evaluation of stream temperature change with reservoir ... - CiteSeerX

EVALUATION OF STREAM TEMPERATURE CHANGE WITH RESERVOIR MANAGEMENT IN THE NAM NGUM BASIN Koshi Yoshida*, Hajime Tantji**, Hiroaki Somura**, Osamu Toda***, and Takao Masumoto** * Japan Science and Technology Agency ** National Institute for Rural Engineering *** Graduate School of Agricultural and Life Sciences, The University of Tokyo ABSTRACT The Mekong is a large international river and many reservoir construction projects are planned in the tributaries. Such human activity raises or lowers natural stream-water temperature and, unfortunately, change of water temperature is often undesirable. Control of water temperature therefore becomes a prerequisite for maximum utilization of water resources. Reservoirs often provide a means for exercising such controls. In this study, to evaluate artificial effects on the downstream environment, changes in river water temperature by dam construction were estimated. Then, the mitigation effect of adopting surface intake on the dam site was examined. Key Words: Mekong, Runoff analysis, Stream Temperature, Dam Management 1. INTRODUCTION The Mekong is a large international river. It flows through 6 countries (China, Myanmar, Thailand, Laos, Cambodia and Vietnam) often presenting conflicts of water resource utilization between upstream and downstream areas. After the establishment of the MRC (Mekong River Committee) dam construction in the main stream was prohibited, but many dam construction projects are planned in the tributaries, especially in Laos. Laos is blessed with good topographical conditions for hydroelectric power sites and recently switched its economic policy from forest resources exportation to electric power exportation. Usually, the generation of electricity by dams is regarded as non-consumptive water use (as opposed to irrigation by dams, which is regarded as consumptive use), while at the same time fish ways are constructed to compensate for cutting off upstream and downstream habitats. In view of water volume, electricity is regarded as non-consumptive use, and thus fish ways are regarded as successfully satisfying the ecological need for a connection between the upstream and downstream habitats. However, what about heat environmental change in the downstream? Temperature is perhaps an important parameter in stream-water quality. Human activity raises or lowers natural stream-water temperature due to impoundments, industrial use, irrigation and global warming (Bartholow 1989, LeBlanc 1997). Water temperature affects nearly every physical property related to water quality management and, unfortunately, change of water temperature is often undesirable. Previously, stream temperature increased by solar energy (solar radiation, convection, conduction) under non-dam conditions as flowing to downstream, and especially decreased by the release of cool water from dams (Delay 1966, Bashar 1993). Control of water temperature, therefore, becomes a prerequisite for maximum utilization of water resources. Reservoirs often provide a means for exercising such controls. In this study, to evaluate artificial effects on the downstream environment, changes in river water temperature by dam construction were estimated. Then, the mitigation effect of adopting surface intake on the dam site was examined. 2. STUDY AREA The Nam Ngum River Basin was selected as the study area. Nam Ngum River is a large tributary of the Mekong River, having a length of 420 km and a drainage basin of 16,400 km2. It flows through Vientiane (the capital of Laos) and has a large dam (Nam Ngum), shown in Figure 1. The Nam Ngum Dam is located just upstream of the confluence of the Nam Ngum and Nam Lik rivers and has an 8,280 km2 catchment area.

The Nam Ngum Dam has a height of 75 m, a capacity of 9_109 m3 and a depth of 30 m (Hori 2000). In this study, discharge data from two gauge stations on the Nam Ngum River (Naluang, Pakkagnoung) and one gauge station (Hinheup) on the Nam Lik River were used. Stream temperature was evaluated for cases with and without a dam, and with surface intake.

3. DATA SET In this study, topographic, hydrological, and meteorological data sets were used (shown in Table 1). Stream flow line, watershed area, hill slope, and stream length were calculated (shown in Table 2) from the Global Map DEM data. A basic model and schematic of the study area was prepared based on the calculated subbasin (shown in Figure 2).

4. RUNOFF MODEL In this study we employed TOPMODEL to calculate runoff discharge from each separated grid (1km resolution). TOPMODEL, proposed by Beven and Kirkdy (1979), is based on the contributing area concept in hill slope hydrology (shown in Figure 3). This model is based on the exponential transmissivity assumption that leads to a topographic index ln(a/To/tanb). To is lateral transmissivity under saturated conditions, and a is upstream catchment area draining across a unit length, and b is local gradient of the ground surface. This model combines lumped and distributed characteristics using a topographic index, so many improved models have been applied to the Mekong River basin (Ao 1999, Nawarathna 2001). Figure 4 shows distribution of topographic index in this area. High topographic index value means that the area is easy to saturate.

The average saturation deficit D(t) is distributed to the local saturation deficit S(i,t) at grid cell i, according to the magnitude of local topographic index relative to its average _ as follows: S ( i ,t ) = D( t ) + m ↔(ƒÉ− ln( a / To / tanb )) (2) where m is the decay factor of lateral transmissivity with respect to saturation deficit in meters. Rainfall on grid cell i is first input to the root zone. Storage in the root zone Srz(i,t) changes as follows: Srz( i , t ) = Srz( i , t − 1 ) + R( i , t ) − E( i , t ) (3) where R is precipitation and E is evapotranspiration. The excess of root zone storage Srz(i,t) is input to the unsaturated zone and its storage Suz(i,t) is calculated using Eq.(4). Suz( i ,t ) = Suz( i ,t − 1 ) + Srz( i ,t ) − Sr max( i ,t ) (4) where Srmax is the maximum height of the root zone tank. Overland flow from grid cell i, qof(i,t) can be estimated as follows: qof ( i , t ) = Suz( i , t ) − S ( i , t ) _______ (5) Groundwater discharge is considered semi-steady depending on the saturation deficit. The hydraulic gradient

is assumed parallel to the ground surface. Groundwater discharge from grid cell i is determined from Eq.(6): qb( i ,t ) = To ?exp( − S ( i ,t ) / m ) ?tanb (6) The discharge from grid cell i to the stream is the summation of qof(i,t) and qb(i,t). 5. ROUTING MODEL The above-mentioned TOPMODEL was applied to the Nam Ngum River basin and the Muskingum-Cunge Method was employed for flow routing along the main stream. Consider the continuity equation: ƒA ƒQ + =0 ƒt ƒx

(7) where Q is the flow discharge, A is the cross-section. Introducing flood wave celerity c, this equation can be expressed in terms of Q only as follows: dA ƒQ ƒQ 1 ƒQ ƒQ √ =0 + = √+ ƒx ƒ dQ √ t ↵ ƒx c ƒt ↵

(8) By centering the time derivative (putting _=0.5 ) and keeping _ a variable, this can be expressed in finite differences as follows (referring to Figure 5 for the positions of the points): 1 [θ ( Q2 − Q1 ) + ( 1 − θ )( Q4 − Q3 )]/ ∆t + [Q3 + Q4 − Q1 − Q2 ]/ 2∆x = 0 c

(9)

By solving this equation for Q4 the following equation is obtained, Q 4 = C1Q1 + C 2 Q 2 + C 3 Q 3

where C1 =

:∆t + 2Kθ

∆t + 2 K ( 1 − θ )

K = ∆x / c

:

:

c=

C2 =

(10)

∆ t − 2 Kθ ∆t + 2 K ( 1 − θ )

C3 =

2 K ( 1 − θ ) − ∆t ∆t + 2 K ( 1 − θ )

dQ 5 S 3 / 2 = h dA 3 n

Used parameters for runoff analysis and flow routing were shown in Table 3. Figure 6 and Figure 7 show results of model calibration at the Naluang station and the Hinheup station respectively. Figure 8 shows calculated hydrograph in 1997 at the Pakkagnoung station, in each case of with and without a dam. In the with dam case, base flow in dry season increases and flood peak in rainy season is cut. In assessment based on water volume, this indicates improvement of water utilization by dam construction.

6. STREAM TEMPERATURE ANALYSIS In this study, SSTEMP (Stream Segment Temperature Model, Bartholow 1989) was employed for calculating stream temperature changes. SSTEMP solves the one-dimensional heat advection-dispersion equation and includes the heat exchange with the atmosphere. When applied to an open channel of constant cross section and flow, the heat transport equation takes the form A

WS ƒT ƒ( QT ) ƒ  ƒT ADL + = √+ ƒt ƒx ƒx ƒx ↵ ρC p

(11)

where T is water temperature, x is distance, t is time, D L is a dispersion coefficient in x direction, S is a source or sink term which includes heat transfer with the surrounding environment, A is cross sectional area, W is surface width, Q is flow rate. For numerical test, dispersion D L of 15-200 m2/s were applied, but increases of temperature did not change model simulation results because of the relatively high velocity of flow and the low longitudinal temperature gradient along the modeled stream, indicating that the system was advection dominant. Therefore DL was set equal to zero in the model. Equation (7) can be used to predict water temperatures in steady flow streams that are well mixed and have no significant transverse temperature gradients, the main variation is in the flow direction (Bashar 1993).

Stream cross sectional area A and width W are a function of stream flow rate Q, _ is the density of water, and Cp is the specific heat of water. The source or sink term S expresses the heat transfer rate with the surrounding environment: S = S a + Sb (12) with Sa = H s − H l − H e − H c (13) where Sa is the net heat exchange rate between the water and the air, Sb is the net heat exchange between the water and the streambed, H s is the net short-wave (solar) radiation, H l is the net long-wave radiation, H e is the evaporative heat transfer, Hc is the convective heat transfer In general terms, SSTEMP calculates the heat gained or lost from a parcel of water as it passes through a stream segment. This is accomplished by simulating the various heat flux processes that determine that temperature change. These physical processes include convection, conduction, evaporation, as well as long wave radiation (heat to or from the air), direct solar radiation (short wave), and radiation back from the water. First, solar radiation and interception by shading are calculated. To calculate solar radiation, SSTEMP computes the radiation at the outer edge of the earth’s atmosphere. This radiation is passed through the attenuating effects of the atmosphere and finally reflects off the water surface depending on the angle of the sun. Next, sunrise and sunset times are computed by factoring in local east and west side topography. Thus the local topography results in a percentage decrease in the level plain daylight hours. From this local sunrise/sunset, SSTEMP computes the percentage of light that is filtered out by the vegetation. SSTEMP requires inputs describing the average stream geometry, as well as (steady-state) hydrology and meteorology, and stream shading. SSTEMP optionally estimates the combined topographic and vegetative shade as well as solar radiation penetrating the water. Vegetation data was estimated using Global Map Vegetation data along the stream (shown in Figure 9). The model then predicts the mean daily water temperatures at specified distances downstream. It also estimates the daily maximum and minimum temperatures.

To apply the model, the Nam Ngum River Basin was schematized (Figure 2) and three analyses were conducted: with a dam, without a dam, and with surface intake. The Nam Ngum Dam releases about 300 m3/s discharge for electricity and the release water temperature is about 23°C (Hori 2000). In the surface intake case, we estimated that 20% of 300 m3/s was taken from the surface of the reservoir. Figure 10 shows the result of temporal mean stream temperature change at Pakkagnoung in 1997. And Figure 11 shows spatial distribution of monthly averaged stream temperature affected area in (a) April (dry season) and (b) September (rainy season). In Figure 10, the added vertical axis on the right shows the discharge at Pakkagnoung in each case. Generally, when the river keeps natural stream condition (in the case without the dam), the stream temperature is strongly affected by the mean air temperature. In the dry season (Jan-June), the total discharge is small so that the percentage of dam release discharge, which usually has a low temperature, increases. Thus, stream temperature in the case with the dam is about 5°C cooler than in the one without the dam.

In the rainy season (July-Oct.), the total discharge was so large that the stream temperature was not as sensitive to the dam-released cool discharge. Thus, stream temperature did not differ greatly between the two cases during the flood period (beginning of September). Even in the rainy season, however, stream temperature decreases under conditions with the dam when the percentage of dam-released cool discharge increased. In the surface intake case, the stream temperature is about 2°C warmer than in the case with the dam. Ecological knowledge leads to appropriate stream temperature and the required conditions for the river can be managed by the adoption of surface intake. 7. CONCLUSION In this study we examined stream temperature change by dam release water management. For detailed assessment, not only water volume but also water temperature should be considered. Stream temperature is an especially important parameter for evaluating any artificially induced impact on the environment and ecosystem. The results show that dam construction provides cooler stream temperature and can be controlled by appropriate surface intake from the reservoir. Dam management strongly affects the downstream

environment; so more detailed monitoring is essential for estimating the impact of artificial activity. Since June 2003, this is now being measured in the field. In this study, an open source, steady-state model (SSTEMP), was applied directly to discharge data to calculate stream temperature change. For further improvement, therefore, surface flow and temperature should be solved dynamically at the same time during the physical process.

ACKNOWLEDGEMENTS This research was partially supported by two Grant-in-Aids: Revolutionary Researches 2002 of the Ministry of Education, Science, Sports and Culture, and CREST of the Japan Science and Technology Corporation. REFERENCES Ao, T., Ishihira, H. and Takeuchi, K. (1999) Study of Distributed Runoff Simulation Model based on Block Type Topmodel and Muskingum-Cunge Method, Journal of Hydrological Engineering, 43, 7-12. Bartholow, J. M. (1989) Stream temperature investigation: Field and analytical methods. Instream Flow Information Paper No.13, US Fish Wildlife Service Biological Report 89 (17). Bashar, A. S. and Stefan, H. G. (1993) Stream temperature dynamics (Measurements and modeling), Water. Res. Res., 29 (7), 2299-2312. Beven, K. J. and Kirkby, M. J. (1979) A physically based, variable contributing area model of hydrology, Hydrological Science Bulletin, 24 (1), 43-69. Delay, W., and Seaders, J. (1966) Predicting temperature in rivers and reservoirs, Journal of Sanitation Engineering Division, ASCE, 92, 115-134. Hori, H. (2000) The Mekong – Environment and Development-, The United Nations University. LeBlanc, R. T., Brown, R. D. and FitzGibbon, J. E. (1979) Modeling the effects of land use change on the water temperature in unregulated urban stream, Journal of Environmental Management, 49 (4), 445-469. Nawarathna, N. B., Ao, T. Q., Kazama, S., Sawamoto, M. and Takeuchi, K. (2001) Influence of human activity on the BTOPMC Model Runoff Simulations in large-scale watersheds, XXIX IAHR Congress Proceedings, Theme a, 93-99.