (autoclaves) at 1210C for 20 minutes. Effluents were collected from drains. Samples were preserved by addition of HNO3 prior to laboratory analysis; other preservations were done by refrigeration to avoid loss of nitrate. Dissolved oxygen was on site by precipitation. The effluent was examined for physicochemical and microbiological characteristics. The results of the effluents were compared with WHO standards and the pollution determinants were determined.

226


All the chemical analysis were carried out in accordance with guidelines of the Environmental Protection Agency (2002).

3. Results and analysis The results of the study looked at the physicochemical and microbiological characteristics of the effluent from the abattoir. 3.1 Physicochemical characteristics The result of the physicochemical analysis of the effluent is summarized in Table 2. Table 2: Physicochemical characteristics of effluent from Araromi abattoir Parameter Tested Mean Temperature 27.30C (±9.250C)* pH 7.41(±0.26) Colour Dark brown Odor Foul Conductivity 19.0 x 102 (±8.25) Acidity (ppm) 0.9 (±0.13) Alkalinity (mg/l 0.4(±0.05) Dissolved Oxygen 2,000 (±231) (ppm) Total dissolved solid 240 (mm/l) Total suspended solid 480 (mm/l) Total Solid (mm/l) 685 Biochemical Oxygen 42 Demand (mm/l) Total Hardness 172.5 Chloride (mm/l) 4.6 Calcium 83 Magnesium ND Aluminium ND Lead ND Iron 7.3 ¡ Standard deviation in parenthesis

227


The World Health Organization (WHO) classifies the extent of chemical levels in water for drinking purpose as either inoffensive or unobjectionable when it is within acceptable range for human consumption. It is considered inoffensive for any specific physical quantity when the value is far below the maximum value and/or when that specified physical quantity will have no detrimental effect on human health when it reaches up to the maximum level. It is also unobjectionable when the value of a specified quantity is within the range of the highest desirable level but not up to the maximum permissible level. The result in Table 2 showed that the composition of the slaughterhouse is affected by the number of animals slaughtered and the disposal method employed. A significant part of the variation can be seen in the Total Suspended Solids (TSS) and the exhibited high dissolved oxygen and Biochemical Oxygen Demand (BOD). These are characterized by the varying amount of washed water, heads of animals and solid waste. The mean pH in sample was 7.41. WHO defines 6.5-8.5 as the suitable range for hydrogen concentration (pH) levels. The range in the effluent is considered inoffensive. However, due to the high proportion of total solids in the samples, one possible management option would be to include screens as a primary treatment process thereby reducing the total washdown to the rivers and streams. 3.2 Analysis of bacteriological quality Result from the bacteriological analysis is presented as Table 3 Table 3: Counts of biological organisms in sampled effluent per 100ml Sample

Coliforms

A B C D E

184 191 188 185 190

Other Enteric Bacteria (cfu/mu) 7.30 x 106 7.64 x 106 7.10 x 106 7.43 x 106 7.25 x 106

The result indicated that the various samples were contaminated in one way or the other with E.Coli and other enteric bacteria. It can be deduced from Table 3 that all the water samples used for the test were polluted biologically beyond permissible limits. The presence of coliform staphylococcus aurens was confirmed in the abattoir. The presence of this bacteria in intolerable number obviously constitute a serious public health hazards as the presence of these micro organisms is associated with water borne diseases since the waste is discharged into the streams. The seepage of the effluent to well and borehole also constitute a serious health hazard to the public. Further observation also showed that the surroundings of the abattoir gave offensive

228


odours and breed mosquitoes due to the pile up paunch contents and other solid wastes, faeces, carcass, horns, scraps of tissue and other solid waste. Waste products are at best an embarrassment or nuisance and at worst serious pollutants. Such systems of production are not sustainable in the long-term and it is possible to develop integrated systems where local inputs are optimized and recycled, with a reduction in external inputs. Sustainable animal production means, that we are able to produce food animal and animal products without lasting damage to the environment, which means that essential elements like water, air and soil are left without dead loads and that by-products of animal production creates no animal and human health risks through environmental protection and animal waste management (TIELEN, 2000). The role of livestock manure as a source of fertilizer should not be underestimated because all organic wastes are evidently needed in many parts of the world to restore the land to maximum productivity. However, few epidomological studies have established definitive adverse health impacts attributable to pathogenic organisms in agricultural reuse of wastewaters from abattoirs. Helminthic diseases caused by Ascaris and Trichuris spp. are endemic in areas of the world where raw untreated sewage is used to irrigate salad crops and vegetables eaten uncooked (SHUVAL et al., 1985; 1986). If sustainable agricultural systems are to be developed that are largely independent of external inputs, solid waste from slaughtered animals can be fermented in a tank, this produces compost and biogas. Biogas was being produced as early as the 1920s in a number of communal sewage farms in central Europe; but the primary consideration was not so much how to obtain additional energy, but rather the problem of rational and hygienic waste disposal (SPORE, 1993).

4. Conclusion The results obtained from the investigation showed that effluents from the abattoir constitute potential hazards to the environment. The high level of Total Suspended Solid (TSS) and Conductivity indicates that the samples were heavily loaded with colloidal, organic, inorganic and suspended matters. Clean technologies have been promoted to serve many purposes, such as the reduction of pollution generated by conventional abattoir operations, the improvement of process efficiency and energy conservation, leading to more cost-effective and profitable operation, and the optimization of the use of raw materials, thereby promoting a more efficient use of natural resources. The maintenance of good environmental conditions by disposing sewage and refuse in a sanitary manner in the abattoir starts with the definition of the minimum requirement for all the links in the production chain and these includes:


229

– Installation of necessary standard equipment and major functional units of the abattoir such as cold rooms, skinning machines, slaughtering machines and changing rooms for workers; – Thorough and adequate training on sustainable animal production for the people involved in animal trade from farms to abattoirs and slaughterhouses, including periodic continuing education programs; – Maintenance of proper hygiene within the abattoir and the environment, target areas for sanitization include: infrastructures and facilities contained therein, equipment, surrounding areas, abattoir workers and visitors; – Periodic Sanitary-hygienic evaluation of abattoirs and slaughterhouses; – Enforcement of existing health and hygiene regulations.; – Development of appropriate technology, which will take care of all the wastes being generated in the abattoir, including abattoir wastewater treatment and recycling for irrigation; and – compost and biogas production.

References Abiola, S. S., 1995: Assessment of Abattoir and Slaughter Slab Operation in Oyo State. Nigerian Journal of Animal. Production 5: 54-62. Alonge, D. O., 1991: Textbook of Meat Hygiene in the Tropics. - Farmcoe Press, Ibadan, Nigeria. 58pp. Barrett, J. R., 2001: Livestock Farming: eating up the Environment? - Environ. Hlth. Perspec. 109 (7): A312-A317. Bellani, I., A. Mantovani and I. Ravaioli (eds.), 1978: Proceedings of the W HO Expert Consultation on some Veterinary Public Health Problems. - Annali Istituto Superiore di Sanità, Rome. FAO, 1989: Urban Food Consumption Patterns in Developing Countries. By H. Delisle. Rome. Filani, M. O., 1994: Ibadan Region, Re-Charles Publications in Conjunction with Connell Publications, Ibadan, Nigeria. Lewicki, P., 1993: Higiena Produckcji. - Czesc I. Przem. Spoz. 47 (10): 275-276 Meadows, R., 1995: Livestock Legacy. Environmental Health Perspectives 103 (12): 1096 – 1100. Mitchell, R. and I. Chet, 1978: Indirect Ecological Effects of Pollution, In R. Mitchell (Ed.), In Water Pollution Microbiology. - John Wiley and Sons, New York, 2: 177-199. Moran, J. M., M. D. Morgan and J. H. Wiersma, 1980: Introduction to Environmental Science. - W.H. Freeman and Co., San Francisco. Odeyemi, O., 1991: Consequences of Water Pollution by Solid Wastes and Faecal Materials In Nigeria. - In: Akinyele, L, Omueti, J. and Imevbore, T. (eds.): Proceedings of the Third National Conference on Water Pollution, June, 1991. Port Harcourt, Nigeria. Olugasa, B. O., S. I. B. Cadmus and N. N. Atsanda, 2000: Actualization of Strategies for Beef Quality Control in reply to: South Western Nigeria. In M.J.M. Tielen and M.T.H. Voets TH (eds.). - In: Proceedings of the X International Congress on Animal Hygiene. Maastricht, Netherlands, ISAH 1: 67-71. Pezacki, W., 1970: Mikrobiologiczne Problemy Technologii Miesa. - Post. Mikrobiol. 9 (3): 472441.

230


SPORE, 1993: The Thiès abattoir: from pollutant to fertilizer. CTA No. 47. Shuval, H. I., P. Yekutiel and B. Fattal, 1985: Epidiomological evidence for helminth and cholera transmission by vegetables irrigated with wastewater. Jerusalem – case study. - Water Science and Technology. 17(4/5): 433-442 Shuval, H. I., A. Adin, B. Fattal, E. Rawitz and P. Yekutiel, P., 1986: Wastewater irrigation in developing countries: health effects and technical solutions. - Technical paper No. 51. World Bank, Washington D.C. Tielen, M. J. M., 2000: Animal Hygiene: the Key to Healthy Animal Production in an Optimal TH Environment. - In: M. J. M. Tielen and M. T. H. Voets (eds.): Proceedings of the X International Congress on Animal Hygiene. Maastricht, Netherlands, ISAH 1: 3-10. UNFPA, 1995: The State of World Population. New York. Von Braun, J., J. McComb, B. K. Fred-Mensah and R. Pandya-Lorch, 1993: Urban Food Insecurity and Malnutrition In reply to: Developing Countries: Trends, Policies and Research Implications. - Washington, DC, IFPRI. WHO, 1981: WHO/WSAVA Guidelines to Reduce Human Health Risks Associated with Animals in Urban Areas. - VPH/81.29. Geneva. Windyga, B., A. Grochowska, B. Urbanek-Kalowska and B. Napiorkowaka, 1996: Preparaty Dezynfekujace Dopuszczone Do Stosowania W. Zacladach Przetworstwa Spozywczego. - Przem. Spoz. 50 (6): 27-34.

Address of Authors: A. O. Akinro Department of Civil Engineering Rufus Giwa Polytechnic Owo, Nigeria Department of Agricultural Engineering Federal University of Technology Akure, Nigeria Tel. +234 803 574 1488 Email address: [email protected] I. B. Ologunagba Department of Civil Engineering Rufus Giwa Polytechnic Owo, Nigeria Olotu Yahaya Right Foundation Academy P.O. Box 50 Ikare-Akoko, Ondo-State, Nigeria


Seiten 231 - 238

Bacteriological and physico-chemical analysis of some domestic water wells in peri-urban areas of Akure, Nigeria A. O. Akinro and I. B. Ologunagba

Keywords Well water quality, water analysis, contamination, well protection, formulation and implementation of guidelines Abstract Quality tests were conducted on some domestic water wells in some locations in Akure, the capital city of Ondo State, Nigeria to evaluate the suitability of the water for consumption. Wells were selected for sampling based on their closeness to poorly constructed or ageing septic tanks, drains, animal housing and pit privies. The data indicated that the calcium hardness ranged from 33 – 146 mg/l and total hardness also ranged from 42 – 165 mg/l. According to APPA (1995), none of these values exceeded the permissible level of 500 mg/l. Chloride levels ranged from 10 – 30 mg/l. Chloride level could therefore be considered inoffensive or unobjectionable. The values of iron present in samples were negligible and therefore considered as traces. However, from the bacteria counts per 100 ml of water sample, organisms were visible in four categories of colours, indicating that four different kinds of organisms were present in the wells surveyed.

1. Introduction Water is essential to human life. Access to safe water supply facility is crucial to ensuring good health and essential for sustainable development. Inadequate quantity and/or quality of drinking water, lack of sanitation facilities and poor hygiene directly and indirectly affects public health. Access to safe drinking water is an important indicator of risk exposure to water related diseases (WHO/UNICEF, 2006a). “Improved water supply” is defined to include “reasonable access” to protected water resources which include protected springs and dug wells, boreholes, public stand pipes and household connections. In addition, it involves the application of measures to protect the water source from contamination (HURTON and HALLER, 2004). Reasonable access means at least 20 liters/person/day, accessible within 1 km of that person’s dwelling. (WHO/UNICEF, 2005), while “improved sanitation” involves access to sanitation facilities which allow for safe disposal of excreta. It is defined to include connection to a sewer or septic tank system, pour-flush latrine, simple pit or ventilated improved pit-latrine (WHO/UNICEF, 2005). The excreta

232


disposal system is considered adequate if it is private or shared (but not public), and if excreta are hygienically separated from human contact. (WHO/UNICEF/WSSCC, 2000). The Nigerian government has long considered the provision of water supply and sanitation services to be the domain of the federal, state, and local governments (FGN, WS&SISN, 2000). However, these tiers of government have not been able to meet the domestic water needs of the people both in the rural and urban centers of Nigeria. Services are critically short supplied. For example, out of the 85 million people living in urban and semi-urban areas of Nigeria, less than 50% have reasonable access to reliable water supply (FGN, WS&SISN, 2000). Water supply services, where they exist, are unreliable and of low quality and are not sustainable because of difficulties in management, operation and maintenance, pricing and failure to recover costs. As man cannot do without this essential utility, people tend to seek for water whichever source they can find it. One of such sources is through construction of shallow wells. Available statistics showed that wells- either dug wells with or without aprons constitute over 45% of the available water sources in Nigeria as shown in Table 1. Table 1: Sample of water use patterns by available water sources Spring/stream 32% Hand dug well (w/apron) 30% Hand dug well (w/out/apron) 27% Rain 20% River 16% Pipe borne 14% Borehole 14% Vendors 6% Source: FGN, Small Towns Water Supply and Sanitation Program development studies, 1997

However, many of the constructed wells tend to have poor water quality due to contaminations from the surface or from nearby septic tanks and other sources. These may make such water not ideal for human consumption. Thus the need to check the unwholesomeness of this water source. The study therefore, were undertaken to evaluate the physico-chemical characteristics and bacteriological properties of some shallow wells used for domestic purpose in peri-urban centers of Akure, Ondo State, Nigeria in other to ascertain their wholesomeness.

233


2. Materials and method The study area is Akure Municipal, the capital of Ondo State of Nigeria. The population of Akure based on the 1999 census is 386,550. The technique adopted for this study is largely quantitative and it utilizes data that is collected through household interview using standard quantitative technique Quantitative research allowed the selection of a representative sample from among the population to be investigated, which then allows an analysis that generate inferences for the entire population under investigation (NEWMAN, 2000). The study area was first divided into 5 zones (see Table 2). The households were randomly selected based on the map of the study area and demarcation of zones. Size of households as well as residential density of each zone determines the number of questionnaires administered in the area. Table 2: Zoned areas of Akure selected for study A1 Futa area Ilesha Road

A2 Ilesha Garage area Aule area Leo Hospital area Champion

A3 Idanre Road Danjuma

A4 Shagari Village

A5 Ijoka area

Reconnaissance visits were made to the selected households for the assessment of domestic water wells in their residences. Informal interviews were conducted with house owners and these were followed up with the administration of questionnaires. The study questionnaires were admitted on 18th March, 2008 to heads of household and organizations. 50 questionnaires were administered out of which 40 were received a week after. Samples of water from the wells were taken to the Federal Ministry of Water Resources Regional Water Laboratory, Akure, Nigeria for quality tests. The aim was to obtain thorough information about the wells and the quality of water from them. Water samples were taken into sterilized bottles and screened thoroughly to obtain the bacteria count of the water. To ensure that no organisms were admitted into the bottles other than that in the water, the containers used to draw the water were also sterilized on the site using cotton wool damp with ethanol and in flames. Samples were securely covered, labeled and stored in a refrigerator for less than 24 hours. The duration of storage ensured that organisms present survived. The stan-dard methods described by APHA (1995) for pH, turbidity, colour, chloride, sulphite, total dissolved solids and taste were used for all the samples. The heavy metals were determined as described in standard methods (APHA, 1995) using Atomic Absorption spectrophotometer (A bulk Model 200A flame AA systems). Isolation, enumeration

234


and identification of bacteria in the water samples were done using standard microbiological procedures (HARIGAN and McCANCE, 1976; BUCHANAN and GIBBONS, 1974; COWAN and STEEL, 1974; TWORT et al., 1974).

3. Results and discussions The results of the study looked at wells environment, water lifting procedure, physicochemical and bacteriological quality of the wells water. About 83% of the wells had some concrete well heads while the rest had no structural protection whatsoever. No boreholes were found in the selected areas because of the prohibitive cost. All the wells were thus of the open type and the number of people depending on a particular well varied, ranging from 5 to 12 persons/day. This is because people in the neighbourhood without any source of domestic water depended on the wells of neighbors. Those households without wells attributed it to prohibitive cost. It was also inferred from the survey that 74.7% of the households depended entirely on the well water for every use including drinking without any treatment while the remaining 25.3% used the water only for washing and cooking. Such owners purchased treated sachet water for drinking. All the wells were also close to pit privies. Five of such wells were considered for detailed study. Table 3 shows the physical parameters of the wells selected for analysis. The diameters of the wells were not standardized but ranged from 1.5 to 2.1 m and their depths were generally less than 20 m. All the wells had concrete lining, covering the inner wall either in the form of pre-cast concrete rings or brick/masonry, the thickness generally varies from 0.5 to 0.75 m according to the depth of the well. The rate of water recharge was found to be very rapid in most of the wells visited, taking less than 24 hours to recharge after voluminous water withdrawal. The aquifers were unconfined. All the wells considered were near some source of pollution, mainly septic tanks and animal housing. In most cases, these structures were located topographically at higher elevations making the wells susceptible to contaminations from groundwater flow. These structures, in all cases, were less than 50 m from the well, with some being as close as 4.5 m. The wells had some protective headwalls erected to about 1. 5 m heights. The survey also indicated that siltation occurred regularly in various wells. Many had a fair knowledge of its implications hence; they regularly de-silted these wells and scrubbed the walls of the wells every year. The duration of one year for desilting and scrubbing was common to all the well owners. Wastewaters from households were channeled into open lined drains. The drains were located at distances of about 10 to 25 m from the wells. Most of the wells were found with major cracks.

235


Table 3: Physical parameters of typical wells Well Parameter W1 1.60 11.5 Rapid 20.6

Well Identification W2 W3 W4 1.75 1.80 2.10 18.3 17.5 18.9 Rapid Moderate Rapid 25.4 40.5 37.2

W5 Diameter 1.9 Depth (m) 16.7 *Rate of Recharge Rapid Distance from contamination 29.5 (m) Height of Headwall (m) 1.2 1.5 1.45 1.5 1.4 Distance from drainage (m) 12.3 19.7 10.4 18.2 23.6 *Rate of recharge is considered rapid if well recharges within 24 hours after rapid water draw and moderate if it occurs within 48 hours.

3.1 Analysis of bacteriological quality Quality tests were conducted to evaluate the suitability of the water for consumption. Wells were selected for sampling based on their closeness to poorly constructed or ageing septic tanks, drains, animal housing and pit privies. From the bacteria counts per 100 ml of water sample, organisms were visible in four categories of colours, indicating that four different kinds of organisms were present in the wells surveyed as shown in Table 4. The counts of biological organisms in sampled water are shown in Table 5. Table 4: Category of well water and colour representation No. 1 2 3 4

Colour Violet Purple Greenish-blue Cream

Inference Coliforms Escherichia coli Salmonella Other enteric bacteria

Table 5: Counts of biological organisms in sampled water per 100 ml Label W1 W2 W3 W4 W5

Coliforms 65 1730 924 1125 210

Counts per 100 ml of well water sample Salmonella Other enteric bacteria 12 >300 7 >300 11 >300 13 >300 5 >300

236


The result showed that all the water samples were contaminated in one way or the other with organic material and bacteria Coli per 100 ml of water sample. Table 4 showed that all the wells sampled were polluted biologically beyond permissible limits by 100% and above. All samples were infected by E.Coli, Salmonella and other enteric bacteria. The detection of E.Coli in water sample is an indication of feacal contamination which is harzardous to health. It can result into several types of illnesses, including gastroenteritis (GI), acute respiratory disease (ARD), eyes, ears and skin infections (PRÜSS, 1998; WHO, 2003). 3.2 Physico-chemical analysis The levels of chemicals in the well water samples screened are shown in Table 6. Table 6: Chemical properties of waters from sampled wells Parameter W1 pH 5.9 Apparent colour 11.5 Turbidity 3.5 Conductivity 257.5 Total hardness 42.4 Calcium hardness 67.8 Chloride 10.3 Silica 13.4 *STD = standard deviation

W2 7.1 16.31 3.2 1094 164.7 33.2 20.6 23.2

Samples W3 6.4 12.45 3.7 448.2 63.5 145.7 28.5 30.4

W4 6.51 14.3 3.4 542.4 123.3 88.7 30.4 26.7

W5 7.2 15.74 3.6 967.3 72.6 86.4 20.7 32.5

Mean 6.62 14.06 3.48 661.88 93.30 84.36 22.10 25.24

STD* 0.54 2.07 0.19 354.78 49.76 40.86 7.96 7.51

WHO defines 6.5-8.5 as the suitable range for the hydrogen ion concentration (pH) levels in water. The pH of the water samples ranged from 6.42 - 7.1. Hence, all samples are within the unobjectionable range. Colour as measured on the Hazen scale showed that samples were either within inoffensive range or unobjectionable range. None exceeded the defined water quality assurance limit. The well waters could generally be considered acceptable with regards to turbidity. WHO defines the total hardness as Calcium Carbonate in mg/l of a sample. The data indicated that the calcium hardness ranged from 33 – 146 mg/l and total hardness also ranged from 42 – 165 mg/l. According to APPA (1995), none of these values exceeded the permissible level of 500 mg/l. Chloride levels ranged from 10 – 30 mg/l. Chloride level could therefore be considered inoffensive or unobjectionable. The values of iron present in samples were negligible and therefore considered as traces.

237


4. Conclusions and recommendations It is evident from the above analysis that the physical qualities of the wells water were all within tolerable limits. The samples were also tasteless and odourless. The chemical tests also indicated that the levels of chemicals in the wells were within acceptable limits. In contrast, the biological tests indicated that the wells water was contaminated with bacteria and other micro-organisms. These were likely to have come from nearby pit privies and unprotected septic tanks and nearby unconfined aquifer. These were considered the only possible sources of pollution since the area had no other sources of contaminants. Wells with high bacteriological content need some chlorination or should not be used for drinking for health reasons. Also, policy makers at state and local government levels should ensure the formulation and implementation of guidelines on siting of wells and other sanitary services in other to safeguard the health of the people.

References th

APPA, 1990: Standard Methods of the Examination of water and waste water, 15 Edition. American Public Health Association, Washington D.C. 200p. th Buchanan, R. E. and N. E. Gibbons, 1974: Bergeys Manual of determinative bacteriology, 8 Edition. - The Williams and Wilkins Co. Baltimore. 1268p. Cowan, S. T., 1974: Manual for the identification of medical bacteria. - Cambridge University Press, Cambridge. Federal Government of Nigeria, 2000: Water Supply & Sanitation Interim Strategy Note, 38p. Harrigan, W. F. and B. McCance, 1976: Laboratory Methods in food and dairy microbiology. - Academic Press, London. 451p. Hutton, G. and L. Haller, 2004: Evaluation of the Costs and Benefits of Water and Sanitation Improvements at the Global Level, Water, Sanitation and Health, Protection of the Human Environment. - World Health Organisation, Geneva. Prüss, A., 1998: Review of Epidemiological Studies on Health Effects from Exposure to Recreational Water. - International Journal of Epidemiology, Vol.27, No.1, pp.1-9. WHO/UNICEF, 2006a: Joint Monitoring Programme for Water Supply and Sanitation. - http:// www.wssinfo.org/en/welcome.html, accessed October 2006. WHO/UNICEF, 2006b: Meeting the MDG Drinking Water and Sanitation Target: the Urban and Rural Challenge of the Decade. WHO/UNICEF, 2005: Water for Life: Making It Happen. - Joint Monitoring Programme report. WHO/UNICEF, 2004: Joint Monitoring Programme for Water Supply and Sanitation: Meeting the MDG Drinking Water and Sanitation Target: a Mid-Term Assessment of Progress. http://www.unicef.org/wes/mdgreport, accessed 20 March 2006. WHO, 2003: Guidelines for Safe Recreational Water Environments: Coastal and Fresh Waters,Vol.1. - World Health Organisation, 220 pages. WHO/UNICEF/WSSCC, 2000: Global Water Supply and Sanitation Assessment 2000 Report. nd Twort, A. C., R. C. Hoather and F. M. Law, 1974: Water supply, 2 Edition. Arnold, London.

238


Address of Authors: Akinro, A. O. and I.B. Ologunagba Department of Civil Engineering Technology Rufus Giwa Polytechnic P.M.B 1019 Owo. Ondo State, Nigeria E-mail: [email protected]


Seiten 239 - 248

Potential to enhance the extent of paddy cultivation using domestic and municipal wastewater harvesting – a case study from the dry zone of Sri Lanka U. S. C. Udagedara and M. M. M. Najim

Keywords Municipal wastewater, domestic wastewater, paddy cultivation, dry zone

Abstract Paddy cultivation is constrained due to shortage of rainfall or irrigation water supply. With the population growth, farmers have to increase the paddy production to meet the existing as well as future demand with the same or less land cultivated at the present. Wastewater from domestic and municipal sources which are harmless and that can be harvested can be diverted to irrigate paddy fields thereby increasing the extent of land under paddy. A study was conducted in selected Divisional secretariat divisions from Puttalam District, Sri Lanka in order to explore the potential increment of land that can be brought under paddy cultivation with the utilization of municipal and domestic wastewater. The wastewater generation was estimated based on the population and water consumption data. The extent of paddy land under minor irrigation and rainfed agriculture, potential asweddumized extent, land area cultivated in Maha (main) season and Yala (off) season and the areas harvested were collected from district agriculture department. It is noted that all the available lands were not cultivated and not harvested in both Maha and Yala seasons. Lands left without any production is high in the Yala season compared to the Maha season. Only a part of the land area is harvested from the total area sown. One of the major reasons for crop failure is shortage of supplementary irrigation water, inadequate rainfall and irrigation water. About 22% of the uncultivated lands in Maha can be brought under cultivation if 55% of the gray water generated is collected and diverted to irrigate paddy fields. In Yala season, only very small portion of uncultivated land (5% of uncultivated land) can be brought under cultivation with the gray water irrigation.

1. Introduction Rice (Oryza sativa L.) is the largest irrigated crop and ranks second only to wheat as the most extensively grown crop in the world. Rice is the staple food in Sri Lanka and paddy cultivation is practiced in 40% of the total agricultural lands. Paddy production requires a large quantity of water. With the increasing population, farmers have to increase the paddy production to meet the existing as well as future demand with the same or less land cultivated at the present. Farmers give up paddy cultivation

240


due to different problems faced by them such as water shortage, high cost of production etc. Water availability is a problem in the dry and the intermediate zones, especially during the dry spells. In additions, average paddy water requirement is about 1200 mm which is high when compared with other field crops. High paddy water demand is mainly met with supplementary irrigation water. Rainfed cultivation is practiced in areas where irrigation infrastructure is not available but most of these lands will not give good yields as of irrigated crops due to shortage of water. The water shortage is another issue which limits the irrigation water supply. The water shortage is fulfilled in some localities by using wastewater, treated or untreated. Wastewater agriculture has important environmental and economic advantages such as supply of reliable water quantities at water scarce periods, increase the crop yields by providing nutrients to plants, thereby reducing the need for artificial fertilizers, low-cost method for sanitary disposal of municipal wastewater, prevention or minimization of the pollution of rivers, canals and other surface waters and addition of organic mater to the soil and improvement of soil properties such as moisture holding capacity. Unawareness of existing nutrient levels in the wastewater and the temporal variation of these concentrations are some problems faced by farmers (RAJAPAKSHE and NAJIM, 2007). Traditionally, domestic wastewater generated from houses, especially grey water from kitchens, bathrooms and washrooms, from villages in Sri Lanka are flown along open unlined wastewater drains and collected in a pool. In the traditional system, there were many beneficial plants (leafy vegetables and medicinal plants) grown along these drains. These plants utilized the wastewater and the nutrients contained in the wastewater as an input for growth. This system of wastewater disposal and utilization was kept away from the black water disposal system so that the grey water system was not harmful in any means. The wastewater collected in the drains or canals get diluted with the storm water or irrigation water in some period but concentrated in some other depending on the water flows. With the increased population pressure, these water sources are contaminated with black water, industrial wastewater in some cases and wastewater from different service providers pausing different levels of threats. This results in much of the urban domestic (and industrial) wastewater being discharged untreated into urban waterways in spite of water quality guidelines and discharge standards being available in these countries. In Sri Lanka, wastewater or diluted wastewater is being used intentionally or unintentionally as a source of irrigation water from recent past (JAYAKODY et al., 2006; RAJAPAKSHE and NAJIM, 2007). This could be due to many reasons such as pollution of irrigation canals with wastewater, reduction of supply from reservoirs due to increased demands from other sectors (domestic and industrial sectors), increased intensity of cultivation, increased extent of cultivation, utilization of


241

nutrients in wastewater etc.. Wastewater use in agriculture in Sri Lanka would likely occur due to pollution of perennial water sources that were always used for irrigation, by urban wastewater and/or under water scarcity conditions (as in the dry zone), where farmers would use it if it were the only available source of water.

2. Objectives Lands left without any production is very high in the Yala season compared to the Maha season mainly due to water shortage. Wastewater generated from cities could be utilized in increasing the extents asweddumized if there is a possibility to divert city wastewater to those lands or diluted with available irrigation water to meet the total irrigation water requirement. In additions, the wastewater could be used in increasing the cropping intensity of already asweddumized lands. Therefore, a study was conducted in Puttalam district selecting all the divisional secretariat divisions to explore the potential increment of land that can be brought under paddy cultivation with the utilization of municipal and domestic wastewater.

3. Methodology Puttalam district where paddy is cultivated was selected for this study (Figure 1). Data from all the divisional secretariat divisions within the Puttalam district was collected from different sources such as provincial agriculture department, department of census and statistics and published literature. The extent of paddy lands under major (more than 80 ha command area) and minor (less than 80 ha command area) irrigation and under rainfed agriculture, potential asweddumized extent, land area cultivated in Maha and Yala seasons and the areas harvested were collected from the Department of Census and Statistics. Wastewater generation in each divisional secretariat division was estimated based on the population and water consumption data. The population data was collected from the Department of Census and Statistics (DCS) (2001) and the NWS&DB water consumption data was combined to estimate the quantities of probable gray water discharges except the black water. Assuming a consumptive use of 20-30% for water use in general, it is considered, in an average 75% of urban water returns as wastewater which is an internationally accepted figure for wastewater return flows. There can be variations in this amount according to geographical location of cities, climate, and people’s behavior, level of industrialization, urban population is served with pipe borne water etc.. According to the Sri Lanka Standard 745-2003 standard for “design and construction of septic tanks and associated effluent disposal systems”, the wastewater generation is 240-160 L/C/day out of which 75% is grey-water and

242


25% is black-water (SLSI, 2003). In an average, NSW&DB takes 120 L/capita/day as the water consumption figure. The NWS&DB water consumption and Sri Lanka Standard Institute grey-water generation data were combined to estimate potential available wastewater for re-use.

Figure 1: Location of Puttalam District


243

The paddy water requirement for Puttalam district is estimated using evaporation data from Puttalam district and the paddy crop coefficient considering the peak requirement at the mid season stage. It is assumed that the land preparation requirement is supplied mainly by rainfall.

4. Results and discussions 4.1 Puttalam district and the climatic zones Puttalam district is mainly located within the dry zone of Sri Lanka which is having an annual rainfall of 1450 mm to 2400 mm. The agroecological zones within the Puttalam district are DL3, DL 1f, DL 1b, IL 3, IL 1b and IL 1a. The major rainy season in DL 3, DL 1f and DL 1b is September to December. These agroecological zones record some showers during May to June. Other months are mainly considered as dry months. The agroecological zones DL 3 and DL 1f receives less than 800 mm of annual rainfall while DL 1b receives less than 900 mm of annual rainfall. IL 3 and IL 1b receive less than 1100 mm of annual rainfall while IL 1a receives less than 1400 mm. September to December is the main rainy season of the IL 3, IL 1b and IL 1a while May to June also experiences some showers. The rainfall received in a month within these rainy seasons varies widely and the rainfall received in some months is not sufficient to fulfill the crop water requirement. Irrigation supplement becomes essential during such periods. 4.2 Paddy cultivation in Puttalam district There are 18,712 ha of paddy lands asweddumized in Puttalam district out of which 7,152 ha are coming under major irrigation schemes, 9,718 ha are coming under minor irrigation schemes while the remainder of 1,842 ha are coming under rainfed paddy cultivation (Table 1). In Maha season 60.7% of the asweddumized area is cultivated and only 36% is cultivated in the Yala season. In the Yala season, only 46 % of the major irrigated areas are cultivated. The area under cultivation is very low for the minor irrigated areas (33%) and rainfed areas (15.3%). The water shortage is the major reason for the lowest extent under cultivation. Out of the areas cultivated, 99.1% of the areas are harvested during the Maha season as there is no crop failure due to water shortage. In the Yala season the extent of lands that were harvested is 97%, which means 3% of the cultivated area is affected by the shortage of water. The extent of land that is not sown in Maha season under the major irrigation, minor irrigation schemes and under rainfed cultivation is 25.7%,44.8% and 62.5%, respectively (Table 1). Under the major irrigation schemes the extent not sown is low that is mainly due to availability of sufficient supplementary water for irrigation under the irrigation tank or under the diversion scheme. The extent of land areas that were not brought during Maha could be mainly due to other reasons than that of water shortage.

244


During Yala season, the extents that are not sown under the major schemes increased from 27.1% to 54.1% which is a considerable increment (Table 1). This increment is mainly due to the shortage of water in the reservoir or the lack of water in the river for the diversion. Under the minor irrigation schemes, the extent of not sown lands increased from 44.8% to 66.4% which is mainly due to the insufficient irrigation water available in the minor tank system or the anicut scheme. The low rainfall in the area contributes to the lower water level in the minor reservoirs. The rainfed agriculture is the worst affected sector in the Yala season where 84.6 % of the lands are not cultivated (Table 1). This is mainly due to insufficient rainfall for a paddy crop. 4.3 Wastewater generation in Puttalam district The NWSDB takes 120 L/capita/day as the water consumption figure. According to the Sri Lanka Standard Institution (SLSI), the amount of gray water generated is 75% of the water used (SLSI, 2003). Considering these two values, the amounts of gray water generated at each Divisional Secretariat (DS) division is estimated and given in Table 2. Table 1: Asweddumized, cultivated and harvested paddy lands in Yala and Maha seasons in Puttalam district Asweddumized Sown Major (ha) (%) Minor (ha) Yala (%) Rainfed (ha) (%) Total (%) Major (ha) (%) Minor (ha) (%) Maha Rainfed (ha) (%) Total (%)

7,152 9,718 1,842 18,712 7,152 9,718 1,842 18,712

3,276 46 3,223 33 283 15.3 6,782 36 5,309 74 5,361 55 689 37.4 11,359 60.7

Not sown 3876 54.1 6459 66.4 1559 84.6 11930 63.7 1843 25.7 4357 44.8 1153 62.5 7353 39.2

Harvested 3,119 95 3,194 99 261 92.2 6,574 97 5,287 100 5,296 99 680 98.6 11,263 99.1

Not harvested 157 4.7 29 0.89 22 7.7 208 3 22 0.41 65 1.2 9 1.3 96 0.84

245


In Kurunegala DS division, the gray water generated is estimated as 8005 m3/day and the amount of gray water generated from the Kurunegala city is 4620 m3/day (RANAWEERA, 2005). That amount of gray water generated from the city alone is about 55% and this amount can be harvested / collected and diverted to irrigate paddy fields easily as those are flowing through storm water drains or natural streams that are passing through the cities. Table 2 also shows the 55% of the gray water that can be harvested and diverted from different DS divisions from Puttalam district. Table 2: Estimated amounts of wastewater generation from different Divisional Secretariat divisions in Puttalam district

DS Division Anamaduwa Arachchikattuwa Chilaw Dankotuwa Kalpitiya Karuwalagaswewa Madampe Mahakumbukkadawala Mahawewa Mundel Nattandiya Nawagattegama Pallama Puttalam Vanathavilluwa Wennappuwa Total

Population 33,302 38,092 59,890 59,386 81,780 20,225 43,522 16,905 48,861 56,294 57,686 12,956 22,410 71,091 16,460 70,817 709,677

Amount of gray water generation (m3 /Day) 2664.2 3047.4 4791.2 4750.9 6542.4 1618.0 3481.8 1352.4 3908.9 4503.5 4614.9 1036.5 1792.8 5687.3 1316.8 5665.4 56774.16

4.4 Paddy water requirement For a 110 paddy variety, the crop water requirement is 696 mm in Yala season and 325 mm in Maha season, respectively. The seepage and percolation loss will be about 4 mm/d which needs about 380 mm for a cropping season. The peak irrigation water requirement will be 9.88 mm/d in Yala and 3.56 mm/d in Maha in Puttalam. The crop water requirement is estimated based on the crop coefficients given in ALLEN et al. (1998) and average pan evaporation values in Kurunegala (IMBU-LANA et al., 2006).

246


Due to the dry climatic conditions, the crop water requirement in Yala season is more than the Maha season. This is one of the reasons why the extent of paddy cultivated is less during Yala as the available water in the reservoirs are limiting. The amount of rainfall received is also less during the Yala which has an influence in the low land extend cultivated. 4.5 Potential paddy cultivable area with gray water irrigation Considering the peak irrigation water requirement, the extent of extra land that can be cultivated in Yala and Maha is given in Table 3. About 22% of the uncultivated lands in Maha can be brought under cultivation if 55% of the gray water generated is collected and diverted to irrigate paddy fields. In Yala season, only very small portion of uncultivated land (5% of uncultivated land) can be brought under cultivation with the gray water irrigation. Table 3: Extent of extra land that can be cultivated with gray water irrigation from different DS divisions in Puttalam district DS Division Anamaduwa Arachchikattuwa Chilaw Dankotuwa Kalpitiya Karuwalagaswewa Madampe Mahakumbukkadawala Mahawewa Mundel Nattandiya Nawagattegama Pallama Puttalam Vanathavilluwa Wennappuwa Total

Harvested / Diverted Gray Water (m3/d) 33,302 38,092 59,890 59,386 81,780 20,225 43,522 16,905 48,861 56,294 57,686 12,956 22,410 71,091 16,460 70,817 709,677

Cultivable extent (ha) Yala Maha 27.0 74.8 30.8 85.6 48.5 134.6 48.1 133.5 66.21 183.8 16.4 45.5 35.2 97.8 13.7 38.0 39.6 109.8 45.6 126.5 46.7 129.6 10.5 29.1 18.1 50.4 57.6 159.8 13.3 37.0 57.3 159.1 574.6 1594.8


247

5. Conclusion Wastewater agriculture is practiced in very few localities in Sri Lanka, most of which are informal usages. There are few cases where wastewater is diverted to cultivate paddy but the extents cultivated is very small. This study shows that if the generated wastewater, specially the gray water from domestic water usage can be collected and diverted for agricultural production, additional areas can be brought under cultivation that are uncultivated due to water shortage. The land extend that can be brought under paddy cultivation is about 1594.8 ha in the Maha season and 574.6 ha in the Yala season. The authority has to make sure that the gray water which is diverted for agricultural production is not contaminated with black water and hazardous industrial wastewater in order to minimize the adverse health hazards the farmers and the consumers could face.

References Allen, R. G., L. S. Pereira, D. Raes and M. Smith, 1998: Crop evapotranspiration - Guidelines for computing crop water requirements. - FAO Irrigation and drainage paper 56. Food and Agriculture Organization of the United Nations, Rome. DCS, 2001: Census of Population and Housing 2001: Kurunegala District - Final Results (CD). - Sri Lanka: Department of Census and Statistics. Sri Lanka. Imbulana, K. A. U. S., N. T. S. Wijesekera and B. R. Neupane, 2006: Sri Lanka National Water Development Report. - UNESCO and Ministry of Agriculture, Irrigation and Mahaweli Development, Sri Lanka. Jayakody, P., L. Raschid-Sally, S. A. K.Abayawardana and M. M. M. Najim, 2006: Urban growth nd and wastewater agriculture: A study from Sri Lanka. - 32 WEDC International Conference, Colombo, Sri Lanka. Rajapakshe, I. H. and M. M. M. Najim, 2007: Water and Nutrient Balance in a Paddy Field Irrigated by Wastewater During Off (Yala) Season in Kurunegala, Sri Lanka, Journal of Applied Irrigation Science, 42 (1): 77-9. Ranaweera, R. A. D. U. P., 2005: Wastewater Generation, Its Quality and Conjunctive use for Paddy Irrigation in Kurunegala. Peradeniya, Sri Lanka. - BSc. Thesis Department of Agricultural Engineering, Faculty of Agriculture, University of Peradeniya. Sri Lanka. SLSI, 2003: Draft Sri Lanka standard - Code of practice for the design and construction of septic Tanks and associated effluent disposal systems. - Sri Lanka Standards Institution, No. 17, Victoria Place, Off Elvitigala Mawatha, Colombo 08. Sri Lanka.

248


Address of Authors: Dr. M. M. M. Najim (Corresponding Author) Senior Lecturer in Environmental Conservation and Management Faculty of Science, University of Kelaniya, Kelaniya. Sri Lanka Email: [email protected] U. S. C. Udagedara Environmental Conservation and Management Degree Program Faculty of Science, University of Kelaniya, Kelaniya. Sri Lanka Email: [email protected]


Seiten 249 - 265

Evaluation of Water Poverty Index in Ondo-State, Nigeria O. Yahaya, A. O. Akinro, O. Mojaji Kehinde and I. B. Ologunagba

Keywords Water Poverty Index, Ese-Odo, Water stress, Freshwater, Access, Dry season, Wet season, Household

Abstract An increasing world population exerts a continually growing demand on usable freshwater resource and matching the demand with supply of safe drinking water has resulted to serious social-economic constraints. Women and girls suffer more from this problem. They spend hours on daily basis trekking long distances in search for portable water. Time and drudgery involved to access safe drinking resulted to loss of human capital, thus affects nearly every household activity. This paper focuses on the evaluation of Water Poverty Index (WPI) as an integrated tool veritable for all the local government areas in OndoState of Nigeria to address their water sector. Simple time analysis and composite index approaches were employed to compute WPI values in all the sampled areas. Variables such as water resource, access to safe water, use of water and environmental impacts were considered. The results show that composite index gives more comprehensive output over simple time analysis, the former encapsulate more variables. There is a clear distinction between the WPI values obtained during the dry and wet season. Water stress is experienced more in dry season in all the investigated areas. The ranking of WPI values from the two approaches shows that Ese-Odo is the most water-stressed with least WPI values of 10.1 points (composite index) and highest value of 1.4minsl-1 (simple time analysis), while Owo, Ondo-West and Ose local government areas are less water stressed with WPI values of 0.55 minsl-1,17.8; 0.53 minsl-1,16.2; and 0.5 minsl-1,17.1 respectively. Dwellers enjoy fairly supply of freshwater in this region. The results obtained indicate that constructive investment in water and sanitation improves Human Development Index (HDI). However, this paper concludes that to prevent the occurrence of virtual water situation and improve water supply, researches of this nature should be conducted from time to time and government at all levels should holistically address the problem.

1. Introduction During the last few years, water has become an increasingly important issue in developing nations. In order to attain the Millennium Development Goals of halving the population of people without access to safe water by 2015, integrated water

250


management approaches are required. In monitoring the achievement of portable water at the local level, appropriate indicators are needed that allow measurement of progress of water sector for each community to be made (CLAUDIA, 2006). However this situation is very complex to explain in a simple language, therefore an index has been found to be a feasible way to express such complex condition (STEVEN et al., 2002). The Water Poverty Index (WPI) was identified as the possible indicator for monitoring progress at the local level as it puts access to water in a wider waterrelated context (SULLIVAN 2000, 2002). The index has been designed to identify and evaluate poverty in relation to water resource availability (STEVEN et al., 2002). Water shortages may relate to the inadequate ability of society to access the small volumes of water needed for drinking and domestic purposes. In most cases in developing world, women and children particularly girls spend most of their productive time trekking long distances sourcing for water. The evidence shows that women’s livehoods are constrained by being tied to sporadic and expensive water supply in urban slums or hours of water-fetching labour in rural areas (UNDP, 2004). The World Bank’s Global Monitoring Report states that although primary school enrolment rates are up, completion of primary schooling, especially for girls, remains a major concern. UNESCO reports that in one third of countries for which data was available, less than two thirds of children enrolling in primary education are reaching the last grade. Findings have shown that 1 in 10 girls still do not complete primary education (UNDP, 2004). Progress in health and education is dependent on access to affordable sanitation and safe water. Children, most especially (girls) educational prospects are similarly constrained. Public health systems are over-burdened by diarrhoeal diseases- the UN says that at anyone time; half the hospital beds in the developing world are occupied by patients suffering from diarrhoea and other water related ailments. (UNDP, 2004). In analysing the reasons for water problems, it is important to recognise that water scarcity can be considered in two ways. First order scarcity is the shortage of water itself, while second order scarcity is that resulting from lack of social adaptive capacity. The poor lack social adaptive capacity and this suggests that this aspect of development in the water sector is most pertinent to poverty alleviation (SULLIVAN et al., 2001a). Also, the poor frequently put affordable access to safe water and sanitation at the top of their priorities. Due to the acute water-stress condition in most of the developing world, this paper attempts to develop integrated water management tool veritable in running mathematical- based model useful in arresting shortage of water supplies. For the purpose of the study, Composite Index and Simple Time analysis approaches are employed to develop and test the Water Poverty Index (WPI) in all eighteen local government areas in Ondo-State, Nigeria.


251

2. Materials and methods 2.1 Study area Ondo-State is located in South-Western part of Nigeria, West-Africa between 70 14’ North and longitude 50 8’ East. The state has a surface area of 20,956 km2 of which 2,020km2 are covered by water. It shares boundaries with Kogi and Edo to the east, Oyo and Ogun and the south by the Atlantic Ocean as shown in Figure 1.

Figure 1: Map of Ondo State showing investigation locations The state has a population of 4,475,316 rising about 351m above sea level. OndoState enjoys tropical climate with two distinct seasons. These are the wet season (April-October) and the dry season (November-March). It lies in the rain forest zone with mean annual rainfall between 1300-1600 mm and with average temperature between 27.50-32.50C. The relative humidity ranges between 85% and 100% during the rainy season and less than 60% during the harmattan period. Several rivers run

252


through the state, of which the most important ones are the Owena River in the east (150km), the Awara (120 km) and Ogbese River (20 km). There are other smaller rivers and lagoons across the state. The WPI is mainly designed to provide a tool with which water engineers and managers can evaluate the water situation in different locations in a holistic way (CLAUDIA et al., 2005). However, in order to make strong comparison in the values of Water Poverty Index estimated in each of the selected communities, WPI values were computed using both adjusted Composite Index and Simple Time analysis approaches. 2.2.1 The Composite Index Approach In this approach, the index was constructed from a series of variables which captured the essence of what is being measured using national scale (Rodiya, 2007). A simple relationship was constructed for computing WPI taking into consideration all the key variables as follows: WPI • Wa A †W sS † Wt (100 ‡ T )

(1)

Where A is the adjusted water availability (%). The value of A should recognize the seasonal variability of water availability), S is the population with access to safe water and sanitation (%) and T is the index to represent time and effort taken to collect water for the household and WPI is the water poverty index. For the purpose of this study, (T) was modified to take account of gender and child labour issues as follows: (100-T). Since A, S, T are all defined to be between 1 and 100; Ws, Wt is 0.25 by weight and Wa is given 0.5. Therefore, Wa † Ws † Wt • 1.0

(2)

The relationship in equation (1) is finally modified as follows: 1 WPI • [Wa A †W sS † Wt (100 ‡ T )] 3

(3)

where Wa, Ws and Wt are the weight given to A, S and T respectively. 2.2.2 Simple time analysis approach WPI is constructed using bottom-up approach considering variables such as total time taken in collecting water including queuing time, volume of water collected in each trip. For the household with pipe borne water, the volume and time taken to collect water per head is assumed to the same across the members of the household. Using time-analysis approach, the index is determined as follows:


WPI •

T (min sl ‡1 ) V

253

(4)

where T is the total time (in minutes) spent per person in a day to collect water, while V is the volume of water collected in litres. Based on a reconnaissance survey of eighteen local government areas in OndoState, four most water-stressed towns in each of the local government areas were randomly selected for sampling purposes. Two hundred scientifically-structured questionnaires were randomly administered to 200 households in each of the 72 sampled communities in all the local government areas. Data obtained were subjected to statistical analysis to determine the following variables; percentage of people with access to safe water and not, water availability, total time spent to collect water and Human Development Index (HDI). Data were collected for wet and dry seasons for two consecutive years (2007 and 2008). 2.2.3 Human Development Index (HDI) The HDI gives a measure of social and economic progress which is built from an average of three separate indicators: life expectancy at birth; knowledge, as measured by the adult literacy rate (with a two-thirds weight) and the combined primary, secondary and tertiary gross enrolment ratio (with a one-third weight); and a decent standard of living, as measured by per capita GDP. HDI values for two consecutive years in all the local government areas were computed using Goalposts method of calculating the HDI developed by UNDP (2004) as follows: X i ‡ X min 1 • Xi X max ‡ X min

(5)

Where the Xi1 for all three indicators are measured to derive the HDI, while Xmax and Xmin are the maximum and minimum values of standardized dataset of indicator component such as life expectancy, adult, GDP per capita, education, health e.t.c

254


Plate 1: The students queuing for portable water from solarpowered borehole at Adekunle Ajasin University, Akungba, Nigeria

Plate 2: A lady fetching water from traditional well

Plate 3: Young girls queued up to fetch water from the municipal water supply

Plate 4: Reticulated solar-powered borehole

255


3. Results and discussions 3.1 Estimate of water poverty index using simple time analysis approach From a test-bed of collected data in 72 water-stressed communities in all the 18 local government areas in Ondo-State, WPI values were calculated using composite index and simple time analysis approaches. The summary of the dataset, showing average mean, total volume of water collected (TV), total time spent to collect water (TT), total number of local government areas (LGA), is presented in Table 1 and 2. Table 1: WPI values for all the local government areas in Ondo-State in wet season S/N

L.G.A

1. 2. 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14. 15. 16 17 18.

Akoko N-E Akoko N-W Akoko S-E Akoko S-W Akure-North Akure-South Ese-Odo Ifedore Ifelodun Ilaje Ileoluji/Okegbo Irele Odigbo Okitipupa Ondo-East Ondo-West Ose Owo

TT (min)

TV (litres)

WPI

110 130 140 110 102 105 240 130 130 226 130 200 180 150 125 89 90 100

200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

0.55 0.65 0.70 0.55 0.51 0.53 1.20 0.65 0.65 1.13 0.65 1.00 0.90 0.75 0.63 0.44 0.45 0.50

Source: Field Data The results of water poverty index (WPI) obtained during wet and dry season using simple time analysis approach presented in Table 1 and 2 showed that Ese-Odo local government has the highest value of WPI (1.20 minsl-1), while Ondo-West has the least value of 0.44 minsl-1 during the wet season as presented in Table 1. It was also found that Ese-Odo local government still recorded highest WPI value of 1.40 minsl1 , while Ose local government has the lowest WPI value of 0.5 minsl-1 during the dry season as presented in Table 2. This simple analysis shows that Ese-Odo local government is the most water-stressed region in Ondo State followed by Ilaje, while water stress was considered to be least at Ondo-West and Ose local government during wet and drying seasons respectively. During the consultation process, it was

256


discovered that the dwellers derive their drinking water from a variety of sources such as; direct withdrawal from pond, streams and river, traditional wells of up to 1.5-2.0 m diameter with local construction, modern wells that are usually filled with concrete in order to prevent outside contamination and seepage flow and reticulated solar-powered boreholes of cleaner and high quality water. However, the presence of Owena multipurpose dam reduced the water-stress condition of Ondo-West, OndoEast and Akure South local government. In addition, Awara dam serves Akoko N-E and some part of Akoko N-W, while Ose dam serves Owo and Ose local government areas as presented in Table 3. Aquifer in this region discharges sufficient amount of water which improves the yield of an average borehole in the above-named local government areas. Most of the faulty boreholes happened as a result of mishandling by dwellers, minor electric and mechanical problems which could be easily corrected by community project management team. Table 2: WPI values for all the local government areas in Ondo-State in dry season S/N

L.G.A

1. 2. 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14. 15. 16 17. 18.


TT (min)

TV (litres)

WPI

120 140 135 130 130 120 273 152 150 240 152 216 190 165 145 106 100 110

200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

0.60 0.70 0.68 0.65 0.65 0.60 1.40 0.76 0.75 1.20 0.76 1.08 0.95 0.83 0.73 0.53 0.50 0.55

Source: Field Data Water from pond, stream, river, sea and traditional well is generally considered unsafe for drinking. Due to the presence of abundant salty seawater at Ese-Odo, Irele and Odigbo local government areas, development of surface and underground water becomes a problem. Despite the huge financial resource expended on provision of portable water at Ese-Odo and Irele local government areas, majority of the boreholes were not functioning and most of the functional ones are not very good for drinking

257


as shown in Table 6. Finding also reveals that seepage of salty seawater into boreholes has contaminated most of the boreholes in the region and becomes highly unsafe for drinking. Thus, in turn make the development of underground water to be highly difficult and expensive. Converting seawater to safe drinking water either by desalination or any other processes has not been developed in this part of the world and this makes the exploitation of surface water impossible. Table 3: Source of functional safe drinking water in Ondo-State S/N

L.G.A

1. 2. 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14. 15. 16 17. 18.


No of solarpowered Boreholes 8 9 7 10 12 10 4 8 10 5 9 6 7 8 9 10 11 12

No of hand pump Boreholes 32 26 23 33 37 30 14 23 20 20 24 22 23 22 30 35 32 34

No of modern grouted wells 24 22 20 25 44 40 9 21 20 14 21 17 21 23 23 28 27 30

No of dams 1 1 1 -

Source: Data from the survey

3.2 Calculation of water poverty index using composite index approach Composite index approach draws on the structure and methodologies used by the Human Development Index, and it is based on the combination of relevant variable components collected and summed, to an index, based on the range of values on each variable in that location (STEVEN et al., 2002). Several indicators have been used to describe water availability or access and composite approach focused on water stress, water productivity, or crop productivity (CLAUDIA, 2006). The development of composite indexes combining these elements needs to be done in a transparent manner. To develop an appropriate and transparent indicator, standardized data set is required Due to the wide coverage of composite index approach; it is preferred to simple time analysis approach. Table 4 and 5 show the computed WPI values for wet and dry season using composite index approach.

258


Table 4: WPI values for all local government areas in Ondo-State during wet season S/N

L.G.A

Water Availability (%)

Access to Water (%)

TIndex

Index to time spent (100-T) Weight:0.25

WPI

30.0 35.0 50.2 56.1 29.3 28.6 93.6 49.9 57.3 91.6 75.0 86.7 80.1 75.3 36.7 29.0 25.0 26.2

70.0 65.0 49.8 43.9 70.7 71.4 6.4 50.1 42.7 8.4 25.0 13.3 19.9 24.7 63.3 71.0 75.0 73.8

15.9 14.6 15.0 15.5 19.1 20.9 13.5 13.9 14.4 13.5 15.4 13.6 14.3 14.5 17.1 20.9 20.3 19.6

Weight:0.25 Weight: 0.5 1. Akoko N-E 2. Akoko N-W 3. Akoko S-E 4. Akoko S-W 5. Akure-North 6. Akure-South 7. Ese-Odo 8. Ifedore 9 Ifelodun 10. Ilaje 11. Ileoluji/Okegbo 12. Irele 13. Odigbo 14. Okitipupa 15. Ondo-East 16 Ondo-West 17. Ose 18. Owo Source: Field data

59.6 28.7 40.3 45.9 50.1 57.8 74.6 33.2 40.4 71.8 60.2 68.2 67.9 65.5 43.4 59.6 50.2 48.1

56.1 53.2 51.0 49.8 58.1 63.2 6.0 50.1 49.3 10.0 39.8 14.0 16.0 18.0 55.2 60.3 68.3 55.1

Table 5: WPI values for all local government areas in Ondo-State during dry season S/N

L.G.A

Water Availability (%)

Access to Water (%)

TIndex

Weight:0.25

Index to time spent (100-T)

WPI

Weight:0.25

Weight: 0.5 1. 2. 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14. 15. 16 17. 18.

Source: Field data


30.3 25.1 24.0 22.6 41.6 32.6 55.1 29.6 26.1 50.3 45.6 49.3 37.2 45.3 40.3 40.1 40.1 42.1

52.1 49.6 49.8 42.3 50.1 60.0 5.3 48.2 40.3 10.2 33.2 20.8 30.6 30.1 51.6 53.2 60.2 62.1

40.5 46.9 46.4 50.1 44.6 35.4 94.6 48.9 56.7 82.4 65.9 78.6 68.9 69.6 42.0 39.6 35.2 32.4

59.5 53.1 53.6 49.9 55.4 64.6 5.4 51.1 43.3 17.6 34.1 21.4 31.1 30.4 58.0 60.4 64.8 67.6

14.4 12.7 12.6 14.5 15.7 15.8 10.1 13.2 11.3 10.7 13.2 11.7 11.3 12.6 15.9 16.2 17.1 17.8

259


The comparison of water poverty index using the composite index approach in Table 4 and 5 show that Akure-South and Ondo-West local government areas recorded highest WPI value of 20.9 (index point) each during the wet season. This indicator shows that the two local government areas experienced lowest degree of water stress. In addition, the water situation in Akure-South and Ondo-West can still be improved focusing on access to water, water availability and also on environmental aspects, particularly water recycling which have not been converted into use in these areas and Ondo State in general. Ese-Odo and Ilaje local government areas recorded the lowest WPI value of 13.5 (index point) each. The region is heavily water-stressed and special attention should be given to the use components and also to increase investment to improve access to water resource. The values of WPI obtained during the drying season period showed that Owo local government area has the highest value of 17.8 (index point), while Ese-Odo local government recorded the least value of 10.1 (index point). This development showed that Ese-Odo local government and its environs are strongly water-stressed at both dry and wet seasons, while OndoWest, Ose, Owo, Akoko N-E, Akoko N-W, Akoko S-W, Akoko-South and Akure North are generally less water-stressed with fair access to safe drinking water at all season. The summary of HDI values computed for year 2007 and 2008 in all the local government areas are presented in Table 6. Table 6: Ranking of Human Development Index in all the local government areas in OndoState S/N

L.G.A

1. 2. 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14. 15. 16 17. 18.


HDI 2007

Ranking 2007

0.34 0.30 0.28 0.34 0.35 0.35 0.15 0.30 0.30 0.19 0.30 0.22 0.24 0.25 0.32 0.47 0.41 0.37

5 7 8 5 4 4 13 7 7 12 7 11 10 9 6 1 2 3

Source: Data from the survey

HDI 2008 0.36 0.33 0.29 0.35 0.37 0.41 0.27 0.31 0.32 0.22 0.33 0.27 0.28 0.31 0.38 0.49 0.43 0.38

Ranking 2008 5 7 10 6 4 3 13 9 8 14 7 12 11 9 3 1 2 3

260


The ranking of computed HDI values showed that highest Human Development Index of 0.47 and 0.49 were obtained at Ondo-West in year 2007 and 2008, while lowest values of 0.15 and 0.22 were recorded at Ese-Odo and Ilaje local government areas for 2007 and 2008 respectively. Fig.2, 3, 4 and 5 show the relationship between Human Development Index (HDI) and Water Poverty Index (WPI) during dry wet and dry season for the year 2007 and 2008 respectively.

W ater Po verty In d ex (W PI)

25

y = 29.291x + 7.3049 R 2 = 0.707

20

15

10

5

0 0

0.1

0.2

0.3

0.4

0.5

Hum an deve lopm e nt Inde x (HDI)

Figure 2: Relationship between WPI and HDI during wet season in 2007 using composite index 20

y = 26.023x + 5.7885 R2 = 0.7615

Water Poverty Index (WPI)

18 16 14 12 10 8 6 4 2 0 0

0.1

0.2 0.3 Hum an developm ent Index (HDI)

0.4

0.5

Figure 3: Relationship between WPI and HDI during dry season in 2007 using composite index

261


25

y = 37.078x + 3.6569 R 2 = 0.8111

W a te r P o v erty In d e x (W PI)

20

15

10

5

0 0

0.1

0.2

0.3

Hum an developm

0.4

0.5

0.6

ent Index (HDI)

Figure 4: Relationship between WPI and HDI during wet season in 2008 using composite index y = 30.562x + 3.354 R2 = 0.752

20

Water Poverty Index (WPI)

18 16 14 12 10 8 6 4 2 0 0

0.1

0.2

0.3

0.4

0.5

0.6

Hum an development Index (HDI)

Figure 5: Relationship between WPI and HDI during dry season in 2008 using composite index The graphs show a gradual increase in WPI values with corresponding increase in HDI values for the two consecutive years during wet and dry season. This fairly strong relationship showed that increase in access to safe drinking improves socialeconomic and human development capacity of the communities. The increment of allocation spent on water and sanitation as presented in Table 7 resulted to increase in access to safe drinking water and Human Development Index (Table 6) in most of the local government areas in the state. The population of people that have no or poor access to safe drinking water was estimated for two concurrent years and the

262


result in Table 8 shows that Ese-Odo was ranked to be the highest with 94.7% and 89.2%, while lowest values of 37.9% and 35.9% for the year 2007 and 2008 respectively. This also explains further the degree of water stress status at Ese-Odo local government area despite the financial commitment on the provision of portable water between year 2007 and 2008 by government at every level and some donor agencies. However, fairly accessibility of portable water at Owo local government area and its environs is not enough to satisfy the water demand of the dwellers and so more technical and financial commitment must be invested to improve the volume of safe drinking water and the percentage of dwellers that can access it. Analysed data in Table 9 shows that the percentage of people that had no access to freshwater water reduced from 58.4% (2,613,584) to 54.8% (2,452,472) between the year 2007 and 2008 respectively. The reduction is strongly correlated to the investment in water and sanitation within the period of assessment. Table 7: Ranking of Investment in Water and Sanitation in all the local government areas in Ondo-State S/N L.G.A 1. Akoko N-E 2. Akoko N-W 3. Akoko S-E 4. Akoko S-W 5. Akure-North 6. Akure-South 7. Ese-Odo 8. Ifedore 9 Ifelodun 10. Ilaje 11. Ileoluji/Okegbo 12. Irele 13. Odigbo 14. Okitipupa 15. Ondo-East 16 Ondo-West 17. Ose 18. Owo Source: Data from the survey

Investment (N) 80,333,245.16 70,900,345.96 63,567,176.00 78,670,200.17 88,540,070.33 89,205,100.56 90,105,255.13 70,000,200.45 68,900,245.12 89,205,070.12 69,540,100.13 59,240,100.43 59,470,214.12 60,120,473.10 75,245,250.77 98,000,582.22 95,325,420.19 93,216,110.10

2007 8 11 15 9 7 5 4 12 14 6 13 17 18 16 10 1 2 3

Investment(N) 89,769,240.23 77,780,236.19 65,467,105.19 86,450,789.16 99,765,129.26 102,134,256.18 105,452,245.10 73,265,105.99 73,451,243.86 106,126,243.23 78,900,733.45 70,106,345.67 67,780,567.88 68,450,453.12 100,500,345.18 124,578,217.24 108,432,106.13 103,221,103.25

2008 9 12 18 10 8 6 4 14 13 3 11 15 17 16 7 1 2 5

263


Table 8: Ranking of population without access to safe water in all the local government areas in Ondo State S/N

L.G.A

1. Akoko N-E 2. Akoko N-W 3. Akoko S-E 4. Akoko S-W 5. Akure-North 6. Akure-South 7. Ese-Odo 8. Ifedore 9 Ifelodun 10. Ilaje 11. Ileoluji/Okegbo 12. Irele 13. Odigbo 14. Okitipupa 15. Ondo-East 16 Ondo-West 17. Ose 18. Owo Source: Data from the survey

Population (%) (2007) 47.9 50.4 50.2 57.7 49.9 40.0 94.7 51.8 59.7 89.8 66.8 79.2 69.4 69.9 48.4 46.8 39.8 37.9

Population (%) (2008) 43.2 47.3 49.6 53.3 45.9 36.3 89.2 49.1 55.3 84.7 66.2 75.2 65.1 65.3 40.8 46.9 36.2 35.9

Ranking 14 10 11 8 12 16 1 9 7 2 6 3 5 4 13 15 17 18

Table 9: Estimated average total population without access to safe water in all the local government areas in Ondo-State Year 2007 2008

Average total population(%) 58.4 54.8

Estimated population 2,613,584. 2,452,472

4. Conclusions The study evaluated water poverty index using two approaches and also some other index such as human development and related the finding together to determine the degree of water stress in all the local government areas in the state and recommend realistic measures to address the pathetic situation. The results obtained from the two approaches indicated that Ese-Odo, Ilaje and Irele local government areas are the most water-stressed region coupled with low Human Development Index in the state, while areas such as Ose, Owo, Ondo-West, and Ondo-East local government areas have fair access to portable water and improved Human Development Index. Heuristic application of composite index approach to test the generated dataset

264


provided flexible and strong decision-making strategies in such a way as to construct a holistic water management tool to address the problems of poverty, and its relation to water access and use. However, simple time analysis cannot link complex multidimensional aspects of water management together as a result of this, composite approach is always preferred. The results presented here using various approaches to test our standardized data sets are expected to enhance our understanding of the significant effects of water poverty to economy, human development, health and education. Many states and local government areas are moving towards a point where water resources are insufficient for agriculture, drinking and other domestic uses and to prevent the occurrence of virtual water, further researches are needed to be conducted from time to time on water problems and proffer realistic and technical solution to enhance the supply of safe drinking water at reasonable distance in all strategic locations across communities and regions.

References Claudia, H., 2006: Development and evaluation of a region water index for Benin. NPC, 2005: Nigerian Population Commission, 2005 census in Nigeria. Rodiya, A. A., 2008: Estimates of water poverty index in Ekiti State. - M.Eng. Thesis. Federal University of Technology, Akure, Nigeria, pp 25-30. Steven, D. M., S. Caroline and M. Jeremy, 2002: Water poverty index: a tool for integrated water management. Sullivan, C. A (ed.)., 2001a: The development of a water poverty index: A feasibility Study. The Central for Ecology and Hydrology (Wallingford). Sullivan C. A., J. R. Meigh and P. Lawrence, 2005: Application of the water poverty index at different scales: A cautionary tale. - Agriculture, Ecosystems and the Environment. Special issue. UNDP, 2004: Human Development Report 2004. Cultural liberty in today’s diverse world. New York: UNDP


Address of Authors: Olotu Yahaya Right Foundation Academy P.O. Box 50 Ikare-Akoko Ondo-State, Nigeria E mail: [email protected] A. O. Akinro, A. O. Department of Agricultural Engineering Federal University of Technology Akure, Nigeria O. Mogaji Kehinde Department of Civil Engineering Rufus Giwa Polytechnic Owo, Nigeria I. B. Ologunagba Department of Agricultural Engineering Rufus Giwa Polytechnic Owo, Nigeria

265