0068-B1

Effect of Land Use/Land Cover Changes and Practices on Sediment Contribution to the Tri an Reservoir of Dong Nai Watershed, Vietnam

Nguyen Kim Loi^[1]

Abstract

This study is aimed at assessing factors contributing to reservoir sedimentation, as well as deriving the predictive models of reservoir sedimentation using land use factors in Dong Nai Watershed as a case study. It is especially important in the Dong Nai Watershed where the soil is highly erodible and forest conversion for agricultural cropping is serious. Data on land use and land cover change in the Dong Nai Watershed during the period 1990 to 2000 was employed in assessing and modelling. The mathematical statistical models for predicting reservoir sedimentation (RS) were formulated through multiple regression analysis using data from Tri An reservoir to correlate with hydrological and various land use parameters during the period 1990 to 1998. All significant variables (annual rainfall, X₁, forestland, X₃, natural forest, X₄, Plantation forest, X₅, and agricultural land, X₆) have correlation to the RS with r-value higher than 0.5. Among the factors contributing to the reservoir sedimentation, natural forestland and RS have a very high negative correlation (-0.92), while agricultural and RS have a positive correlation (0.936). Annual rainfall and RS have the lowest positive correlation (0.78). Among the landuse and hydrological factors used in deriving the relational function, only three parameters were statistically selected as significant factors for the prediction of RS. These are average annual rainfall (X₁), forestland (X₃), and agricultural land (X₆). The most suitable equation suggested for RS prediction for Dong Nai Watershed is:

RS = 1491.5 _* X₁^0.09129_* X₃^{- 0. 02409} _* X₆^{0. 49519}

Based on the above model, it could be postulated that for every 10 percent of forest in the watershed converted into conventional cropping area, RS would increase about 8-9 percent annually.

Introduction

The Tri An hydroelectric power plant, with an installed capacity of 400 MW and 1.7 billion KWh/year generation capacity is located on the Dong Nai river, at 65 km from HoChiMinh City, in the North - East direction. The plant serves as a prime force in enhancing the socio-economic development of the Southern Vietnam. The main function of Tri An power plant is energy production for supplying this part of the country. In addition, it also provides irrigation water to Dong Nai, Binh Duong provinces and HoChiMinh city and contributes in driving back seawater penetration.

The Dong Nai river is originated in Lang Biang High plateau (Lam Dong province) with an elevation of 1770m. The Dong Nai River flows in north-east to south-west and north to south directions though almost all provinces of South Vietnam. The upper part of the watershed at the former waterfall of Tri An (now a hydroelectric power plant) consists of 4,100 km² including the tributary La Nga river.

The Tri An hydroelectric power plant construction were completed in 1991, after 7 years construction with full investment of the Government Vietnam. Since then there have been many factors affecting the life span of reservoir. For example, the conversion of forest area for agriculture and rural development caused remuneration soil erosion and sediment transported to downstream area. The land conversion and farming system upstream of DongNai river alter the flow regime and water quality specially sediment contribution to the Tri An reservoir.

The results of this sedimentation is the decrease in storage capacity of the Tri An reservoir and other alteration in the watershed. In order to prolong the life span of plant, which have been positively affected on the downstream zone, in term of electric production, flood control, etc., sediment transportation to this reservoir should be investigated in relation to change of land use and practices so that mitigation plan by ways of watershed management could be formulated and implemented.

With the aims at providing the means for watershed management and understand the causal effect on sediment contribution to Tri An reservoir, this research is hereby proposed, namely: "Effect of Land use/Land cover Changes and Practices on Sediment contribution to the Tri An Reservoir of Dong Nai Watershed, Vietnam".

Objectives

In order to formulate watershed management plan to reduce sediment contribution to the Tri An reservoir, the specific objectives of this study are as follows:

1. To assess land use/land cover change in Dong Nai watershed upstream the Tri An reservoir during the period from 1991 to 2000.

2. To investigate the relationship between the land use/land cover change and sediment contribution to the Tri An reservoir during the period from 1991 to 2000.

3. To assess the impact of land use practices during the period from 1991 to 2000 in Dong Nai watershed on sediment contribution to the Tri An reservoir during the mentioned period.

Methodology

1. Data Collection

1.1 Types and Sources of Data

Available data and information related to the sedimentation in Tri An reservoir such as map, statistic data, forest area, forest cover, and other the related data have already collected by the offices of local authorities and relevant professional institutions.

Methods

2. Model Formulation

2.1 Mathematical Model Formulation

1) Models for predicting annual land-use changes

In order to obtain year-by-year land use changes, the Markov Chain model. The Markov Chain model was applied to determine probability of land use change based on the land use evolution between two given periods. The general form of the model to predict land use change from 1^st date (year) to the 2^nd date (year) is expressed herein as:

Where g_ij: is probability of change determined from overlaying of two different periods of land use map. The prediction of the next (forward and backward) period of land use distribution can be expressed as:

(Proportion of land use of the first date) * (Matrix of probability of land use change) = (Proportion of land use of the second date)

This can be transformed (backward) in general matrix multiplication as:

2) Analyzing relationship between sediment contribution and driving factors

Simple regression and multi-regression analysis were applied for analyzing the relationship between sediment contribution and land use change (as the remaining forest area in the year considered) together with other driving factors as:

Y = f (X₁, X₃, X₄, X₅, X₆, X₇,)

Where

Y = annual sediment deposition to the reservoir (m³);
X₁ = annual basin rainfall (mm);
X₃ = forestland area (ha);
X₄ = natural forestland (ha);
X₅ = plantation forestland (ha);
X₆ = agricultural land (ha);

The correlation between the dependent (Y) and independent (X) variable was determined by the coefficient of determination (R²) and the F - test ratio. The data analysis of the affirmation recompilation used the STARTGRAPHIC Plus 3.0, available software.

Results of Study

1. Relationships between LUCC and Tri An reservoir sedimentation

Based on Tri An Reservoir Sedimentation data from Tri An hydro-electric power plant and Institute of Water Resources in Ho Chi Minh City and the remaining land use cover together with annual rainfall as shown in Table 2, the correlation matrix among parameters was analysed and shown in Table 4. The relationships between land use/land cover change and reservoir sedimentation were derived as shown in Table 5. The relationships can be described as follows:

Table 2. Observation data Rainfall, Forestland, Agriculture, Openland and Tri An Reservoir Sedimentation in Dong Nai Watershed, Vietnam.

Source: Rainfall from Department of Meteorological in Dong Nai Province;

Note:

Based on the average of suspended sediment of Tri An reservoir observed by Institute of Water Resource in Ho Chi Minh City during 1990 to 1998, annual catchment erosion rate in Dong Nai Watershed can be summarized in Table 2 and interpreted in general as rather low compared to those annual rates of erosion in India, Pakistan, Burma, Indochina, Thailand, and USA as shown in Table 3.

Table 3. Annual catchment erosion rate in India, Pakistan, Burma, Indochina, Thailand, and USA.

Source: Applied Hydrology and Research Section, EGAT (1991)

Table 4. Correlation matrix between land use factors and Reservoir Sedimentation (RS) in Dong Nai Watershed during 1990 - 1998.

Note: (2) & (3) separated from (1).

3.1 Factors Contributing to Reservoir Sediment

Among of factors contributing to reservoir sediment, natural forestland and RS have a very high negative correlation (-0.92), while agricultural and RS have positive correlation (0.936). The annual rainfall and RS have positive correlation with r = 0.78. Logically speaking, the more forest and natural forestland the less reservoir sedimentation was indicated, while the plantation forest and agricultural land contribution greater sediment to the reservoir. In Dong Nai watershed, the plantation forest have just started since 1990, therefore, the first stage of plantation forest and contribute greater sediment to the reservoir. In long-term perspective, the more plantation forest would yield of the less reservoir sedimentation. However, the higher annual rainfall could be considered as major factor causing sedimentation in this reservoir.

According to Table 5, the non-linear regression implies the better prediction of RS than the simple linear with the SEE much lesser even though R² and F-ratios are almost the same values. The best fit of by simple linear regression equation which could be employed to predict RS is:

RS = 369488 + 2.61989 _* X_6; R² = 0.88; F-ratio = 48.17; SEE = 27837.7

where

RS = Reservoir sedimentation;
X₆ = agricultural land.

Parameter X₆, the agricultural land, also produce a good correlation with RS with smallest SEE, when transforming into power functional relation (see Table 5).The derived equation is:

RS = 357.92 _* X₆ ^0.6405; R² = 0.87; F -ratio = 48.17; SEE = 0.0116

Based on the statistical parameters as shown in Table 5, the agricultural land parameter (X₆) indicates the largest influence on reservoir contribution and could be considered as the most suitable predictor for RS estimation in this watershed.

3.2 Suggested Model for Reservoir Sediment Deposition Prediction

In order to reduce SEE with remaining the same value of R² and F-ratio as the linear equation, the non-linear multiple regression analysis was attempted using the same parameters, the power and exponential models were employed and the results are shown in below equations. The derived equations yielded almost the same R² and F-ratio but much less in SEE. The more reliable relationship between RS and contributing factor X₁, X₃, and X₆ should be thus in the non-linear multiple regression model. Equations are thus recommended for predicting RS in Dong Nai watershed. The equations are:

RS = 1491.5 _* X₁^0.09129_* X₃^{- 0. 02409} _* X₆^{0. 49519;} R² = 0.92; F-ratio = 19.21
RS = e ^{13.43 + 0.000048X}₁^{- 0.00000053X}₃^{+ 0.00000172X}₆; R² = 0.92; F-ratio = 18.89

Both models are not much different in F-ratio, while the first equation is easier application in practice, Logically, the second equation is more reasonable than the first equation, because the second equation is exponential function and the first equation is power function. Based on the second equation, the assumption is without agricultural land, reservoir sedimentation (RS) will become zero, while the first equation RS still get some. In this case, I would like to suggest the second equation for predicting reservoir sedimentation, because it has highest R², F-ratio with smallest SEE.

Based on the above equation, it could be postulated that for every 10 percent of converting forest in the watershed into conventional cropping area, about 8-9 percent of RS could be annually increased. In Thailand, Tangtham and Lorsirirat (1993) found that every 10 percent of converting forest in the watershed into conventional cropping area, about 4-5 percent of RS could be annually increased.

Table 5. Relationships between Land use/Land cover Changes and Tri An reservoir sedimentation in Dong Nai Watershed.

Remark: SEE = Standard Error of Estimate.

Conclusion

This study is aimed at assessing factors contributing to reservoir sedimentation as well as deriving the predictive models of reservoir sedimentation using land use factors in Dong Nai watershed. It is especially important in the Dong Nai watershed where the soil is highly erodible and forest conversion for agricultural cropping is in serious condition. Land use and land cover change in Dong Nai watershed during 1990 to 2000 was employed in assessing and modelling. The mathematical statistical models for predicting reservoir sedimentation were formulated through multiple regression analysis using data from Tri An reservoir to correlate with hydrological and various land use parameters. The results can be concluded as follows:

1. Land use and land cover change in Dong Nai watershed during 1990 to 2000 indicate the largest increase agricultural land. It was increased about 75,810ha or 13 percent of the study area, while forestland was decreased about 33,196ha or 6 percent of the study area. The smallest increased category was urban/settlement area (133ha or 0.04 percent of the study area).

2. All significant variables (annual rainfall, X₁, forestland, X₃, natural forest, X₄, Plantation forest, X₅, and agricultural land, X₆), have correlation to the RS with r-value higher than 0.5. Among of factors contributing to reservoir sediment, natural forestland and RS have a very high negative correlation (-0.92), while agricultural and RS have positive correlation (0.936). The annual rainfall and RS have lowest positive correlation (0.78).

3. Among landuse and hydrological factor used in deriving the relational function, only three parameters were statistically selected as significant factors in predicting reservoir sedimentation (RS). These are average annual rainfall (X₁), forestland (X₃), and agricultural land (X₆). The most suitable equation suggested for RS prediction for Dong Nai Watershed is:

RS = 1491.5 _* X₁^0.09129_* X₃^{- 0. 02409} _* X₆^{0. 49519}

Since it has the highest R² with smallest SEE and very highly significant in prediction. It could be postulated basing on the above model that for every 10 percent of converting forest in the watershed into conventional cropping area, about 8-9 percent of RS could be annually increased.

4. The suggested prediction equation containing forestland (X₃) and agricultural land (X₆) parameters also implies management techniques in reducing RS by means of landuse practices in Dong Nai Watershed. The upland soil of the Dong Nai Watershed which have been used for shifting cultivation should be placed under soil conservation such as contour banking for cultivation so that on site soil erosion could be reduced while the farmers could gain benefit from using the area as agricultural land.

5. This case study has been based on data from Tri An reservoir (1990 - 1999) only, which is not adequate for mathematical statistical analysis. At least, 20 year of observed reservoir sedimentation data should be used in order to satisfy a high level of statistical confidence. The long-term analysis of this kind of relationship including other hydrological characteristics such as annual, seasonal and daily discharges, if could be made available, should be taken into consideration. Seasonal land use changes and practices should be also attempted for better prediction.

Literature Cited

Applied Hydrology and Research Section. 1991. Summary of the Result of Sediment Survey in the Reservoirs. Meteorology and Hydrology Division, Survey and Ecology Department. Electricity Generating Authority of Thailand.

Bich, L.N., H.L. Tuy, and N.D.Vuong (1999) Study sedimentation in Tri An Reservoir. Institute of Water Resource Research in Ho Chi Minh City. VIETNAM. Agricultural Pub. 201-206p.

Brooks, K.N., F.F. Peter, M.G. Hans, and L.T. John.1992.Hydrology and the Management of Watershed. Iowa State University Press, Ames - USA,392p.

Dong Nai Statistical Department.2000. Statistical Yearbook. MARD. Vietnam.190p.

Department of Land Development.1999. Report of Land use in Dong Nai in 1999. MARD.Vietnam. 20p. (in Vietnamese).

Tangtham.N. and K.Lorsirirat.1993. Prediction Models of the Effect of Basin Characteristics and Forest Cover on Reservoir Sedimentation in Northeastern Thailand. Kasetsart J. (Nat. Scl.)27: 230 - 235p.