IMPROVING THE HYDROLOGICAL CYCLE IN LAND SURFACE CLIMATE MODELS

Abstract

The hydrological components of land surface climate models have increased greatly in complexity over the past decade, from simple bucket models to multilayer models including separate and distinct soil water and ground water components. While the parameterizations included in these models have also increased in complexity, the fundamental ability of the numerical solution for the vertical movement of soil water in the Community Land Model (or other land surface models) to simply maintain the hydrostatic solution of the original partial differential equation has yet to be determined.Also, the ability of current generation reanalysis products to simulate near surface quantities as gauged by flux tower measurements has yet to be determined.This study demonstrates that the numerical solution as used in CLM3.5 cannot maintain the hydrostatic state. An alternate form of the equation, titled the Modified Richards equation is presented so that the numerical solution maintains steady statesolutions. Also, an improved and simple bottom boundary condition is derived that itself doesn't destroy hydrostatic initial conditions. The new solution is demonstrated to be as accurate as proven numerical solutions while being one to three orders more computationally efficient. The Modified Richards equation together with the new bottom boundary condition is shown to improve the ability of CLM to simulate soil water, water table depth, and near surface turbulent fluxes.Comparison with flux tower observations shows that ERA-Interim better simulates near surface temperature and wind speed than other current generation reanalysis products. Reanalysis products are able to reproduce the flux tower observations on monthly timescales, and the errors between the products and the measurements are primarily due to biases. However, at six hourly timescales the errors are not only larger but also caused primarily by a lack of correlation with the observations.