Evaluation of Nocturnal Temperature Forecasts Provided by the Weather Research and Forecast Model for Different Stability Regimes and Terrain Characteristics

详细信息    查看全文

• 作者：Adriano Battisti ; Otávio C. Acevedo ; Felipe D. Costa…
• 关键词：Nocturnal temperatures ; Stable boundary layer ; Turbulence formulations ; Weather Research and Forecasting model
• 刊名：Boundary-Layer Meteorology
• 出版年：2017
• 出版时间：March 2017
• 年：2017
• 卷：162
• 期：3
• 页码：523-546
• 全文大小：
• 刊物类别：Earth and Environmental Science
• 刊物主题：Atmospheric Sciences; Meteorology; Atmospheric Protection/Air Quality Control/Air Pollution;
• 出版者：Springer Netherlands
• ISSN：1573-1472
• 卷排序：162

文摘

The quality of nocturnal temperature forecasts made by the Weather Research and Forecast (WRF) numerical model is evaluated. The model was run for all July 2012 nights, and temperature fields compared to hourly observations made at 26 weather stations in southern Brazil. Four different planetary boundary-layer (PBL) schemes are considered: Bougeault–Lacarrere (BouLac), Quasi-Normal Scale Elimination (QNSE), Yonsei University (YSU) and Mellor–Yamada–Janjic (MYJ). Additional simulations to assess the role of higher horizontal and vertical resolutions were performed using the MYJ scheme. All schemes, except BouLac, underestimated the 2-m temperature, and in all cases the temperature bias is dependent on wind speed. At high wind speeds, all schemes exhibit a cold bias, which is greater for those that yield lower nocturnal surface-layer turbulent intensity. The elevation difference between each station and the model nearest grid point \(H_{\textit{station}} -H_{\textit{gridpoint}} \) is highly correlated with the temperature simulation error. We found that a consistent cold bias is restricted to conditions with low-level clouds. We concluded that one possible means of improving nocturnal temperature forecast is to use parametrization schemes that allow for higher turbulent intensity in near-neutral conditions. The results indicate that this improvement would partially counteract the misrepresentation of the low-level cloud cover. In most stable cases, a post-processing algorithm based on terrain characteristics may improve the forecasts.