Validation of a multidecadal RCM hindcast over East Asia

Abstract

The precipitation and low-level air temperature in East Asia from a regional climate model (RCM) hindcast for the 22-year period 1979–2000 is evaluated against observational data in preparation for the model use in regional climate change research. Emphasis of the evaluation is placed on the RCM capability in capturing the temporal and spatial variability of precipitation and low-level temperature, especially in conjunction with important climatological events such as, ENSO and East Asian monsoon, at three spatial scales of continental, subcontinental, and river basins. Spatial anomaly correlation time series of geopotential height and temperature show that the simulated upper-air fields remain consistent with the driving large-scale fields, NCEP Reanalysis 2 (R2), throughout the period. The simulated seasonal shifts in 850 hPa winds also agree well with R2 over eastern China and the western Pacific Ocean although the magnitudes of the shifts are overestimated, especially over the eastern slope of the Tibetan Plateau and in northern Manchuria. The simulated precipitation climatology agrees reasonably with that from two analysis datasets based on station- and remote-sensing data. Outstanding characteristics of precipitation including the location of the main rainband, climatological means, and the spatiotemporal variability in association with East Asian Monsoon, ENSO, and extreme events, are well represented in the hindcast. The most notable bias in the simulated precipitation is an overestimation of winter rainfall in southwestern coast of China, near the border with Vietnam. The simulation overestimates the interannual variability of seasonal precipitation especially in southern China, however, the corresponding coefficients of variation agree reasonably with observations except in very dry regions. This suggests that climate sensitivity of scaled precipitation can be useful for projecting climate change signals. The simulated low-level temperature climatology agrees reasonably with observational data as well. The most noticeable biases in the simulated low-level temperature are the warm (cold) biases in southern Siberia (northeastern China) during winter (summer) and the systematic underestimation of low-level temperature in the Tibetan Plateau for all seasons. The daily maximum temperature is underestimated for all seasons by 2−3 K with the largest biases in spring and fall except in the northwestern Mongolia region where it has been overestimated during winter. The daily minimum temperature biases ranges from 0.3 K in spring to 2 K in winter, and are much smaller than those in daily maximum temperature. The evaluation of the multidecadal hindcast shows that model errors mostly confined in the region near the lateral boundaries of the model domain with only minor biases in eastern China. This allows us to be cautiously optimistic about the RCM usefulness for studies of precipitation and low-level temperature changes in East Asia induced by increased emissions of greenhouse gases.