Hydrological Modelling of the. Mountainous ...

CC-HYDRO .

."

....

Impact of Climate Change on River Basin Hydrology under Different Climatic Conditions

Hydrological Modelling of the. Mountainous Mesochora Catchment

by

DIONYSIA PANAGOULIA National Technical University of Athens, Department of Civil Engineering, Division of Water Resources, Hydraulic and Maritime Engineering, 5, Iroon Polytechniou, 15780 Athens, GREECE. e-mail: [email protected]

and GEORGE DIMOU Dr civil engineer, 7 Voutyra, 16673 Voula, Athens, GREECE e-mail: [email protected]

November 1998

INTRODUCTION

1

1. SELECTION OF THE BASIN

2

1.1 DESCRIPTION OF THE MESOCHORA CATCHMENT 1.1.1 General remarks 1.1.2 Topography 1.1.3 Geology 1.1.4 Soils 1.1.5 Vegetation 1.1. 6 Climate 1.2 COLLECTION OF PRECIPITATION, TEMPERATURE AND RUNOFF DATA 1.2.1 Precipitation 1.2.2 Temperature 1.2.3 Wind, Sunshine and Humidity 1.2.4 Snow conditions 1.2.5 Streamflow

3 3 4 6 7 7 7 9 9 9 10 10 11

2. HDROMETEOROLOGICAL DATA: PROCESSING AND BASIC STATISTICS

11

2.1 STATISTICAL CHARACTERISTICS 2.1.1 Precipitation 2.1.2 Temperature 2.1.3 Wind 2.1.4 Streamflow

11 11 12 13 13

2.2 DATA PROCESSING 2.2.1 Precipitation 2.2.2 Temperature 2.2.3 Streamflow 2.2.4 Wind 2.2.5 Sunshine 2.2.6 Humidity 2.3 STATISTICS OF PROCESSED DATA

13 13 14 14 15 15 15 15

3. DETERMINISTIC CONCEPTUAL MODELS: THE US NWSRFS SNOWMELT AND RAINFALL-RUNOFF MODELS

15

3.1 SNOW ACCUMULATION AND ABLATION MODEL (SAA) 3.2 SOIL MOISTURE ACCOUNTING MODEL (SMA) 3.3 INPUT DATA 3.3.1 Precipitation and temperature stations-data 3.3.2 Reference evapotranspiration 3.3.3 Streamflow 3.4 MODEL CALIBRATION 3.4.1 SAA model parameter estimation 3.4.2 SMA model parameter estimation 3.4.3 SMA model validation

17 19 20 21 22 23 23 23 26 28

3.5 RESULTS 3.5.1 Snow Water Equivalent 3.5.2 Runoff 3.5.3 Evapotranspiration 3.5.4 Soil moisture storage

32 33 33 35 37

3.6 Conclusions

41

4. REFERENCES

42

INTRODUCTION The modelling of a mountain hydrology is the most challenging object of hydrological simulation, since measurements of meteorological variables in mountainous regions are the most difficult to make and the processes governing mountain hydrology cover the greatest range of demands on theoretical understanding. Regarding the meteorological variables, three major problems are posed: the accessibility to the mountains on a continuous basis, the accuracy of measured meteorological variables, and the areal representativeness of measurements (Klemes, 1990). The problem of accessibility is not connected only to measurements involving water (i.e. rainfall, snow cover and snow water equivalent, etc), but also the measurements of surface energy (at least temperature), wind speed and all other phenomena in the boundary layer. In mountainous areas the accuracy of measured meteorological variables is lower than it is in flat regions. The high slopes and strong wind affect the catch of the true precipitation by the gauges, while harsh conditions cause instruments to malfunction more frequently, thereby leading to gaps in the records, as well as to non-homogeneity, when instruments have to be changed or recalibrated. Even if measurements could be made where they were necessary and their accuracy could be submitted to control, the problem of their areal representativeness would still remain a formidable obstacle for an accurate description of meteorological variables. The areal representativeness of point measurements is a general and pervasive problem in hydrology and especially in mountainous one. The above cited problems are associated with mountain hydrology models inputs. The input modelling from scarce and ineffectively located point measurements of precipitation and energy components, as well as the areal and elevation distribution of precipitation amounts and temperature degrees, represents the most important and elaborate part of a mountain hydrology model (Klemes, 1990). In this study, the input data of mountain hydrology models are surface- and elevationWise assessed from incomplete point records (precipitation and temperature) with a combinatorial technique of the Thiessen method and available station data (Panagoulia, 1991a). The so obtained areal information is corrected for elevation variation. By this double method we can preserve the real nature of meteorological information which is, also, released from errors that indroduce the filling up data techniques (Georgakakos & Krajewski, 1991; Wallis et Panagoulia & Dimou Hydrological Modeling of the Mountainous Mesochora Cautclunent

1

al. 1991). The purely statistical filling methods are not totally reliable nor promising (i.e. the climatological kriging methods of Delfiner & Delhomme, 1973; Chua & Bras 1982; Kitanidis, 1983; Hevesi et aI., 1992 and Phillips et aI., 1992), because they make partial use only of the

global statistical information contained in the overall precipitation and temperature data. Furthermore, the statistical methods cannot preserve the physical structure of meteorological data. The used method can also handle successfully any change in the gauge network, that was the greatest limitation of the classic Thiessen method (Linsley et aI., 1988). As regards the theoretical understanding of the mountain hydrology it includes the disciplines of hydrology, climatology, boundary layer meteorology, geophysics and geology, as well as the interaction of these processes. The models that simulate the mountain hydrology are the conceptual and direct parameterization ones. While there is a great discussion about the intercomparison of snowmelt-runoff models (WMO, 1986), the deterministic conceptual models, despite the lack of a sounder physical basis, remain attractable for medium-size catchments due to the detailed hydrological description of the catchment. The deterministic conceptual models used in this work to simulate the hydrology of a medium-size mountainous catchment (in this instance the 633 km2 Mesochora catchment in Central Greece) and which assume as inputs data over area and elevation integrated with the aforesaid mentioned method, are the US NWSRFS-Snow Accumulation and Ablation Model, and the Soil Moisture Accounting Model. The results of this simulation have determined the ability of US NWSRFS- snowmelt and soil moisture models to represent the hydrological dynamics of a medium-size catchment during dry to wet transitions and extreme rainfall. The .snowmelt model predicted successfully the initiation of snow accumulation in the fall and the gradual melting of the snowpack in the late winter and spring, while the rainfall-runoff model, which accepted as input the snowmelt model output "rain plus melt", accurately reproduced also accurately both the magnitude and timing of the annual and monthly runoff. On a, daily basis, the runoff model reproduced satisfactorily the historic data, while some discrepancies arose due to antecedent dry conditions and extreme rainfalls (Panagoulia, 1992b).

1. SELECTION OF THE BASIN The catchment features and the general characteristics of the hydrometeorological data inside and around the study catchment are described in this section. Furthermore, the reasons of catchment selection are analyzed.

Panagoulia & Dimou Hydrological Modeling of the MOlmtainous Mesochora Cautchment

2

1.1 DESCRIPTION OF THE MESOCHORA CATCHMENT 1.1.1 General remarks

The Mesochora catchment, which is drained by the upper Acheloos river, was selected for analysing the basin hydrology. The catchment with an area of about 633 km2 lies in the central mountain region of Greece (Fig. 1) and extends nearly 32 km from north (39 0 42') to south (39 0 25') with an average width of about 20 km. The Pindus Mountains, with peak elevations of about 2300m, form the western boundary of the catchment, while the eastern one formed by

! )

)

N.\$)

I

FORMER Y U GO S L A V I A '

,,-./' r',J

\

\ '.

'"\(~'-

ALBA N I A

"\. .

"

"

BULGARIA

l

I'''''''

\.

\.,/".

,...

\

)

,

1

/'

'\...........

'-'-'~ ,J

j

.,I" GREECE

,

,

'"

.r'

. ~"'IIIr.i:

.~ ARDIT~

•

o ".

"1-

f.>

o

SEA

OF

CRETE

tXlkm

Figure 1. Localtion of Me sochora catchment

Panagoulia & Dimou Hydrological Modeling of the Mountainous Mesochora Cautchment

3

the Koziakas Mountains with peaks of about 2000m. Inside the catchment presents intense topography with strong interchanges of lower and higher elevations. The mean elevation is 1390m, while the elevation from the highest point to the catchment outlet ranges from 2200 m to 780 m. The wild Acheloos river traverses the catchment formed by a number of lower and higher order streams. The length of the main stream of the Acheloos river is about 28 km. The hydrology of the catchment is controlled by snowfall and snowmelt at high elevations. The catchment mean annual precipitation (weighted average over elevation bands) is about 1900mm and its mean annual runoff is 1140mm (22.8 m3/s). The mean January and July daily temperatures (weighted average over elevation bands) are 0.10 and 17.SoC correspondingly (Panagoulia, 199h, 1992a,b). Daily temperature and precipitation data are available at 4 temperature and 11 precipitation stations installed within and around the Mesochora catchment. In addition, monthly sunshine and humidity data are available at 3 and 2 stations correspondingly, which are installed outside but near to the catchment boundary. The daily streamflow data are available at the Mesochora gauge station which lies at the outlet of the catchment (Fig.2). The Mesochora catchment is of great significance for Greece because the river will be partially diverted at the outfall of the catchment through Pindus mountains to irrigate the arid Thessaly plain. The river's water will be also used to boost hydropower generation in the surrounding area. It is the largest project in Greece, comprising five dams (one is Mesochora's dam), 40 kilometres of large tunnels and about 8000 kilometres of buried irrigation pipes. The operation and performance reliability of the costly hydraulic works largely depends on the hydrological regime of the catchment through its streamflow. Obviously, a downward alteration to the catchment hydrology, resulting from climate change, would have major effects on the reliability of water resources management and generally on the social and physical environment of the region. In addition, Mesochora catchment was selected for study purposes because there is no upstream diversion or flow regulation of the river which is rather important since the effects of climate change on hydrological regimes are also affected by man-made interferences in the river flow. 1.1.2 Topography

The Mesochora catchment located between the Pindos and Koziakas mountains presents many elevation peaks such as the Tsiasianika (el. 1943m), Kapo Graso (el. 1930), Pahtourniasina (el. 1976), etc (PPC, 1972). In the catchment the elevation range is in the youthful stage of the

Panagoulia & Dimon Hydrological Modeling of the MOlUltainous Mesochora Cautchment

4

geomorphic cycle, and as it is characteristic of this stage, streams with steep gradients and sharply incised canyons are common. Surface drainage is usually flowing south parallel to the elevation structure or roughly east-west as joint on fault controlled streams.

21'10'

1'05'

+

21'15'

21"30'

21' 25'

21'20'

METSOVO

ME SOCHORA CATCHMENT ED

PRECIPITATION GAUGE STATIONS

~

PRECIPITATION TEMPERATURE

- ' - ' -CATCI+o4ENT

S> •

I

AND STATIONS

eouNDARY

STREAt*1.OW

STATION

SUN9-IINE

STATIONS

+ HUMIDITY

STATIONS

I

3~' 40'1

....===-..-===3a__ km o

2

4

6

a

10

IOANNINA

39"35'

39'35'

a MATSOUKI

+ PRAMANDA

39"30'

~2!1'

21' 25'

21· 30'

Figure 2. Topography of Me sochora catchment and recording stations


5

The development of the present topography of the catchment area apparently started during the Pleistocene epoch when glaciers occupied the northern part of the Pindos mountains. In the study area, broad valleys which have been completely destroyed by subsequent erosion were apparently the dominant topographic form. Later episodes of mountain uplift have resulted in stream rejuvenation and formation of deeply incised canyons. The final episode of uplift appears to have occurred after the streams had established their grades at an elevation of 100 to 150m above· present stream grades. This is demonstrated by the change in topography which commonly occurs at this level. Of practical importance related to this final episode of uplift is the contrast in rock condition between the «higher level» and the «near river level» rock. Generally, «higher level» rock has experienced moderate to severe degradation due to extensive weathering which is in contrast to the relatively unweathered «near river level» rock. This last episode of uplift caused to be significantly cut to depths of approximately 25m below their present grade as evidenced by the depth of fluvial gravel deposited in the major streams of the area. To a degree, this dates the last period of downcutting by the streams and provides a mean of measuring the intensity of erosinal processes. The main course of the Acheloos follows the north-northwest structural trend of the Pindos mountain range. The very extensive Mesochora thrust fault forms a boundary between the rugged limestone and chert mountains and the more subdued topography of the treecovered flysch. At the site of the Mesochora dam the river, which often flows parallel to the regional trend, cuts across one of the thrust units, 3km west from the major Mesochora thrust fault, to follow an east-west joint controlled direction. The site is located in a deep gorge cut through limestone beds. 1.1.3 Geology

The geological structure of the Mesochora catchment area is represented regionally by four thrust sheets delineated by the Mesochora, Nea Petki and Avgo thrust faults (PPC, 1972). The thrust sheets, i.e. sheets bounded by thrust faults vary in thickness and structural detail, but they are alike in broad outline and composition, and they are generally repetitive. The normal sequence of rocks in a thrust sheet of the Upper Acheloos region is, from top to bottom, Eocene Flysch, Cretaceous limestone, Jurassic limestone and rocks of the Radiolarite series and Triassic limestone.


6

The thrust faults have developed commonly in the rocks of the Jurassic formations, particularly in the Radiolante series, due to the shale-chert constituents, or in the Triassic limestone, probably due to interbedded shales. In many places thrust sheets of limestone and other older rocks have overridden the youngest fotmation of the Pindos zone, the Flysch. The thrust sheets generally dip to the north-northeast, causing ridges to have a hog-back profile with a gentle dip slope and a steep scarp. Thrust sheets are intensely folded with principal fold axes parallel to the north-northwesternly strike, and gentle secondary fold axes normal to the regional strike. The major thrust faults of Mesochora, Nea Pefki and Avgo are subparallel and trend north-northwest. The main course of the Upper Acheloos River has been controlled by the linear thrust fault with the Jurassic Radiolarite series forming the left bank and the foreground of the thrust, the Flysch and underlying Cretaceous limestones forming the right bank. 1.1.4 Soils

The soils of the Mesochora catchment have been formed from decay of hard limestones and flysch (clayish, psammitic and mixed). Over these weathered mass of rocks, rendzinas of variable colour, brown forest soils and litho soils have been formed under the influence of meteorological and biological factors (M.E.P.PW, 1995). 1.1.5 Vegetation

The formation and development of the vegetation in the Mesochora catchment is strongly depended on temperature and precipitation. Taking into account: (a) the biological mean temperature and the extreme temperatures mainly of the colder month, and (b) the mediterrenean bioclimate, two classification floors have been developed, i.e. the Emberger and UNESCO - F AO levels. According to these levels and vegetation observations, mountainous oak trees are observed in the central and north areas of the catchment with high elevation, while beech trees are met in the areas near to the streams. 1.1.6 Climate

The climate of the Mesochora catchment is between mediterranean and continental (middle european) type (M.E.P.PW, 1995). In the mountainous areas, the continental type presents features similar to the climatic

type of Middle-Europe, which are the long period of snowfalls, extensive cloudiness, plentiful summer storms and numerous days with hail.


7

2400

2200 2000 1800 1600

E

800

50~--~----L---~---L--~~--~--~--~----~--~

o

10 Area

2 above

30

40

elevations

50

60

70

so

90

100

(0/0)

Figure 3. Hypsometric curve for Mesochora catchment In the flat areas, the continental type presents clearly continental characteristic, i.e. large temperature range, very low temperatures in winter, moderate height of rainfall with uneven annual distribution, and strong winds with small duration during the spring and autumn. Due to the relatively large elevations two distinct seasons appear, the winter and summer. Rainfalls occur until the summer, while dry and warm atmospheric conditions there are only during June, July and August. The catchment area covers a large range of elevations (Fig. 3) and due to the strong relation of precipitation, snowfall and snowmelt to the elevation three elevation zones are distiguinshed (Table 1) (Panagoulia, 1992b). The microclimate of the catchment is elevation-dependent with hot summers and mild winters at low elevations and mild summers and cold winters at high elevations.


8

Table 1. Elevation zones lZone tupper IMiddle !Lower

Elevation range Median elevation (m) (m) 1580-2200 1830 1400 1280-1580 780-1280 1080

Area (%) 30.54 29.51 39.95

1.2 COLLECTION OF PRECIPITATION, TEMPERATURE AND RUNOFF DATA 1.2.1 Precipitation

The precipitation data have been obtained from observations of eleven stations mainly situated within the catchment. The stations are listed in Table 2 together with their coordinates, elevations and data recording period.

Table 2. Precipitation stations Station

Water Division

River

Katafyto

West Cent Greece

Acheloos

Matsouki

Epirus

Arachthos

39° 34'

21° 10'

1079

1967-92

Mesochora

West Cent Greece

Acheloos

39° 29'

21°20'

780

1962-92

Metsovo

Epirus

Arachthos

39° 46'

21° 11'

1157

1950-92

Pahtouri

West Cent. Greece

Acheloos

39° 28'

21° 15'

950

1960-92

Paleochori

Thessaly

Pinios

39° 37'

21° 25'

1050

1960-92

Pertouli

West Cent. Greece

Acheloos

39° 33'

21° 28'

1160

1952-92

Stoumareika

Thessaly

Pinios

39° 28'

21° 29'

860

1960-87

Tyma

Thessaly

Pinios

39° 31'

21° 32'

900

1951-92

Vakari

West Cent. Greece

Acheloos

39° 30'

21° 22'

1150

1959-92

Vathyrema

West Cent. Greece

Acheloos

39° 27'

21° 25'

920

1960-92

Latitude Longitude Altitude Data Period (m) 980 1954-92 39° 38' 21° 15'

1.2.2 Temperature

Four temperature stations are installed within and around the Mesochora catchment. The stations are listed in Table 3 together with their coordinates, elevations and data recording period.


9

1.2.5 Streamflow

The daily streamflow data are obtained from the Mesochora gauge station situated at the outfall of the catchment. The coordinates, elevation, data recording period, and mean annual observed runoff of the station are listed in Table 7.

Table 7. Sreamflow station Station

Water Division

River

Latitude Longitude Altitude Data Period (m) Mesochora West Cent Gree Acheloos 39° 29' 1972-92 780 21° 20'

Mean (m3/s) 22.8

2. HDROMETEOROLOGICAL DATA: PROCESSING AND BASIC STATISTICS 2.1 STATISTICAL CHARACTERISTICS

The collection of the hydro meteorological data has already described in the first section. The basic statistical characteristics (mean and standard deviation) of the raw data for all the described stations are presented here on a monthly and annual basis. 2.1.1 Precipitation

The mean and standard deviation of monthly precipitation series calculated for the existing data are given in Tables 8 and 9 respectively.

Table 8 Mean monthly and annual precipitation (mm) March

Ma,

Station

January February

Katafyto

188.69 178.40 126.50 104.72 80.62 51.16 34.83 33.45

64.45 132.98

210.98

219.48 1438.75

Matsouki

216.23 212.30 160.80 143.34 118.39 65.42 53.97 47.30

70.15 156.66

232.10

257.26 1782.7

Mesochora

228.23 211.12 161.39 128.37 89.62 40.64 34.07 30.8

67.65 155.05

239.69

301.73 1715.52

Metsovo

171.34 160.90 128.65 136.92 112.71 63.52 44.34 42.54

71.19 140.69

196.95

193.81 1465.60

'Pahtouri

288.04 259.75 190.60 154.15 108.80 49.42 36.35 37.74

66.68 18l.60

292.57

355.14 2043.46

'Palaichori

155.43 141.26 114.61 108.08 78.43 42.71

32.34 28.47

52.28 129.23

201.18

20l.55 1273.91

'Pertouli

185.55 172.05 140.14 123.01 90.49 53.79 39.99 32.22

63.67 158.69

214.13

226.08 151l.55

Stournareika 232.06 217.19 181.18 150.13 97.01 44.08 31.09 32.03

73.55 175.02

26l.90

315.59 1813.67

Apri

July Augus1 September Octobe November December

Annual

Tyrna

189.5 1

44.83 31.86 34.96

71.56 229.27

243.63

248.78 1675.72

Vakari

215.11 192.89 153.59 129.25 91.78 43.89 34.38 30.63

64.46 16l.58

244.82

293.90 1662.95

Vathyrema

247.58 244.43 185.51 160.23 109.25 51.04 31.83 32.68

69.74 185.98

287.67

363.24 2006.53

174.85 177.99 139.43

99.5 1

June


11

Table 9 Standard deviation of monthly and annual precipitation (mm) Station

January February

Marc!

~atafyto

113.26 116.49

77.19 51.31 36.74 26.93 26.99 27.87

45.04

81.83

107.88

115.65 300.97

lMatsouki

158.05

92.46 69.01

59.41 44.91 46.62 26.63

51.18 115.41

105.07

167.25 290.00

lMesochora

166.78 131.04 105.58 80.42 48.24 24.08 21.94 20.37

49.78 117.02

122.74

179.22 337.99

lMetsovo

106.95

47.38

78.86

90.32

96.80 341.92

lPahtouri

202.74 129.58 131.08 95.25 54.62 34.14 31.67 23.66

50.70 146.38

194.90

230.29 462.18

Pa1eochori

101.03

77.95

65.12 75.24 42.12 33.05 26.58 28.44

54.60

73.43

109.30

115.91 263.73

Pertouli

102.50 106.33

73.36 77.87 54.23 31.16 32.95 31.58

62.47

88.90

104.88

116.28 270.30

Stournareika 128.95 110.36

79.08 87.34 43.94 27.34 27.34 24.68

56.88 113.53

148.95

164.22 395.84

Tyrna

112.42 104.52 111.67 110.71 67.62 34.07 37.77 46.56

70.64 181.78

123.20

143.06 385.60

Vakari

143.70 113.00

91.32 66.02 46.72 27.80 27.27 22.04

46.99 100.01

146.02

157.92 356.70

Vathyrema

181.07 162.22 119.43 77.70 52.01 32.84 30.11 25.25

54.89 117.74

151.37

212.64 553.36

97.30

94.99

April

May

June

July Augusl September October November December

76.98 60.57 48.26 39.96 39.56 33.76

Annua

2.1.2 Temperature The mean and standard deviation of monthly min/max series of temperature calculated for the existing data are given in Tables 10-13.

Table 10 Mean minimum monthly and annual temperature (ae) Station Ioannina Metsovo lPahtouri Vakari

January February

0.05 -1.18 -0.06 -1.54

1.12 -1.01 1.35 -0.63

March

April

May

June

July

3.37 0.91 3.40 1.24

6.10 4.44 5.97 4.75

9.53 8.93 9.40 8.33

12.64 12.33 12.44 11.14

14.88 15.06 14.09 13.53

Augusl September October Novembe

14.97 14.7'1 13.59 12.94

12.28 11.89 11.78 10.21

8.66 7.80 8.25 6.83

Decembe

4.68 3.61 4.52 3.32

Annual

1.43 0.41 1.32 0.07

7.48 6.58 6.78 5.74

Octobe November Decembe

Annual

Table 11 Standard deviation of mean minimum monthly and annual temperature cae) Station Ioannina Metsovo Pabtouri Vakari

January February

2.1 1.35 1.10 1.45

1.96 1.78 1.62 1.80

March

April

Ma'y

June

July

1.39 2.21 1.88 1.90

0.91 1.53 1.76 0.97

1.05 1.26 1.15 0.95

1.42 1.32 0.89 0.82

1.04 2.03 1.15 1.12

Augusl Septembe

1.13 1.57 1.48 1.43

1.26 1.59 1.67 0.96

1.46 1.88 1.89 1.88

1.88 1.85 2.40 2.77

1.74 1.34 1.24 2.65

0.62 1.19 1.21 1.15

Decembe

Annual

Table 12 Mean maximum monthly and annual temperature (ae) Station Ioannina Metsovo Pabtouri Vakari

January February

8.82 4.66 6.50 7.28

10.40 5.03 7.97 8.41

March

April

Ma,

June

Jul;

13.72 7.84 11.62 11.52

17.58 12.01 14.68 16.13

22.68 17.32

26.92 21.11 26.57 23.26

30.66 24.37 30.29 25.68

21.0~

20.74

Augusl September Octobe Novembe

30.69 24.06 29.14 24.68

26.61 20.86 24.50 21.07

20.65 15.30 17.49 15.29

14.55 9.63 11.46 11.82

9.79 6.12 7.95 8.31

19.42 14.09 16.96 16.01

Table 13 Standard deviation of mean maximum monthly and annual temperature (ae) Station Ioannina Metsovo Pahtouri Vakari

January February

1.47 1.50 2.59 1.80

1.61 2.09 1.66 2.33

March

April

May

June

Jul;

2.00 3.05 2.12 2.91

1.71 2.15 2.83 2.70

1.96 1.48 2.28 2.21

2.45 1.29 3.82 1.32

1.62 1.61 3.68 1.64

Augusl September October Novembe

1.70 1.43 3.70 1.24

1.76 1.83 2.50 1.41

1.60 1.77 3.15 2.06


1.50 1.73 2.23 2.29

Decembe

1.31 1.62 1.79 2.44

Annual

0.67 1.00 1.67 1.66

12

2.1.3 Wind The annual frequency (%) of wind direction and forces in Beaufort scale from observations over the whole period (1956-1993) are presented in Table 14. Table 14 Annual frequency (per cent) of wind direction and forces in Beaufort scale. BEAUF 0 1 2 3 4 5 6 7 8 9 10 >ll SUM

N

NE

E

SE

S

SW

W

NW

0.45 1.36 1.02 0.285 0.066 0.022 O.Oll O.Oll 0.000 0.000 0.000 3.225

0.153 0.559 0.603 0.625 0.482 0.296 0.153 0.066 O.Oll O.Oll 0.000 2.959

0.175 0.899 1.063 0.789 0.47 0.296 0.164 0.088 O.Oll O.Oll O.Oll 3.978

0.504 1.710 1.853 0.888 0.164 0.033 O.Oll O.Oll O.Oll 0.000 0.000 5.196

0.34 0.899 1.02 0.428 0.088 0.022 0.000 O.Oll O.Oll 0.000 0.000 2.808

0.153 0.636 0.724 0.219 0.011 O.Oll 0.000 0.000 0.000 0.000 0.000 1.754

0.899 1.886 2.028 0.713 0.077 O.Oll 0.000 0.011 0.000 0.000 0.000 5.625

0.46 2.182 2.171 0.625 0.099 O.Oll O.Oll O.Oll 0.000 0.000 0.000 5.570

CALM 68.885

68.885

SUM 68.885 3.134 10.131 10.482 4.572 1.458 0.702 0.350 0.209 0.033 0.033 O.Oll 100.000

2.1.4 Streamflow The mean and standard deviation of monthly discharge series calculated for the existing data are presented in table 15. Table 15 Monthly and annual discharge (m3/s) lMesochora station

JanuaI) February

lrvIean value

28.82 34.28 33.32 41.45 32.66 11.91 5.38 3.13

2.92

10.30

25.99

38.95 22.77

Standard deviation

21.68

1.25

9.39

13.53

25.31

March

April

Ma\

June

Jul August September October November Decembe

19.54 15.58 15.03 13.36 4.64 1.37 0.92

Annual

6.21

2.2 DATA PROCESSING Beyond the daily and monthly mIssmg values that are included in all types of data (precipitation, temperature, discharge, etc), enough of them are inconsistent for some years in their recording periods. The missing values of precipitation and temperature are not interpolated for the presented reason. Instead of, consistency checks are applied to these data. 2.2.1 Precipitation The monthly values of raw precipitation for all stations are listed in Tables Al to All in the Appendix A of the Improved Working Report (Panagoulia & Dimou 1997). The consistency of the data was checked through the double-mass curve which has been plotted for all the stations in Figures Al to All (Panagoulia & Dimou 1997). Because of, the Metsovo, Pertouli and


13

Vakari are the most representative stations with high quality data, their average precipitation values were used as data of a base station. For the implementation of consistency checking only, if there were missing monthly values, these were interpolated by the long-term average values of their existing monthly data (panagoulia, 1992b). Actuallly, some deviations from the straight line were observed for some station data. The inconsistent data were corrected by applying a corrective factor calculated from the ratio of the slopes of two segments of the double-mass curve (slope for the recent years/slope for the prior years). The corrected doublemass curves are plotted in Figures 9 to 19 of the Improved Working Report (Panagoulia & Dimou 1997). 2.2.2 Temperature

The monthly values of raw minimum and maximum temperature for all stations are listed in Tables B1 to BS in the Appendix B of the Improved Working Report (Panagoulia & Dimou 1997). The consistency of the data was checked through the mass curve and double-mass curve which have been plotted for all the stations in Figures B 1.1 to BS.2 (Appendix B of the Improved Working Report (1997)). For a consistent record the mass or double-mass curve should be a straight line, or a straight line with waves on it if a seasonal variation between stations exist. The mass curve showed that the Ioannina station was the most consistent station while all the others were not. Thus, the Ioannina station was used as a base station for consistency checking. The consistency of the data was checked by the double-mass curve on a monthly basis (Panagoulia, 1992b), and some deviations from a straight or wave line were observed for minimum data. The inconsistent data were corrected by applying a corrective factor of + 1DC. The corrected double-mass curves are plotted in Figures 20 to 25 of the Improved Working Report (Panagoulia & Dimou 1997). 2.2.3 Streamflow

The daily streamflow data of the Mesochora station for the period 1972-1992 were used. The monthly data of the station are presented in Table C 1 in the Appendix C of the Improved Working Report (Panagoulia & Dimou 1997). Most of the streamflow records were complete, but some missing daily values were included. For estimating the missing data, the Avlaki station was used as a backup station, while the average monthly streamflow for both stations was computed. The Mesochora missing daily data were estimated by multiplying the complete daily data of the Avlaki station by the ratio of the monthly average streamflow (Mesochora! Avlaki). In the case of missing monthly data for the Mesochora station, the drainage basins of


14

the two river stations were used. However, there are gaps of five months in the daily records of Avlaki station. The corresponding missing values of the Mesochora station were interpolated from the daily average values of this station for whole the available study period. 2.2.4 Wind The monthly data (directions and forces) of wind are complete and of good quality for all the period (1956-1993). 2.2.5 Sunshine The monthly relative sunshine is expressed as the ratio of monthly sunshine duration over the maximum sunshine duration corresponding to the spesific latitude (40° N). The three stations were used in combination for data correction and filling (linear correlation). Finally the closest two stations (Ioannina, M. Kerasia) are chosen for the project. The monthly data of the two stations present a good quality after corrections and fillings. 2.2.6 Humidity The two stations were used in combination for correction and flliing (linear correlation) of monthly data of relative humidity. The monthly data of the stations present a good quality after corrections and fillings. 2.3 STATISTICS OF PROCESSED DATA

The basic statistical parameters (mean and standard deviation) of the monthly time series of station data after their corrections are presented in the form of histograms or graphs (Figs 2650 ofImproved Working Report, Panagoulia & Dimou 1997).

3. DETERMINISTIC CONCEPTUAL MODELS: THE US NWSRFS SNOWMELT AND RAINFALL-RUNOFF MODELS The hydrology of a catchment from precipitation to stream discharge at the lowest outfall, can be conceived as a series of interlinked processes and storages. In conceptual simulation, the catchment processes are described mathematically, and the storages are considered as reservoirs, for which water budgets are kept (Shaw, 1984). Many conceptual catchment models have been developed over the last two decades (e.g. Stanford IV, SSARR, USDAHL-74, DISPRIN, et al.), each one structured by different combinations of processes, storages and interchanges, and each requiring some specific sets of input data. Panagoulia & Dimou Hydrological Modeling of the Mountainous Mesochora Cautchment

15

Most of these models are deterministic (the modelled processes do not include stochastic part) and operate on discrete time segments related to a precipitation event (event models), or operate continuously for moisture accounting storage and release regardless of precipitation events (continuous models). Others of these models are classified as lumped parameter models and some others as distributed ones. The lumped parameter models take it that the spatial variability of governing parameters is "lumped" to form an "effective value", which is applied to the entire catchment (Huggins & Burney, 1982), while the distributed ones incorporate spatial variability of parameters directly in the model formulation through partial differential equations, which describe system behaviour with an infinite set of boundary conditions (Huggins & Burney, 1982). Given that there is a relatively large number of conceptual models, the problem of choice of the best one comes up. The required information for such a choice, in first level, includes time step of model operation, model availability and purpose of model use, as well as modelled process and basin size, plus data and temporal resolution of input and output of model. It is evident from the above information that more than one models can simulate successfully the same catchment. For this reason, the selection problem of the optimum model is resolved in a second level, based on the following criteria (Woolhiser & Brakensiek, 1982; James and Burges, 1982; Linsley, 1982, and James, Bowles & Hawkins, 1982): •

accuracy of prediction,

•

ease of use,

•

applicability to the problem,

•

generality of implementation,

•

consistency in parameter estimation,

•

sensitivity of model results to changes in parameters and input variables. However, there is not any better model. Although the conceptual watershed models reqUIre vanous sets of data and much

computation time for their parameters' calibration, said models are nonetheless widely used, not only to simulate the entire behaviour of catchment, but to investigate the interaction of the hydrometeorological variables among them, i.e. the influence of temperature, rainfall, etc on runoff (WMO, 1975). Thus the US National Weather Service models of snow accumulation and ablation, as well as the soil moisture accounting, have been selected for the purposes of


16

this study. Both models are the most representative types of deterministic conceptual models. They also are well documented and have been widely used, the first for snow accumulation and snowmelt, the second for streamflow simulation and forecasting. 3.1 SNOW ACCUMULATION AND ABLATION MODEL (SAA)

The (SAA) model was developed by Eric Anderson (US National Weather Service Hydrologic Research Laboratory, Anderson 1973) and has been tested in numerous mountainous watersheds in the United States and elsewhere. It is a deterministic, continuous conceptual model consisting of a set of mathematical formulations which describe explicitly the accumulation and ablation of the snowpack. The SAA itself can be coupled with almost any soil moisture accounting (rainfall-runoff) model. The output "rain plus melt" from the SAA model can be the input to the rainfall-runoff model. The model inputs are air temperature and precipitation at a six-hour time step. For the SAA analyzed model, daily precipitation was interpolated to six-hour increments and six-hour temperature was estimated from daily temperature maxima and minima using equations given by Anderson (1973). The calibration of SAA was performed concurrently with any rainfall-runoff model since the SAA provides the input to the rainfall-runoff model, which in the examined case is the US NWS soil moisture accounting model. The structure of the SAA model presented in Figure 4 can be summarized as follows. When air temperature [TaJ is less than the delineation temperature (O°C or other), accumulation of snowpack occurs. In the opposite case (Ta> delineation temperature), the SAA assumes that the precipitation is rain. The ablation of snowpack is controlled by the heat exchange at the airsnow interface. For the heat exchange computations there are two basic conditions: (a) when

Ta>O°C, melt takes place at the snow surface, and (b) when Ta O°C, melt occurs at a proportional rate to a seasonally varymg melt factor and the difference between Ta and O°e. During non-melt periods, an antecedent temperature index (AT!) is used as an index to the temperature of the surface layer of snowpack. The heat exchange is assumed proportional to the temperature gradient which varies seasonally as does the melt factor of non-rain periods. The SAA model accounts also for the areal extent of snow cover. During the periods of snow accumulation, this is assumed to be 100%. During periods of depletion, the model uses an areal depletion curve of snow. The SAA involves six major and six minor parameters that are described in the section of model calibration.

Preci pitatti on and

Accumulated

Rain on Bare Ground

Areal Extend of the Snow Cover

Energy Exchange at Snow-Air Interface Snow Cover Heat Deficit

Deficit = 0 Liquid Water Storage Transmission of Excess Water

Figure 4. Snow accumulation and ablation model.


18

3.2 SOIL MOISTURE ACCOUNTING MODEL (SMA)

The soil moisture accounting (SMA) model was developed by Burnash et al. (1973) and forms the basis of the US National Weather Service's basic catchment hydrological response model for operational forecasting. At first the SMA was used for the Sacramento River basin simulation and since then it has been widely used (e.g. WMO, 1975; Nemec & Kite, 1981; Gupta & Sorooshian, 1983; Lettenmaier & Gan, 1988; Panagoulia, 1991 a, 1992a,b). The SMA can be classified using Clarke's (1973) scheme as a deterministic, explicit, continous, lumpedparameter, conceptual model. However, the dynamically expanding saturated surface area

(ADIMP) as one of the SMA model important parameters and associated with the additional direct runoff genesis gives to the model spatially distributed characteristics. The original SMA model was designed for daily precipitation input but later versions allow finer time increments (six-hour or less). The input to the SMA model is the daily pseudoprecipitation (rain plus SAA model output) and the reference evapotranspiration (actual or . long-term average). The latter, in the examined model, was estimated through the Penman method (Veihmeyer, 1964). The SMA model is based on a system of percolation, soil moisture storage, drainage and evapotranspiration characteristics, in order to represent the significant hydrological process in a rational manner. The definition of the model parameters is achieved by establishing a soil-moisture computation which allows the determination of watershed streamflow from watershed precipitation. Effective moisture storage capacities in the soil profile are estimated not by sampling of the soil profile, but by inference from the pseudoprecipitation and streamflow records. The final values of the model parameters are determined through the calibration procedure. Figure 5 illustrates the components of the SMA model. The model is represented by an upper and lower zone. Each zone stores moisture in two forms, tension moisture and free moisture. Tension moisture denotes water closely bound to the soil particles while free moisture is the moisture that fills up the soil pores. The upper zone includes one free moisture storage while the lower zone includes two free moisture storages (primary and supplemental). Free water from the upper zone can descend to the lower storages by percolation or can move laterally to produce interflow. Percolation is controlled by the contents of the upper zone free water and the deficiency of lower zone moisture volume. When the precipitation rate exceeds the percolation rate and the maximum interflow drainage capacity, then the upper zone free water capacity is filled completely and the excess rainfall will result in surface runoff.


19

. .LOW

TENSION

__TER

SIOUT

LZTW

Figure 5 Soil moisture accounting model components.

The two lower free water storages fill simultaneously from percolated water, and drain independently at a different rate, giving a variable groundwater recession. Moisture is also extracted from the upper and lower tension zones and from free water surfaces by evapotranspiration. Direct runoff from impervious areas, surface runoff, interflow and primary and supplemental baseflow contribute to generate the channel flow. The SMA model employs about 21 parameters which are discussed in the section of model calibration. 3.3 INPUT DATA Three categories of stations and data, according to their use by the models, are described. These are: 1.

precipitation and temperature stations with daily data,

11.

temperature, sunshine, and humidity stations with monthly data, and

iii. streamflow station with daily data.


20

Table 16. Precipitation gauge stations by elevation zone, and temperature stations for the entire Mesochora catchment.

STATION

ZONE

YEARS OF RECORD ZONE ZONE ZONE STATION ELEVATION MEDIAN ELEVATION RANGE ELEVATION AREA PRECIP TEMPERATURE [m] [m] [m] [%] max min

METSOVO KATAFYTO P ALAIOCHORI P ALAIOCHORI MIDDLE MATSOUKI PERTOULI ELATI(TYRNA) VAKARI MESOCHORA LOWER PACHTOURI STOURNAREIKA VATHYREMA UPPER

1157 980 1050 1050 1079 1160 900 1150 780 950 860 920

1580-2200

1830

30.54

1280-1580

1400

29.51

780-1280

1080

39.95

21 21 21 21 21 21 21 21 21 21 21 21

18

21

21

21

11

11

In addition, the methods used to average the station measurements data over area and elevation are described. 3.3.1 Precipitation and temperature stations-data

Eleven precipitation stations are installed within and around the Mesochora catchment, with the greater density at the lower part of the catchment (Fig. 2). The general characteristics of the stations for the study period (1972-1992) are presented in Table 16. The precipitation stations are consistent and representative of the catchment, but in some precipitation records daily data are missing. The consistency of the existing data was checked by the double-mass curve on a monthly basis (Anderson, 1973). The inconsistent data were corrected by applying an appropriate corrective factor of the order 0.982 to 1.037 (Panagoulia & Dimou 1997). As regards the missing data ,we did not interpolate them for two reasons:

(a) in order to preserve the real nature of precipitation-temperature series elements, (b) in order to avoid the computing errors that are introduced by the estimation techniques of daily missing data. Thus, the technique used to estimate the mean areal daily precipitation was a combinatorial one of a similar Thiessen method and station daily availability, including elevation correction (Panagoulia, 1992, 1995). The correction areal precipitation factor for a given elevation (e.g. midpoint catchment elevation) is obtained from the following algorithm.


21

(1) where, pe is the multiplying corrective precipitation factor for elevation e, P is mean areal precipitation, ews is the weighted mean station elevation based on station daily availability, and pfis the variation rate of precipitation with elevation.

Concerning the network of temperature stations there are three stations which used in the study. One is installed inside the catchment, while the other two are installed outside and present significant deficiencies of daily maximum and minimum data. The consistency of the existing data was checked by the double-mass curve on monthly basis (Anderson, 1973) using the Ioannina station as basis (another neighbouring station to the catchment with reliable and complete data, see section 2.1.1). Some deviations from straight line were observed for minimum data. The inconsistent data were corrected by applying an appropriate corrective factor of the order -0.6 to -l.O°C. For the implementation of consistency checking only, if there were missing monthly minimum temperatures, these were interpolated by the long-term average values of their existing monthly data. The technique used to estimate the mean areal maximum and minimum daily temperature was, also, a combinatorial technique of the Thiessen method and station daily availability. The correction areal temperature factor for a given elevation is obtained from the following algorithm: (2) where,

Te

is the additive corrective temperature factor for elevation e, ews is the weighted mean

station elevation based on station daily availability, and

If is the rate of temperature decrease

with elevation (lapse rate). 3.3.2 Reference evapotranspiration

In order to obtain the catchment reference evapotranspiration, the sunshine, temperature and humidity data on a monthly basis were considered. Their measurement stations have been described in Tables 3 to 6 in section l.2 (see also Improved Working Report, 1997). The sunshine stations are outside the catchment, the Ioannina is at the north-west boundary of the catchment, while the M.Kerasia station is at the eastern boundary. The catchment sunshine is computed from the long-term monthly arithmetic average of the sunshine of the two stations. By the above manner it was calculated, also, the catchment humidity from the two stations. The Pramanta station is installed outside the catchment near the western boundary. Panagoulia & Dimou Hydrological Modeling of the Mountainous Mesochora Cautchment

22

The catchment temperature was estimated by the above Thiessen method combining it with the monthly station availability. The areal temperature was corrected for catchment mean elevation by applying a monthly dependent lapse rate according to equation (2). The long-term monthly mean catchment sunshine, temperature, and humidity were used as inputs to Penman equation which is given in Veihmeyer (Chow, 1964) to estimate the catchment reference evapotranspiration. Other inputs to Penman equation were the average wind speed, (200 miles /day), the monthly percent of reflecting surface and the solar radiation for the midpoint catchment latitude. The sunshine was entered to Penman equation as monthly ratio of duration of bright sunshine to maximum possible duration of bright sunshine. 3.3.3 Streamflow The daily streamflow data of the Mesochora gauge station for the period 1972-1992 were used in this study. The streamflow records of the Mesochora station have been described in sections l.2.5, 2.1.4 and 2.l.3.

3.4 MODEL CALIBRATION The procedures used for model calibration (parameter estimation) and SMA model validation and verification are developed as follows: 3.4.1 SAA model parameter estimation The mountainous catchment was divided into elevation zones, and the SAA model was applied to each zone separately, since low elevations are likely to receive rain, while higher elevations receive snow from the same storm. The weighted mean value of the pseudo-precipitation from all zones was treated as the mean areal precipitation which is the input to the SMA model. In general, the weighting factors used were equal to the ratios of the elevation zone subareas for the total catchment area. The elevation bands were delineated as follows: First hypsometric curve (elevation versus area fraction) was developed (see section 1.1). The catchment was then divided into three zones of area, depending on the elevation range, and the elevation of the midpoint of each band was identified (Tables 1 and 16). The SAA model was manually calibrated for the three elevation zones. According to available station data on that particular day, for every elevation zone, the precipitationelevation correction factors pe, were estimated through a trial and error approach, which was carried out concurrently with the calibration of Soil Moisture-Accounting model. The final precipitation-elevation correction factors are given in Table 17. Panagoulia & Dimou Hydrological Modeling of the Mountainous Mesochora Cautchment

23

Table 17. Station weights and precipitation correction factors. Lower zone

Availability station 000001 000010 000011 000100 000101 000110 000111 001000 001001 001010 001011 001100 001101 001110 001111 010000 010001 010010 010011 010100 010101 010110 010111 011000 011001 011010 011011 011100 011101 011110 011111 100000 100001 100010 100011 100100 100101 100110 100111 101000 101001 101010 101011 101100 101101 101110 101111 110000 110001 110010 110011 110100 110101 110110 110111 111000 111001 111010 111011 111100 111101 111110 111111

Tyrna .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 1.000000 .139466 .143422 .115727 .437191 .139466 .142433 .115727 .255193 .139466 .124629 .115727 .255193 .139466 .124629 .115727 .173096 .127596 .103858 .103858 .173096 .127596 .103858 .103858 .173096 .127596 .103858 .103858 .173096 .127596 .103858 .103858

Vakari .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 1.000000 .641939 .713155 .629080 .765579 .407517 .478734 .394659 .654797 .298714 .367953 .285856 .654797 .298714 .367953 .285856 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .826904 .617211 .701286 .617211 .592483 .382789 .466864 .382789 .481701 .273986 .356083 .273986 .481701 .273986 .356083 .273986

Mesochora .000000 .000000 .000000 .000000 .000000 .000000 .000000 1. 000000 .539070 .639960 .539070 .800198 .339268 .440158 .339268 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .345203 .343225 .345203 .343225 .145401 .143422 .145401 .143422 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .744807 .539070 .634026 .539070 .545005 .339268 .434224 .339268 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .000000 .345203 .343225 .345203 .343225 .145401 .143422 .145401 .143422

Pachtouri .000000 .000000 .000000 1.000000 .372898 .492582 .372898 .000000 .000000 .000000 .000000 .199802 .199802 .199802 .199802 .000000 .000000 .000000 .000000 .234421 .234421 .234421 .234421 .000000 .000000 .000000 .000000 .199802 .199802 .199802 .199802 .000000 .000000 .000000 .000000 .562809 .372898 .492582 .372898 .000000 .000000 .000000 .000000 .199802 .199802 .199802 .199802 .000000 .000000 .000000 .000000 .234421 .234421 .234421 .234421 .000000 .000000 .000000 .000000 .199802 .199802 .199802 .199802

prO.02

Stournareika Vathyrema .000000 1.000000 .161227 .000000 .000000 .507418 .161227 .000000 .000000 .360040 .161227 .000000 .000000 .360040 .161227 .000000 .000000 .286845 .152324 .000000 .000000 .286845 .152324 .000000 .000000 .286845 .152324 .000000 .000000 .286845 .152324 .000000 .000000 .856578 .060336 .000000 .000000 .364985 .060336 .000000 .000000 .241345 .060336 .000000 .000000 .241345 .060336 .000000 .000000 .194857 .060336 .000000 .000000 .194857 .060336 .000000 .000000 .194857 .060336 .000000 .000000 .194857 .060336

1.000000 .000000 .838773 .000000 .627102 .000000 .465875 .000000 .460930 .000000 .299703 .000000 .460930 .000000 .299703 .000000 .358061 .000000 .218595 .000000 .358061 .000000 .218595 .000000 .358061 .000000 .218595 .000000 .358061 .000000 .218595 .000000 .860534 .000000 .823937 .000000 .487636 .000000 .451039 .000000 .321464 .000000 .284866 .000000 .321464 .000000 .284866 .000000 .255193 .000000 .218595 .000000 .255193 .000000 .218595 .000000 .255193 .000000 .218595 .000000 .255193 .000000 .218595


Pe 1. 001730 1.002379 1. 001834 1.001405 1.001609 1. 001899 1. 001713 1. 003243 1. 002546 1. 002932 1. 002650 1. 002876 1. 002179 1.002565 1.002283 .999243 1.000134 1.000143 1. 000264 .999750 1. 000640 1.000649 1.000771 1. 000624 1. 001507 1. 001523 1. 001637 1. 000257 1. 001139 1. 001156 1. 001270 1. 001946 1.001760 1. 002316 1. 001794 1. 001642 1. 001639 1. 001838 1.001673 1. 002912 1. 002576 1. 002873 1. 002610 1. 002545 1. 002209 1.002506 1.002243 .999711 1.000223 1. 000135 1.000257 1. 000218 1.000730 1.000642 1.000764 1. 001092 1. 001596 1.001516 1. 001630 1.000725 1. 001228 1.001149 1. 001262

24

Table 17. (Continued) pj=1.20

Middle zone Availability Palaiochori station 001 010 011 100 101 110 111

.000000 .000000 .000000 1.000000 .832664 .597055 .429719

Matsouki .000000 1.000000 .520750 .000000 .000000 .402945 .402945

Pertouli 1.000000 .000000 .479250 .000000 .167336 .000000 .167336

pe 1.189706 1. 253732 1. 223047 1. 276654 1. 262105 1.267418 1. 252868

Table 17. (Continued) pj=O.63

Upper zone

Availability station

Metsovo

Katafyto

001 010 011 100 101 110 111

.000000 .000000 .000000 1.000000 .586028 .107374 .107374

.000000 1.000000 .848642 .000000 .000000 .892626 .741268

Palaiochori 1.000000 .000000 .151358 .000000 .413972 .000000 .151358

pe 1.352739 1. 384395 1. 379604 1. 304351 1. 324382 1.375801 1. 371009

As with the function pe, the lapse rate is usually nonlinear. It was found that average lapse rates for four periods (each 6 hours long) of each day is -0.80DC/l00m. The lapse rates,

Tf and the temperature correction factors, Te , were also estimated concurrently with the calibration of SMA model, and their final values are given in Table 18.

Table 18. Station weights and temperature correction factors Te for lapse rate Tr-0.80. Availability station 001 010 011 100 101 110 111

Station

Metsovo .000000 .000000 .000000 1.000000 .304228 .263927 .258396

Vakari .000000 1.000000 .734492 .000000 .000000 .736073 .582379

Zone

Pachtouri 1.000000 .000000 .265508 .000000 .695772 .000000 .159226

Lower -1.040000 .560000 .135188 .608000 -.538633 .572668 .317642

Middle

Upper

-3.600000 -2.000000 -2.424812 -1.952000 -3.098633 -1. 987331 -2.242358

-7.040000 -5.440000 -5.864812 -5.392000 -6.538633 -5.427331 -5.682358

Seasonal melt factors were interpolated between MFMAX and MFMfN and were estimated to 0.9 and 0.4 rnrn / DC per 6 h, respectively. PXTEMP, the temperature eC) above which precipitation was assumed to be rain was assumed to be 1DC for the first six months of calendar year, and ODe for the rest ones. Sf, the mean areal water-equivalent above which


25

Table 19. Snow melt model parameter description and parameter calibrated values.

Description

Parameter SCF MFMAX MFM1N UADJ

SI NMF T1MP PXTEMP MEASE PLWHC PAYGM EFC

Calibrated values

A multiplying factor to correct for gauge catch deficiency in the case of snowfall Maximum melt factor during non-rain periods which occurs on June 21 Minimum melt factor during non-rain periods which occurs on December 21 Average wind function during rain on snow periods Mean areal water-equivalent above which there is always 100% areal snow cover Maximum negative melt factor Antecedent temperature index parameter Temperature which delineates rain from snow Base temperature for snow melt computation during nonrain periods Percent liquid-water holding capacity of ripe snow Average daily ground at the snow-soil interface Percent area over which evapotranspiration occurs when ~here is 100% snow cover

1.10 0.90 mm/°C/6hr 0.40 mm/°C/6hr 0.10 mmlmb/6hr 100mm 0.12 mme/oC/6hr 0.30 1.0 - O°C O°C 0.05 0.020mm 0.61

100% areal snow cover always exist, was assumed to be 100mm. The depletion snow curve was formed from the pairs of the following values: AREAL J:lAEAN WATER-EQUIVALENT/Ai

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

l.0

SNOW COVER EXTENT

.05

.11

.18

.20

.22

.25

.33

.42

53

.71

l.00

Description and calibrated values of the SAA model parameters are shown in Table 19. 3.4.2 SMA model parameter estimation

Parameter estimation for the SMA model was based on a process of initial parameter estimation, as suggested by Peck (1976). The final values of parameters were obtained by the manual model calibration on the Mesochora catchment. The description of model parameters and their final values are listed in Table 20. The model was calibrated for the period from 1972 to 1986, which included dry, medium and wet years, so that the model could be subjected to a broad range of changes in conceptual storages. Following calibration, verification was performed using the periods from 1972 to 1989 and from 1972 to 1992. The results of error analysis of daily flow, as well as the three day volume error analysis including peaks, are presented in Tables 21 and 22, respectively. Furthermore, the standard deviation and correlation coefficient of monthly runoff are described in Table 23. The values


26

taken from the statistic parameters of Tables 21 and 23, as well as the comparison between simulated and observed flow components of Table 22, suggest that both hydrological models were proved capable of reproducing the observed streamflow (the SAA model indirectly, and the SMA model directly). Also, the typical monthly simulation errors expressed in per cents of observed flows, were of the order of 10-15 % higher in low runoff months (August and September) and lower in high runoff months. A more detailed comparison of the SMA model with actual runoff conditions is presented below in "Model validation".

Table 20. Soil moisture accounting model parameter description and parameter final values for three periods (1972-86, 1972-89, 1972-92). Soil moisture Parameters phase IpeTIM Direct runoof f4. DJJvfP fsARVA Upper zone

UZTWM UZFWM UZK '(zPERC

1972-86

1972-89

1972-92

.010 .010 .000

.010 .010 .000

.010 .010 .000

4.50cm 4.61 cm 0.57 6.0

4.50 cm 4.61 cm 0.57 6.0

4.50 cm 4.61 cm 0.57 6.0

Percolation

Initial water

~

l.800

1.800

1.800

jLZTWM jLZFSM

25.00 cm 9.00 cm

25.00 cm 9.00 cm

25.00 cm 9.00 cm

LZFPM LZSK

30.00 cm 0.15

30.00 cm 0.15

30.00 cm 0.15

'1-ZPK

0.01

0.01

0.01

IPFREE

.200

.200

.200

If?SERV

.100

.100

.100

fsIDE

.000

.000

.000

UZTWC UZFWC '1-ZTWC '1-ZFSC

4.50 cm .ll cm 21.41cm .09cm

4.50 cm .05 cm 25.00 cm .31 cm

4.50 cm .04cm 25.00 cm .18 cm

jLZFPC f4.DIMC

2.26cm 25.91 cm

7.37 cm 29.01 cm

6.28 cm 28.83 cm

Description Minimum impervious catchment Additional impervious catchment Catchment covered by streams, lakes and riparian vegetation Upper zone tension water capacity Upper zone free water capacity Daily upper zone free water drainage rat~ Proportional increase in percolation from saturated to dry condition Exponent affecting rate of change of percolation between wet and dry conditions Lower zone tension water capacity Lower zone supplementary free water capacity Lower zone primary free water capacity Daily lower zone supplementary free water drainage rate Daily lower zone primary free water drainage rate Percolation water fraction passing directly to lower zones free water Fraction oflower zone free water unavailable for transpiration Ratio of non-channel baseflow to channel baseflow Upper zone tension water content Upper zone free water content Lower zone tension water content Lower zone supplemental free water content Lower zone primary free water content Tension water content of the additional impervious catchment


27

Table 21. Soil moisture accounting model Error analysis for daily runoff for three periods (1972-86, 1972-89, 1972-92). RUNOFF [m3/s/km2]

TOTAL

SURFACE

UPPER LEVEL

LOWER LEVEL

72-86 72-89 72-89 72-86 72-89 72-92 72-86 72-89 72-92 72-86 72-89 72-92 STANDARD ERROR

0.528 0.595 0.632 0.756 0.901 0.972 0.655 0.753 0.816 0.475 0.523 0.552

AVERAGE BIAS

0.017 -0.014 -0.029 0.525 0.583 0.633 0.134 0.184 0.249 -0.056 -0.108 -0.138

DAYS

5479

6575

7671

680

608

711

811

4380

5272

6180

:MEAN DISCHARGE

3.70

3.67

3.51 11.12 10.72 10.34

8.00

7.81

7.52

2.37

2.27

2.14

491

592

Table 22. Soil moisture accounting model error analysis three-day volumes including peaks for three periods (1972-86, 1972-89, 1972-92). observed [m3/s/km2] forecast [m3/s/km2] 72-86 72-89 72-92 72-86 72-86 72-92

72-86 72-89 72-92 13.24 15.37 15.02 STANDARD ERROR 444

515

610 PEAKS-HYDROGRAPHS MEAN THREE-DAY VOLUME

20.54

20.83

19.73

21.30

20.24

18.63

2.56

2.51

2.46

3.74

4.23

4.03

124

140

183

LESS EQUAL TO 5.

161

187

214

OTHERS 20. OR LESS

11.96

11.99

11.93

16.11

15.84

14.93

159

188

213

GREATER THAN 20.

43.26

40.25

36.55

34.89

8.90

43.26 9.05

42.40

:MEAN PEAK-DAY FLOW

8.53

8.29

7.94

7.33

DAY OF FIVE-DAY DISCHARGE CENTROID

3.00

3.00

3.00

3.07

3.06

3.04

Table 23. Soil moisture accounting model monthly runoff for three periods (1972-86, 1972-89, 1972-92). 1972-86

1972-89

1972-92

STANDARD ERROR

2.584 cm

2.847 cm

2.883 cm

CORRELATION COEFFICIENT

0.954

0.940

0.936

3.4.3 SMA model validation While there are many criteria for describing how well a particular model performs the task for which it was employed (i.e. criteria cited in the above section, root mean square error, coefficient of efficiency, etc), we selected a graphical display (time series plot) to show the good fitting of observed and simulated streamflow in a summarised scheme (Figures 6,7 and 8). However some particular differences between simulated and observed daily runoff series there are (panaroulia 1992b).


28

1.4

1972-86

1.2

1.0

5.5:!.. ~

0.8

0

..::: .",

......'"

0.6

'"'"

,Q

0

0.4

0.2

0.0 1.0

0.8

=

.5:!.. ~

0.6

0

..::: .",

....'"

-==

0.4

ii'i

0.2

0.0 1.6

1.4

=

1.2

.5:!..

;;:: "'"

'" 1.0 = '"

'" '" '"

Q.

.; 0.8 -!::.0

!'s.

0.6

'u

...'"

~

0.4

0.2

0.0 111

112

113

1/4

115

116

IJ7

118

119

1110

1111

1112

Day

31112

Figure 6 Mesochora cautchment: average daily precipitation (rain plus melt) observed flows and soil moisture accounding model predicted flows for the IS-year calibration period.

Panagonlia & Dimon Hydrological Modeling of the Mountainous Mesochora Cautchment

29

1972-89

1.4

1.2

1.0

e ..::.. 1:=

0.8

0

0;::: "0

......'" '"'"

0.6

,.Q

0

0.4

0.2

0.0 1.0

0.8

e

..::.. 1:=

0.6

0

0;::: "0

.. '"

:;

0.4

e

00

0.2

0.0 1.6

1.4

e

1.2

..::.. C' a:i 1.0

e

'" ::

Q.

= = 0.6 E "Q.

"; 0.8 ~

~

0

!~

"c:l

...'" p...

0.4

J

~

V V

IV

~IWV'~

I~ 0.2

0.0 111

~

112

113

114

1/5

116

In

118

119

1110

1111

1/12

31112

Day

Figure 7 Mesochora cautchment: average daily precipitation (rain plus melt) observed flows and soil moisture accounding model predicted flows for the I8-year calibration period.


30

1972-92

1.4

1.2

1.0

'5 ~ ~

0.8

0

!;: "0

"... "'" ~

0.6

.Q

0 0.4

0.2

0.0 1.0

0.8

5

~ ~

0.6

0

!;: "0

~

..:: 0.4 = s 00

0.2

0.0 1.6

1.4

5

1.2

~

Z"

0;

S

1.0

'"

=

Q.

= 0.8 .; .:::"

= 0

0.6 B 'Q. '