Influence of Karst Landscape on Weather Systems: A WRF Model Study on Responses for Different Land and Soil Types

Flood and drought events in karst regions can be more pronounced than those in non-karst regions following a given precipitation distribution. Many karstified rocks are buried under a shallow layer of soil or exposed to the atmosphere, in which case the karst surface would have higher albedo, more evaporation, and less soil moisture. Water and heat capacity, heat transfer, and the configurations of various karst features could provide different land-atmosphere interactions over karst regions. In addition, different land surface types and soils over a karst terrain, such as forest cover, barren scrubland, and sandy soil, may influence local and regional weather events, while providing unique land-atmosphere interactions in karst regions.

This study utilized the Weather Research and Forecasting (WRF) model to study the effects of different land coverage types and soils through regional weather simulations as potential impacts for weather development. Five different convective weather events, ranging from light-moderate rainfall to a historic flooding event, were simulated to test if and how karst landscapes might influence weather events. Considering the fact that the existing model system does not have karst landscapes built-in, we designed modeling experiments to simulate different representations of conditions within karst areas. Model experiments consisted of a Control (default land surface and soil types), Barren (barren ground coverage over the karst region), Forest (forest coverage over the karst region), and a Sandy-soil (surface land types are the same as Control, but use sandy soil over the karst region) run. Model outputs were verified against the Stage-IV precipitation data and compared with the Control run for a sensitivity analysis. Other model-simulated and derived variables, such as temperature, dewpoint, sensible and latent heat fluxes, and convective parameters were used in analyzing mesoscale convective development. The following major conclusions were drawn: 1) for each case, precipitation analysis revealed that the barren ground representation of karst coverage had a more pronounced effect on precipitation than in non-karst regions. The Barren experiment produced more precipitation over dense karst than non-karst areas, where the observed precipitation was light. Evaluation of root-mean-square error (RMSE) and bias (BIAS) scores of precipitation found that the barren ground coverage provided the overall lowest RMSE and BIAS values over the peak-rain period of each event, while both the Forest and Sandy-soil experiments had more intermittent values. Equitable threat score (ETS), as well as the contingency table-based bias score (eBias), suggested that both the Barren and Forest experiments had better skills in simulating light-moderate rainfall amounts. Variables that are dependent basically on local surface conditions, such as surface heat flux, showed the strongest contrast between karst and non-karst regions. Other variables that are dominated by large-scale synoptic systems, such as sea-level pressure, lifting condensation level, and free convection level showed smaller contrasts between the two regions. 2) Significant sensitivity responses were found for the barren ground coverage over the karst region, including a pronounced warming effect on the atmosphere. In contrast, the forested coverage caused a cooling effect over the karst region. Generally, the warming effect was stronger than the cooling effect. 3) Overall, the results indicated that the barren ground coverage in a karst region provides increased temperatures and dewpoints, along with other mesoscale variables, that introduce a persistent atmospheric environment that provides conditions favorable for convective development with improved simulated precipitation.