HYDROLOGIC CHARACTERIZATION OF MANGLA SUBWATERSHED USING REMOTELY SENSED SATELLITE DATA AND CREST MODEL

Abstract

Flood Disaster and Water Resource Management are important issues in Trans- boundary basins. Mangla sub-watershed is an important trans-boundary basin shared by Pakistan and India. The river discharge at Mangla outlet point in the sub-watershed is mainly contributed by snowmelt and rainfall. The sub-watershed needs to be characterized in terms of the contribution of its rainfall-runoff generation for which a distributed hydrologic model Coupled Routing and Excess Storage (CREST) has been calibrated and validated. Rainfall is the main input forcing for the rainfall-runoff model. The satellite rainfall estimates were correlated with the in situ rainfall to know their accuracy. TMPA 3B42 V6, 3B42 V7 and PERSIANN CCS rain estimates respectively showed the average correlation coefficient values of 0.68, 0.83 and 0.42; the average root mean square errors values 56.9, 39 and 91.9 mm/month; the average mean absolute error values 36, 24 and 62 mm/month, and the average relative bias values of 1.98, -2.3 and 26.3%. These statistical values suggest that TMPA 3B42 V7 is the best rainfall estimate for this sub-watershed among the available rain estimates and can be used as a substitute in poorly gauged regions of the higher Himalayas or where data is not readily available. Snow covered areas were mapped using MODIS snow product MOD10A2 for the period from 2000 to 2014. The average monthly results show that maximum snow cover is present during the month of January which is around 55.6% of the sub-watershed at the minimum elevation of 716 meters a.s.l. Minimum snow cover is observed during the month of August covering about 2.2% of the area at an elevation of 3624 meters a.s.l. The snow cover starts accumulating in the month of September till it reaches its maximum in January which then starts melting due to the rise in temperature and snow cover retreats toward higher elevations.

CREST distributed hydrologic model has been calibrated for the time period of 2001 to 2003 at 1 km and 500 m spatial resolution topographic data with NSCE values of 0.64 and 0.62 respectively. The model was validated for the time period of 2004-2012 at the same spatial resolution of topographic data with NSCE values of 0.25 and 0.25 respectively. The model was also validated for a shorter time period of only 2007 giving NSCE values of 0.43 and 0.51. Distributed hydrologic models have the potential to generate discharge values upstream of the outlet for any location along the river. Discharge values for Neelum River at Muzaffarabad were predicted for both the calibration (2001-2003) and validation (2007) modes with the NSCE values of 0.39 and 0.54 respectively. This utility of the model has been used to show its importance in the trans-boundary basins where timely accurate discharge data is needed to deal with the flood disasters of un-expected flows and in the long run water resources management. The model needs an efficient snowmelt component to get more precise estimates of the stream flow at locations where snowmelt is a major contributor toward the river runoff.