Impact of inconsistency of multiple land use/land cover products on permafrost modeling on the Tibetan Plateau

Abstract

The Tibetan Plateau (TP), with an average elevation of over 4000m, has the largest area of permafrost in the mid-latitudes. With global warming, the degradation of permafrost is expected to intensify, which in turn will have significant impacts on regional ecology, hydrology and infrastructure engineering. Accurately modeling the extent of permafrost and permafrost change is now a crucial challenge for climate change research and climate modeling in this region. The simulation of permafrost extent in climate models and land surface models is influence by various factors, including atmospheric forcing, land surface conditions, soil conditions and deficiencies in the model schemes. Land use and land cover (LULC), is a crucial component of surface characteristics in land surface models, is directly affecting the surface energy balance and permafrost simulation results. To investigate the effect of LULC uncertainty on permafrost simulation, we conducted simulation experiments on the TP using the Community Land Model, version 5 (CLM5) with five high-resolution LULC products in this study. Firstly, we evaluated the simulation results using shallow soil temperature data and deep borehole data at several sites. The results show that the model performs well in simulating shallow soil temperatures and deep soil temperature profiles. The effect of different land use products on the shallow soil temperature and deep soil temperature contours is not obvious due to the small differences in land use products at these sites. Although there is little difference in the simulating results of different land use products when compared to the permafrost distribution map, the differences are noticeable for the simulation of the active layer. Land cover had a greater impact on soil temperature simulations in regions with larger land use inconsistency, such as at the junction of bare soil and grassland in the northwestern part of the TP, as well as in the southeast region with complex topography. The main way in which this effect occurs is that land cover affects the net surface radiation, which in turn causes differences in soil temperature simulations. In addition, we discuss other factors affecting permafrost simulation results and point out that increasing the model plant function types as well as carefully selecting LULC products is one of the most important ways to improve the simulation performance of land-surface models in permafrost regions.