Regional mean and variability characteristics of temperature and precipitation over Thailand in 1961–2000 by a regional climate model and their evaluation

Abstract

This study presents the characterization of regional means and variability of temperature and precipitation in 1961–2000 for Thailand using regional climate model RegCM3. Two fine-resolution (20 km) simulations forced by ERA-40 reanalysis data were performed, with the default land covers and with a land-cover modification strategy suggested by a previous work. The strategy was shown to substantially alleviate the problem of systematic underestimation of temperature given by the default simulation, for most part of Thailand in both dry and wet seasons. The degree of bias in precipitation tends to vary differently in every sub-region and season considered. The patterns of seasonal variation of both climatic variables are acceptably reproduced. Simulated 850-hPa winds have general agreement with those of ERA-40, but wind speed is overestimated over the Gulf of Thailand during the dry months, potentially bringing excessive moisture to and causing more rain than actual in the south. Long-term trends in temperature are reasonably predicted by the model while those in observed and simulated precipitations for upper Thailand are in the opposite directions. Apart from the conventional methods used in characterization, spectral decomposition using Kolmogorov–Zurbenko filters was applied to inspect the model’s capability of accounting for variability (here, in terms of variance) in both climatic variables on three temporal scales (short term, seasonal, and long term). The model was found to closely estimate the total variances in the original time series and fairly predict the relative variance contributions on all temporal scales. The latter finding is in line with the results from an additional spectral coherence analysis. Overall, the model was shown to be acceptably adequate for use in support of further climate studies for Thailand, and its evident strength is the capability of reproducing seasonal characteristics and, to a lesser degree, trends.