COMPARISON OF SINGLE- AND MULTI-SITE WEATHER GENERATORS IN SIMULATING DAILY PRECIPITATION FOR THE MEKONG RIVER BASIN

Abstract:

Stochastic weather generators can produce synthetic time-series for meteorological variables to closely approximate their observed statistical characteristics. These weather generators are used for developing time-series data for extended periods assuming that the conditions are stationary. Single-site generators preserve the mean and variance. However, some recently developed multi-site weather generators also claim to preserve spatial correlation in addition to the mean and variance of selected variables. Markov Chain based approaches are used in both types of generators used in this study to preserve wet/dry statistics.

The main objective of this study is to evaluate several single- and multi-site weather generators that are used for the generation of the long-term precipitation at the Mekong River Basin (MRB), one of the largest river basins (~840,000 km²) in the world. The MRB covers a wide array of hydro-climatic conditions, terrain and land-cover types. Thus, it provides a very challenging backdrop to test these weather generators. Daily precipitation of Tropical Rainfall Measuring Mission (TRMM) 3B42 (v7) at 0.25-degree resolution from 1998 to 2012 is used in this study as the observed data set.