An evaluation of the BCC/RCGWG"s performance in simulating daily nonprecipitation variables in China
-
-
Abstract
The BCC/RCGWG is a stochastic weather generator developed for simulating daily precipitations in China. In this study it is further developed by adopting a framework based on the Fourier series with multivariables weak stationary process assumed for daily nonprecipitation variables including maximum temperature, minimum temperature, sunshine hour, relative humidity and average wind speed. These nonprecipitation variables are made dependent on the simulated precipitation by the previous version of the BCC/RCGWG for daily precipitation. Using daily observations of the five nonprecipitation variables from 1971 to 2000, the parameters for each nonprecipitation variable of the further developed weather generator are determined for the 669 stations all over China. Based on the estimated parameters, daily nonprecipitation variables at these stations for any period of time can be simulated. A 100 year simulation is made and compared with observations during 1971-2000 in terms of annual, monthly statistics and extreme climate events. The results of monthly and annual statistics are fairly satisfactory. Root mean square error (RMSE) of annual maximum temperature, minimum temperature, sunshine hour, relative humidity and average wind speed are 0.2℃, 0.2℃, 70.9 h, 0.6% and 0.2 m/s, respectively. The RMSEs of monthly maximum temperature, minimum temperature, sunshine hour, relative humidity and average wind speed are between 0.4-0.7℃, 0.4-0.7℃, 10-20 h, 4%-14% and 0.6-0.9 m/s, respectively. These demonstrate the usefulness of the weather generator BCC/RCGWG for the five nonprecipitation variables. However, simulations of extreme climate events are less satisfactory.
-
-