The Impact of Land Surface Processes on Simulations of the U.S. Hydrological Cycle: A Case Study of the 1993 Flood Using the SSiB Land Surface Model in the NCEP Eta Regional Model

Abstract

This paper describes a methodology for coupling the Simplified Simple Biosphere Model (SSiB) to the regional Eta Model of the National Centers for Environmental Prediction (NCEP), and presents the application of the coupled system in regional simulation studies. The coupled Eta?SSiB model is used to study the impact of land surface processes and land surface initialization on the regional water and energy cycle in an extreme climate event, by comparing the results from the Eta?SSiB with those from the Eta?bucket model. Simulations from both models spanned 3 months via a succession of 48-hr simulations over June, July, and August 1993, a summer of heavy flooding in the United States. The monthly and seasonal means from the simulations in both model runs are compared. The Eta?SSiB model produces more realistic monthly mean precipitation over the United States and the flood areas. The improvements are mainly manifested in the intensity of the heavy rainfall and its spatial distribution. The results demonstrate that even with a short-term simulation, a more realistic representation of land surface processes and land surface initialization improves the monthly and seasonal means of the simulated regional precipitation for the summer of 1993. In addition to precipitation, the simulations of surface air temperature are also evaluated and they show that the Eta?SSiB model produces reasonable results over most of the United States, with the exception of a cold bias at night in the mountainous western region of the United States. To understand the mechanisms of land surface?atmosphere interactions and the causes for the differences in the Eta?SSiB and the Eta?bucket simulations, the water cycle in the atmosphere?land system and the energy balance at the land surface are analyzed. The changes in (a) spatial distribution and diurnal cycle of surface latent and sensible heat, and (b) low-level moisture flux convergence (MFC) in response to these changes in surface heating are the primary factors for the improvement in the precipitation simulation. That is, the different surface models of SSiB and bucket, and their different soil moisture initializations, produce different energy partitioning in the surface heat fluxes of the Eta Model. The changes in both the daily mean and the diurnal variation at the land surface lead to different boundary layer evolutions and atmospheric stability conditions. In response to these differences, the Eta?SSiB model and the Eta?bucket model produce different low-level MFC in the heavy rainfall area. Strong and persistent MFC was one of the major forces that produced the heavy rainfall in the summer of 1993. In the above experiments, the Eta?SSiB model used the global reanalysis of the NCEP?NCAR (National Center for Atmospheric Research) 40-year Reanalysis Project (NNRP) for its initial soil moisture, whereas the Eta?bucket model used a tuned annual-mean fixed field of initial soil moisture as employed in the then-operational Eta Model. Because of this important initialization difference, a further set of simulations was performed in which the Eta?bucket was initialized with the NNRP reanalysis soil moisture employed in the Eta?SSiB. Results show that with similarly derived initial soil moisture states, the differences between the Eta?SSiB and the Eta?bucket are reduced but still evident, suggesting that improved representation of vegetation in the SSiB is at least partially responsible for the overall improvements in the simulations. Given that the NCEP?NCAR reanalysis is used for initial conditions and lateral and lower boundary conditions in these experiments, this study shows that a coupled atmosphere?biosphere regional model imbedded in a global reanalysis has the potential to provide a more realistic simulation of precipitation in extreme climate events.