Thursday, 2 July 2015: 10:45 AM
Salon A-5 (Hilton Chicago)
Climatological dynamic downscaling was examined through high-resolution (1-km grid spacing) simulations conducted using the Weather Research and Forecasting (WRF) model, with initial and boundary conditions from the Global Forecast Model (GFS) reanalysis (resolution 1°) historical simulations. The analysis was conducted over the eastern Mediterranean, with a focus on the country of Lebanon, which despite its small area of 10,452 km2 is characterized by a challenging complex topography that magnifies the effect of orographic precipitation. For example, a 50 km west-east cross section shows stark climate variations: a subtropical coastal climate followed by a typically Mediterranean climate at low elevations and a cold weather at higher elevations covered with snow during the winter, reaching a semi-desert plain, too dry to allow agriculture. This variety means a great diversity in ecosystems and landscapes in a limited surface area, which must be captured by the dynamic downscaling model in assessing future climate change impacts. Two simulation years were selected to capture the natural variability of the system, with 2003 and 2010 as typical wet and dry years, respectively. The WRF simulations were conducted using three one way nested spatial domains with 9, 3 and 1 km resolution. The model results were compared against observation data network of surface variables in complex terrain of weather stations located in eight geo-climatic regions with particular emphasis on the impact of model setup and resolution. Assessment of the quality, comprehensiveness and span of several climatic data sources was undertaken to identify the spatial and temporal data that can be used. While climatic records for various weather stations exhibited variations in span and quality, data from 43 rain gauges, 6 wind and 31 temperature stations were used to assess the ability of WRF to simulate the weather, and over long runs the climate, in the study area. In-depth comparisons between simulated and observed precipitation, 2 m temperatures and wind are performed. These evaluations focused on daily average, maximum and minimum temperatures, precipitation and wind with averages over seasonally and yearly time scales and on corresponding simulated variables from the three different resolution domains. Several statistics are used to evaluate model performance using observational datasets such as the mean bias error (MBE), the root mean square error (RMSE), the coefficient of determination R2, the high ranking percentiles, and an assessment of a set of climate indices (wet days, length of wet period, rainfall intensity, etc.) The downscaled dataset showed large improvements over the coarse GFS reanalysis data. The higher resolution simulations produced a significantly improved representation of the temperature and precipitation fields, at all timescales, particularly for extremes (Figure 1). The wind comparison exhibited a slight improvement in bias and in RMSE. Results compare well with recent studies with other models and/or for other regions, further supporting the use of WRF as tool for regional climate downscaling. This adopted higher resolution allowed a better representation of extreme events that are of major importance to develop mitigation/adaptation strategies by policy makers and downstream users of regional climate models. Figure 1 Regional and seasonal distribution RMSE (°C) for daily maximum temperatures ACKNOWLEDGMENTS The authors thank the United States Agency for International Development for providing support for this work through the USAID-NSF PEER initiative.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner