EVALUATION OF SATELLITE-BASED PRECIPITATION ESTIMATES FOR HYDROLOGICAL MODELLING

Abstract

Effective watershed management is crucial for growth, sustainable development and protection of the environment, and prosperity of nation. Due to lack of watershed-wide hydrological data from in situ platforms, whether they are real time or historical, water management has been quite challenging. Under such circumstances, hydrologic models forced with widely available satellite-based precipitation estimates (SPEs) can be useful for generating high resolution watershed-wide hydrological dataset (discharge, sediment yield, nutrient losses etc.) and for decision making on water management. Keeping this in mind, and availability of hydro-meteorological dataset, the present study has been carried out to evaluate the SPEs with the gauge-based IMD gridded dataset using statistical and contingency table methods for agricultural-based Marol and Muneru watersheds, part of the lower Krishna river basin. The accuracy and performance of Tropical Rainfall Measuring Mission (TRMM) precipitation dataset (TMPA-3B42V7) has been found superior among SPES, and has been used further for modelling purpose in the present study. Further, the TRMM precipitation dataset has been evaluated for prediction of daily suspended sediment load (SL) employing support vector machine (SVM) with wavelet transform (WASVM), as they may provide an alternative source of rainfall data for ungauged or data-sparse regions. Overall, performance of both IMD gauge observations and TRMM precipitation driven WASVM models has been found better than the conventional SVM for both the watersheds. The proposed WASVM models can be used effectively for SL prediction up to 6-days lead specifically in the selected watersheds and can be applied effectively in other watersheds of similar hydro-climatic conditions. Furthermore, modelling-based evaluation of TRMM precipitation dataset has been carried out over Marol watershed using the physically based, Soil and Water Assessment Tool (SWAT) hydrologic model. Calibration, uncertainty analysis and validation of the SWAT model, both on daily and monthly basis, was carried out using SWAT-cup SUFI-2 program under three different simulation schemes for the years 1998 to 2011; Simulation schemes I: IMD gauge data benchmarked SWAT calibration and validation, Simulation schemes II: Impacts of TRMM precipitation forcing on gauge-calibrated SWAT model, and Simulation schemes III: SWAT calibration and validation benchmarked by TRMM precipitation dataset. A good agreement between the simulated and observed hydrographs for both discharge and sediment indicates very good performance of the SWAT model with IMD gauge-based precipitation input. Using the IMD gauge data calibrated SWAT model, TRMM precipitation based simulation shows limited vi | P a g e hydrologic prediction on daily time scale, while results were fairly good on monthly time scale. Recalibrating the model with TRMM precipitation input, simulation results shows remarkable improvements for both discharge and sediment, but not as much improvement for already well performing IMD gauge data based calibrated and validated model. To deal with water management issues, analysis and quantification of the different elements of hydrologic processes taking place within the area of interest is must, therefore detailed water balance study was carried out employing the SWAT model. It was found that evapotranspiration is more predominant and accounts for about 46.28% (748.39 mm) of the average annual precipitation (1616.94 mm) falling over the area. About 41.99% (679.01 mm) of average annual precipitation flows out from the study area as surface runoff. Soil erosion status in the Marol watershed was also accomplished to provide the priority of sub-watersheds for soil conservation measures. It has been found that high and very high erosion prone area falling under the watershed are about 24.86% (126586 ha) and 5.47% (27853 ha), respectively. The average annual sediment yield of the watershed was found to be 12.16 t.ha-1yr-1. The SWAT model was employed for evaluating the effectiveness of different management strategies in reducing sediment yield and nutrient loads considering different crops, tillage implements, fertilizer applications and management operations. Reduction of sediment yield and nutrient loss due to alternate cropping treatments (Groundnut and Soybean) was obtained as compared to paddy and maize cultivation. The field cultivator tillage practice (to replace conventional tillage) and conservation practices viz., contour farming and filter strips are recommended in the critical sub-watersheds of Marol watershed to reduce sediment yield and nutrient losses.