Assessment of change in future flow of Brahmaputra basin applying SWAT model using multi-member ensemble climate data

Abstract

Brahmaputra basin plays an important role in four countries namely China, India, Bhutan, and Bangladesh for their three physiographic zones. The Brahmaputra River contributes about 67% of the total annual water flow of Bangladesh. Different studies predicted that climate change with combinations of land use changes, directly affect the magnitude and timing of the runoff, the intensity of floods, droughts, and fresh water availability over the country.Several studies suggest that SWAT (soil and water assessment tool) is effective tool because of physically based and distributed hydrological model. Moreover, many studies worked on the impact of climate change on water availability for different country using the output of GCMs to SWAT hydrology model.Nevertheless, most of the studies did not cover all sets of GCMs in the hydrology model. Hence, in this study, SWAT hydrology has been simulated using climate predictions generated by a multi member ensemble regional climate model for a high resolution (25km) domain over the Brahmaputrabasin to capture the wide range of uncertainty in the future climate change. A 90-m resolution digital elevation model (DEM) derived from the Shuttle Rader Topography Mission (SRTM) has been used to delineate catchment boundary. A 300-m resolution of land use data of 2009-2010 has been used. This data has been reclassified to match the SWAT land classes. The soil map of the catchment has been extracted from the FAO digital soil map of the world. The gridded rainfall data have been obtained from TRMM (Tropical Rainfall Measuring Mission), GPCP (Global Precipitation climatology Project) and APRHODITE (Asian Precipitation - Highly- Resolved Observational Data Integration towards Evaluation) and temperature data from Era-interim. Discharge data from Bahadurabad gauge station of the Brahmaputra riverhas been used for model evaluation as flow data in the upper part of the catchment inside India are not available. The evaluation process comprises of sensitivity analysis, calibration and validation. Based on the availability of APHRODITE and GPCP data, calibrated period has been selected from 1998 to 2002 and validated period has been selected from 2003 to 2007. However, calibration period has been considered from 2000 to 2004 and validation period has been considered from 2005 to 2009 when model is derived by the TRMM data sets. In addition, one year has been considered as warm-up period for both calibration and validation. A warm-up period allows the model to get a fully operational hydrological cycle and thus helps to stabilize the model. The main methods used in modeling the hydrologic processes in SWAT were curve number method for runoff estimating, Hargreaves method for PET and Muskingum method for channel routing. Finally, flows of the Brahmaputra basin are generated using climate forcing by the multi-member QUMP ensembles experiments of PRECIS for the early century (2011-2040), midcentury (2041-2071) and end century (2071- 2099). Performance of the SWAT models are evaluated using a set of standard and widely used indicators. For the model calibration using daily data, Nash-Sutcliffe efficiencies are found as 0.77, 0.77, and 0.78 using TRMM, APRHODITE and GPCP data, respectively. The coefficient of determination (R2) has been found as 0.83, 0.92 and 0.88 using TRMM, APRHODITE and GPCP data, respectively. The RSR ((RMSE-observations standard deviation ratio) value has been found within an acceptable range (<0.5) using the three data sets. However, for the validation period,except the TRMM data, the RSR values showed unacceptable results (>0.5). When the predicted rainfall considering climate change forced model,the median value of monthly flow from May to July period have been found to be 20% ,17% and 14% increasefrom the baseline for 2020s, respectively. In 2050s, the above values can be found as 12% ,13% and 9% from baseline and in 2080s,they can be found as 10% ,12% and 13% from baseline, respectively.

The chance of change of monthly flow for Augustmight increase 4% for 2020s. Those figures might increase to 5% and 8%for 2050s and 2080s, respectively.However, the inter-quartiles of mean flow in March simulated by the RCM ensemble members are very high. This shows the level of uncertainties of future predictions is high. On the other hand, results obtained by the simulation of SWAT model using multi-ensemble members agree that the possibility of increase of the flow in June and July will be high. Although the percentages of increase of flow in the pre-monsoon season are higher than that of monsoon flow, the volume of monsoon flow will be ten times higher than that of pre-monsoon. Therefore, flooding in this basin will be increased in the future.