Data-Based Mechanistic Rainfall-Runoff Modelling for a Large Monsoon-Dominated Catchment in Thailand

Hydrological modelling for water resource and flood management in large monsoon-dominated sub-tropical catchments has not been the subject of extensive research and it is not clear what the appropriate model structures and data requirements may be. High degrees of seasonality, limited data availability, rapidly changing hydrological regimes as a result of land use change and climate variability and a lack of complete understanding of the details of the physical hydrology in these regimes and regions all contribute to this situation. This paper uses Data Based Mechanistic (DBM) modelling methods to explore the hydrology of the 3,853 km2 Mae Chaem catchment in northern Thailand, where there is an unusually rich database of runoff and rainfall data. This is used to examine the appropriate model structure and parameter values in DBM models and the effects of using the available rainfall and runoff data in a range of different ways. Rainfall data are area-weighted using Theissen polygons, within which altitude adjustment is effected on the basis of evidence for an increase of about 0.5 mm of rain per rainday for each 100 m increase in elevation above 1000 m, in the monsoon season only. The model structure suggests a second-order model and the parameter values seem to be rather stable when higher quality rainfall data are used. Furthermore, it is possible to maintain reliable flow simulations by cascading a series of runoff prediction regression models that predict a downstream flow from an upstream flow and the incremental rainfall between gauging stations.