Effects of spatial resolution on WRF v3.8.1 simulated meteorology over the central Himalaya

Abstract

Abstract. The sensitive ecosystem of the central Himalayan (CH) region, which is experiencing enhanced stress from anthropogenic forcing, requires adequate atmospheric observations and an improved representation of the Himalaya in the models. However, the accuracy of atmospheric models remains limited in this region due to highly complex mountainous topography. This article delineates the effects of spatial resolution on the modeled meteorology and dynamics over the CH by utilizing the Weather Research and Forecasting (WRF) model extensively evaluated against the Ganges Valley Aerosol Experiment (GVAX) observations during the summer monsoon. The WRF simulation is performed over a domain (d01) encompassing northern India at 15 km × 15 km resolution and two nests (d02 at 5 km × 5 km and d03 at 1 km × 1 km) centered over the CH, with boundary conditions from the respective parent domains. WRF simulations reveal higher variability in meteorology, e.g., relative humidity (RH = 70.3 %–96.1 %) and wind speed (WS = 1.1–4.2 m s−1), compared to the ERA-Interim reanalysis (RH = 80.0 %–85.0 %, WS = 1.2–2.3 m s−1) over northern India owing to the higher resolution. WRF-simulated temporal evolution of meteorological variables is found to agree with balloon-borne measurements, with stronger correlations aloft (r = 0.44–0.92) than those in the lower troposphere (r = 0.18–0.48). The model overestimates temperature (warm bias by 2.8 ∘C) and underestimates RH (dry bias by 6.4 %) at the surface in d01. Model results show a significant improvement in d03 (P = 827.6 hPa, T = 19.8 ∘C, RH = 92.3 %), closer to the GVAX observations (P = 801.4 hPa, T = 19.5 ∘C, RH = 94.7 %). Interpolating the output from the coarser domains (d01, d02) to the altitude of the station reduces the biases in pressure and temperature; however, it suppresses the diurnal variations, highlighting the importance of well-resolved terrain (d03). Temporal variations in near-surface P, T, and RH are also reproduced by WRF in d03 to an extent (r>0.5). A sensitivity simulation incorporating the feedback from the nested domain demonstrates the improvement in simulated P, T, and RH over the CH. Our study shows that the WRF model setup at finer spatial resolution can significantly reduce the biases in simulated meteorology, and such an improved representation of the CH can be adopted through domain feedback into regional-scale simulations. Interestingly, WRF simulates a dominant easterly wind component at 1 km × 1 km resolution (d03), which is missing in the coarse simulations; however, the frequency of southeasterlies remains underestimated. The model simulation implementing a high-resolution (3 s) topography input (SRTM) improved the prediction of wind directions; nevertheless, further improvements are required to better reproduce the observed local-scale dynamics over the CH.