VIC model calculations of moisture distributions.

VIC Flux Terms

VIC Flux Terms

VIC Flux Terms

 

    The Variable Infiltration Capacity (VIC) model, a macroscale land surface hydrologic model (Liang et al. 1994), was used to simulate land surface hydrologic variables. The model can be run either in full energy mode as well as water balance mode. In this project, the model was run in water balance mode.

Historical Data Records: Global Runoff Data Center

Historical data records were obtained from the Global Runoff Data Centre (GRDC, http://www.bafg.de/GRDC/EN/Home/homepage__node.html). (see under Naturalized Flow).

Calibration and Validation

    The VIC model was calibrated using multi objective calibration method where the Nash Suctcliffe Efficiency (NS) and Correlation Coefficient (R) are the objective functions used. One station from each basin: Uchku (Syr Darya) and Garm (Amu Darya), were used for calibration. The parameters calibrated were: Ds, Ws, binf, D1, D2 and D3 based on the general guideline to VIC model calibration . The years 1950 – 1960 were used for calibration while the validation was done using 1961 – 1990 data. The observed flow data was obtained from GRDC.The monthly flow time series is bias corrected using quantile mapping bias correction method to avoid systematic bias. The quantile mapping method uses the empirical probability distributions for observed and simulated flows to remove biases. The quantile mapping method adjusts a simulated flow based on the cumulative distribution (CDF) of observed flow; so that the simulated and observed flows have the same non-exceedence probability. This method of bias correction was used by ( Hashino, et. al. 2007; among others). After bias correction, the bias-corrected simulated flow will be replaced by an observed flow with similar CDF value.

    The VIC simulation of the two basins is very promising during both the calibration and validation periods based on the above objective functions used. The R and RMSE values are depicted in Table 1. For Garma station, the RMSE value lower during the validation period is lower than the during the calibration period. However, the R is higher during period. As it is usually the case the simulation performs better during the calibration than the validation period for the Uchku period. The stream flows are bias corrected using quantile mapping method. As a result of bias correction, the RMSE greately reduced (Table 1) while the correlation remains the same as bias correction doesn’t affect the pattern of the flows which affects the correlation between two variables. Figure 2 depicts the monthly flow time series and seasonal flow for the two station both during calibation and validation periods. The graphs are of the bias corrected flows. The VIC simulation is good even without bias correction during calibartion as well as validation periods as can be seen from the high correlation between the observed flows and simulated flows as well as the low RMSE values. The bias correction reduced the RMSE by nearly 15% to 38% from the not bias corrected RMSE.

Water Balance Terms

   All the three water balance terms, P, ET and RF, showed similar pattern for the two basins. P peaks between February and May, while the low season is from August to October. The highest P for Amu Darya occurs during March (109.3 mm) while April is the highest P month for Syr Darya (68.3 mm/month). The average monthly P is generally small for both basins due the fact that they are located deep inside the continental range far from the coastal areas. As a result, water carrying winds are depleted of their moisture by the time they reach to these basins from the coastal areas. The lowest P month is August for Amu Darya (6.3 mm/month) and September for Syr Darya (12.5 mm/month). ET’s peak moths are April through July for both basins. June has the highest ET for Syr Darya while May has the highest ET for Amu Darya. As in the case of P, ET is general very small ranging from 4.6 – 48.2 mm/month for Syr Darya and from 6.9 – 46.8 mm/month for Amu Darya. September to February is the low ET months. RF’s peak moths are April through July for both basins. May has the highest RF for Syr Darya while June has the highest ET for Amu Darya. RF ranges from 13.1 – 54.3 mm/month in Amu and 10 – 41.6 mm/month in Syr Dayra. On a mean annual basis, en annual P, ET, RF and SNMT (snow melt)., all the flux terms are highest towards east and southeast of Aral basin. The values of the flux terms get smaller to the west and northwest of the Aral basin. Particularly, the spatial correlation among P, RF and SNMT is very strong with correlation coefficient higher than 0.9.