H11G-0933:
Evaluating the Potential of Radar-based Rainfall Estimates for Stream Flow Simulation in the Philippines

Monday, 15 December 2014
Catherine Cristobal Abon1, David Kneis1, Axel Bronstert1, Irene Crisologo2, Carlos Primo C. David2 and Maik Heistermann1, (1)University of Potsdam, Potsdam, Germany, (2)National Institute of Geological Sciences, University of the Philippines, Quezon City, Philippines
Abstract:
This study evaluates the suitability of radar-based quantitative precipitation estimates (QPE) for the simulation of stream flow in the Marikina River Basin (535 km2), The Philippines. We used observed reflectivity from the wet season period of 2012 and 2013 from an S-band radar near Subic. Radar data processing and precipitation estimation were carried out using the Open Source library wradlib. To evaluate the potential added value of radar-based QPE, we generated a benchmark precipitation product based on the interpolation of rain gauge observations (GO product). The GO product was also used to quantify rainfall estimation errors at the point scale. For stream flow simulation, we used a semi-distributed conceptual hydrological model based on the Open Source ECHSE framework.

At the point scale, the radar-based QPE was benchmarked against the GO product at daily and hourly accumulation intervals. It turned out that the radar-based QPE outperformed the GO product in the 2012 while the performance was similar in the 2013. For both periods, estimation errors substantially increased from the daily to the hourly accumulation intervals, most likely due to a lack of representativeness at the point scale.

Interestingly, though, the hourly rainfall estimates allowed for a good simulation of observed stream flow when used to force the hydrological model. In particular, the two main flood events, induced by an enhanced South-West monsoon, are well represented using both hourly rainfall products. The results show that the quality of the simulated stream flow was well in line with the point-based verification: while the radar-based QPE clearly outperforms the GO product in 2012, both perform similarly in 2013. The hydrological model had been recalibrated for each rainfall product to allow for a fair comparison of the two competing rainfall products.

As the Marikina River Basin has a comparatively dense rain gauge network, the results of this study are encouraging with respect to usability of the radar-based QPE in other parts of the Philippines where dense rain gauge network are not available.