Incorporating sub-grid heterogeneity into a coarse resolution land surface model

Land surface models are highly susceptive to uncertainty which is largely due to spatial heterogeneity in forcing input and land surface properties. Climate models coupled with land surface models would further enhance uncertainty. Although running a high-resolution land surface model at 1-km or finer resolution is a possible option to reduce uncertainty, the high resolution land surface model is not a viable solution for coupling the land surface model with global or regional climate models. In order to represent heterogeneity at the coarser scale of a climate model, it will be necessary to develop methods of reproducing sub-grid scale variation and patterns without explicitly modeling at a finer scale. Generally sub-grid scale spatial variation in surface properties is made available to the land surface model by using the tiling approach. In this study, we have tested a different approach of using data assimilation and the scaling relationships of key land surface fluxes to reproduce the sub-grid heterogeneity. The scaling relationships are established from running a high resolution land surface model using the NASA Land Information System (LIS) framework. The method is found to be successful in incorporating the sub-grid heterogeneity. The study is expected to help improve simulations by climate models coupled with a coarse resolution land surface models.