Impact of the selected boundary layer schemes and enhanced horizontal resolution on the Weather Research and Forecasting model performance on James Ross Island, Antarctic Peninsula

• Main Authors: Matějka, Michael, Láska, Kamil
• Format: Article in Journal/Newspaper
• Language: English
• Published: Masaryk Univerzity 2022
• Subjects: polar meteorology; numerical simulation; WRF model; air temperature; snow cover; wind speed; Antarctic Peninsula; Antarctic; Ross Island; Antarc*; James Ross Island; Sea ice
• Online Access: https://journals.muni.cz/CPR/article/view/25396

The output of the various Weather Research and Forecasting (WRF) model configurations was compared with ground-based observations in the northern part of James Ross Island, Antarctic Peninsula. In this region, a network of automatic weather stations deployed at ice-free sites (as well as small glaciers) is operated by the Czech Antarctic Research Programme. Data from these stations provide a unique opportunity to evaluate the WRF model in a complex terrain of James Ross Island. The model was forced by the ERA5 reanalysis data and the University of Bremen sea ice data. The model configurations include a novel Three-Dimensional Scale-Adaptive Turbulent Kinetic Energy (3D TKE) planetary boundary layer scheme and a more traditional Quasi-Normal Scale Elimination (QNSE) scheme. Impact of model horizontal resolution was evaluated by running simulations in both 700 m and 300 m. The validation period, 25 May 2019 to 12 June 2019, was selected to cover different stratification regimes of air temperature and a significant snowfall event. Air temperature was simulated well except for strong low-level inversions. These inversions occurred in 44% of all cases and contributed to a higher mean bias (2.0–2.9°C) at low-elevation sites than at high altitude sites (0.2–0.6°C). The selection of the 3D TKE scheme led to improvement at low-elevation sites; at high altitude sites, the differences between model configurations were rather small. The best performance in wind speed simulation was achieved with the combination of the 3D TKE scheme and 300 m model resolution. The most important improvement was decrease of bias at a coastal Mendel Station from 3.5 m·s‑1 with the QNSE scheme on the 700 m grid to 1.2 m·s‑1 with the 3D TKE scheme on the 300 m grid. The WRF model was also proven to simulate a large snowfall event with a good correspondence with the observed snow height.