Evaluating the ETA Model Against Observations at Selected U.S. Sites

• Main Author: Yucel, Ismail
• Other Authors: Shuttleworth, William James
• Language: en_US
• Published: The University of Arizona. 1996
• Online Access: http://hdl.handle.net/10150/626796; http://arizona.openrepository.com/arizona/handle/10150/626796

The performance of the NCEP' s Eta model in providing near-surface weather fields and diagnosed surface exchange fields via the process of four-dimensional data assimilation (4DDA) was investigated in this study. The investigation was made for selected periods in 1994 and 1995 disseminated through the GEWEX Continental-scale International Project (GCIP) as the GIST and the ESOP-95 data sets, respectively, and using preliminary data for 1996. The evaluation of Eta model-derived data fields was made through comparison with fields in observations in two different regions, namely in Oklahoma and Kansas and Southern Arizona. In this way, it was possible to make comparison for humid and semi-arid mid-continental climates. The available observational data for comparison included both meteorological variable and surface energy partition in the case of Oklahoma/Kansas, while in the case of Southern Arizona it restricted to meteorological variables. In the course of making the model changes in February 1996, coding errors were found in that interpolated meteorological variables between the lowest model level and the ground to provide at 2 m. As a part of this study, an attempt was made to devise a correction procedure which applied to the GIST and ESOP-95 data sets to recalculate 2 m estimates from variables present in those data sets. This attempt was largely successful, although there were systematic differences between the corrected and those provided by the Eta model in 1996. Further investigation of this is required. The most important discrepancy between modeled and observed fields is in surface radiation. Modeled radiation is significantly less than measured in the clear sky conditions in Southern Arizona, while it greater than measured in cloudy midday in Oklahoma/Kansas. These discrepancies complicated identification of other potential model weakness in this study. However, the 1996 modifications seem to have greatly improved the Eta model's ability to capture the variations in air temperature and specific humidity.