Added value of kilometer-scale modeling over the third pole region: a CORDEX-CPTP pilot study

High-accuracy meteorological datasets are urgently required for understanding hydrological processes across the Third Pole (Qinghai-Tibetan Plateau, or TP), where meteorological stations are sparse. Low-resolution weather and climate simulations have significant errors in this region due to their inability to resolve meso-microscale processes associated with the complex terrain and convective clouds. This work presents a contribution to CORDEX Convection-Permitting Third Pole (CPTP) using dynamical downscaling of the latest global reanalysis data produced by the European Centre for Medium-Range Weather Forecasts (i.e., ERA5) at very high resolution (approximately 0.033 degrees) based on the Weather Research and Forecasting (WRF) model. The results show that the kilometer-scale horizontal grid spacing simulation (WRF3) outperforms the ERA5 and the High Asia Refined regional reanalysis (HAR v2) in terms of smaller biases and root mean square errors, as well as higher spatial pattern correlation coefficients for 10-m wind speed and precipitation. Furthermore, WRF3 more realistically reproduces observed night-time precipitation peaks in the interior TP, while ERA5 and HAR v2 show erroneous afternoon precipitation peaks. Therefore, the added values achieved by resolving detailed physical processes when increasing grid spacing are considerable. This work demonstrates the potential for developing a reliable high resolution meteorological dataset required for research in this unique region.

Automated summary

High-accuracy meteorological datasets are urgently needed to understand hydrological processes in the Third Pole, where meteorological stations are sparse. This work demonstrates the potential of developing a reliable high resolution meteorological dataset for research in this unique region by utilizing dynamical downscaling of global reanalysis data at very high resolution. The study shows the advantages of kilometer-scale horizontal grid spacing simulation over other methods, with more realistic representation of night-time precipitation peaks in the interior TP.