Model performance improvement from fine resolution satellite LULC/NDVI and spin-up of soil layer

The bottom boundary conditions of numerical models play an important role in model simulation performance as they involve the calculated surface fluxes that invigorate the turbulence in and dominate the diurnal cycle of the lower part of overlying atmosphere. The surface properties, such as albedo, emissivity, vegetated species, roughness, of different types of LULC profoundly affect the surface fluxes partitioning and the turbulent stress. However, the default LULC data that comes with common weather research or prediction model packages may not be up-to-date or with resolution not sufficient for finer scale simulations. On the other hand, soil layer moisture (SM) and temperature (ST) have been shown to significantly alter surface fluxes output. Common simulation practices acquire SM and ST from (re)-analysis fields that are often with much coarser resolution and/or disconnected from the land surface types.

In this study we demonstrate how model performance for applications such as prediction of optical turbulence can be improved with incorporated (updated) fine-resolution satellite products for the LULC and vegetation data for finer scale simulations. We also show how the performance can be further boosted by a preparatory spin-up of the land surface model to produce SM and ST that closely reflect the land cover data in model resolution in our customized modeling practice.