Muditha Dantanarayana, Srikantha Herath, SB Weerakoon
Improving accuracy of extreme rain event forecast is very important for the sustainable development of emerging economies, especially for reducing losses and damages in urban areas, which are often the engines of the economy. This study addresses the evaluation and improvement of near future sub-daily scale precipitation forecast for heavy rainfall events in the Kelani river basin in Sri Lanka. Precipitation forecasts of 16 models in 15 minutes temporal resolution were evaluated using Normalized Root Mean Square Error (NRMSE), Temporal Match Percentage (TMP) and their Normalized Standard Deviations (NSD). Station wise model selection base on TMP delivered better performing forecast compared to models selection based on NRMSE.