资源环境科技发展态势分析平台

The episodes of heat wave events have strengthened in recent decades causing great concern for human health, agriculture and natural ecosystem. In the present study, Regional Climate Models (RCMs) namely, CCAM and RegCM, from Coordinated Regional Climate Downscaling Experiments (CORDEX) for South Asia (SA) are evaluated for simulating heat waves (March–June) for a long-term period (1971 to 2005) over India in comparison with observations from India Meteorological Department (IMD). The statistical analysis (correlation, RMSE, MAE, ECDF) results reveal differences in RCMs in simulating spatial pattern and trends of maximum temperature before bias correction. Variance scaling bias correction is found to remove bias and improve model simulations in capturing temperature variability. An increase in correlation in daily observations from 0.24 to 0.70 and reduction in RMSE from 8.08 °C to 2.02 °C and MAE from 3.87 °C to 2.43 °C after bias correction is observed between model and observation.

LMDZ4 and GFDL-ESM2M are found to perform best in simulating interannual variability of seasonal mean maximum temperature with an underestimation of −7.74% and − 15.41% which improved significantly to around −1.51% and − 0.78%, respectively after bias correction over India. LMDZ4 and GFDL-ESM2M (RegCM ensemble) are best-performing models (ensemble) in significantly reproducing the heat wave frequency and spatial variability in closer proximity with observations over India amongst all models after bias correction. Over NW and western regions, the LMDZ4 and GFDL-ESM2M ensemble models successfully capture the increasing trend of 0.2 events/year and 0.4 events/year accordance to IMD and IITM criteria, respectively. However, the ACCESS1.0, CNRM-CM5 and CCSM4 ensemble experiments overestimated heat waves by ±40 events in most sub-divisions in India. Over the central Indian regions, the ACCESS 1.0 and CNRM-CM5 model output show a negative trend of −0.2 events/year and large spatial variability possibly due to model associated uncertainties. Overall the results show an improvement in capturing maximum temperature and heat waves across the regions of Indian sub-continent in the bias-corrected downscaled CORDEX-SA ensemble RCMs than without bias-corrected output. The study suggests a way forward to assess RCMs performance and uncertainty in extreme weather analysis in future projections.