Close Window

Oklahoma State University (Curtis Marshall, COMET graduate student fellow): "The importance of soil moisture to forecasting the onset, duration, location, and characteristics of deep convection over land"

Final Report

The atmospheric science community recently has begun to appreciate more fully the impacts of land-surface characteristics and processes on weather and climate. Observational and modeling studies have demonstrated this importance on a full range of temporal and spatial scales. Recognizing the need to more adequately simulate land-surface processes in mesoscale models, the National Centers for Environmental Prediction Environmental Modeling Center recently implemented a modern land-surface parameterization scheme in the mesoscale Eta model, the primary operational numerical weather prediction system for providing short-term forecast guidance to the National Weather Service. To facilitate better the new land-surface scheme, changes were also made to the atmospheric surface layer scheme. These new schemes became operational on 31 January 1996. On 18 February 1997, several changes were made to the new schemes to improve their ability to simulate the fluxes of heat and water vapor from the land-surface--processes that are critical in determining the diurnal evolution of the planetary boundary layer (PBL).

The purpose of this study is to evaluate the performance of the new schemes using Oklahoma Mesonet observations. Three months of operational model data have been compared with Mesonet observations. Three months of operational model data have been compared with Mesonet observations -- May, June, and July of 1997. Model fields of soil moisture and soil temperature are examined. To understand how possible inadequacies in the simulation of land-surface variables adversely affects the thermodynamic structure of the model PBL, surface energy fluxes are examined in detail at the grid point corresponding to the Mesonet site at Norman, OK (NORM). For this study, special instruments were installed at NORM for estimating surface energy fluxes. Additionally, model fields of shelter-level air temperature and specific humidity are compared to Mesonet observations across Oklahoma. The vertical structure of the model PBL is compared with rawinsonde observations.

Results show that during the time of this study, there was a severe positive bias in top-layer soil moisture, primarily because of the initialization approach. An excess of net radiation at the surface, partly because of a positive bias in downward shortwave radiation also appears. For numerous reasons, ground heat flux appears to be underestimated during the daylight hours by the Eta model. Because the land-surface scheme is constrained by energy balance, the excess net radiation and underestimated ground heat flux result in too much available energy for turbulent fluxes (i.e., sensible and latent heat flux). The severe positive bias in soil moisture serves to partition much of this excess toward overestimated latent heat flux. Evidence is also presented to suggest the possibility that entrainment is underestimated at the top of the model PBL. The net result of the model inadequacies outlined herein was a PBL that was often too cool, too moist, and too shallow. Thus, the resulting errors in the thermodynamic structure of the PBL can have serious consequences on model stability indices and the prediction of warm season precipitation.

Future work is outlined that will address the model inadequacies documented in this thesis. This work includes the development of a new Land Data Assimilation System (LDAS) for the operational Eta model, and possible testing of a new scheme for vertical diffusion in the PBL.