Improved Atmospheric Modelling of the Oasis-Desert System in Central Asia Using WRF with Actual Satellite Products

Miao Zhang, Geping Luo, Philippe De Maeyer, Peng Cai, Alishir Kurban
2017 Remote Sensing  
Because of the use of outdated terrestrial datasets, regional climate models (RCMs) have a limited ability to accurately simulate weather and climate conditions over heterogeneous oasis-desert systems, especially near large mountains. Using actual terrestrial datasets from satellite products for RCMs is the only possible solution to the limitation; however, it is impractical for long-period simulations due to the limited satellite products available before 2000 and the extremely time-and
more » ... onsuming processes involved. In this study, we used the Weather Research and Forecasting (WRF) model with observed estimates of land use (LU), albedo, Leaf Area Index (LAI), and green Vegetation Fraction (VF) datasets from satellite products to examine which terrestrial datasets have a great impact on simulating water and heat conditions over heterogeneous oasis-desert systems in the northern Tianshan Mountains. Five simulations were conducted for 1-31 July in both 2010 and 2012. The decrease in the root mean squared error and increase in the coefficient of determination for the 2 m temperature (T2), humidity (RH), latent heat flux (LE), and wind speed (WS) suggest that these datasets improve the performance of WRF in both years; in particular, oasis effects are more realistically simulated. Using actual satellite-derived fractional vegetation coverage data has a much greater effect on the simulation of T2, RH, and LE than the other parameters, resulting in mean error correction values of 62%, 87%, and 92%, respectively. LU data is the primary parameter because it strongly influences other secondary land surface parameters, such as LAI and albedo. We conclude that actual LU and VF data should be used in the WRF for both weather and climate simulations. elevations can increase dramatically, from a few hundred metres above sea level in the basin areas to over 5000 m above sea level in the mountainous areas, over a horizontal distance of less than 200 km; thus, there is high heterogeneity in land cover types [3] . Water is scarce and is valuable for both human livelihoods and ecosystems in CA [4]; water resources are largely derived from the mountainous areas, whose rivers are fed by hydrologic processes of snow and glacial melt and precipitation [5] . These rivers flow into artificial lakes and then disappear into the desert areas in the basin [6] . Given the limited amount of runoff [7] and unrestricted groundwater exploitation in the area [8] , oases form at the foothills of large mountains [6, [9] [10] [11] [12] . The geographical and ecological characteristics differ significantly between these oases and the surrounding deserts, causing significant differences in energy budgets, the exchange rate of momentum, and water vapor levels. These differences produce typical oasis effects [13] such as the "cold-wet" island effects of oases (an oasis is a wet, cold island capped by warm-dry air), and the thermal differences between oases and the surrounding deserts result in oasis breeze circulation (OBC). Such oasis effects increase in complexity both in and near mountain ranges. Although oases account for only a small proportion of the land surface (e.g., a proportion of 4-5% in Xinjiang, a typical region of the hinterland of the CA), more than 90% of the population and 95% of the socioeconomic wealth are concentrated there [14] . Therefore, CA, because of its large elevation differences and the importance of oases, can be divided into mountainous region, oases, and desert areas, often named the Mountain-Oasis-Desert System (MODS). The northern Tianshan Mountains (NTM), the core section of the Silk Road, is a typical geomorphological part of CA; it is also sensitive to climate change [15] . Recent studies have indicated that annual mean air temperature in the NTM has been increasing at an average rate of 0.8 • C decade −1 [16], which is greater than the average rate in CA (0.39 • C decade −1 from 1979 to 2011) and the global land surface (0.27-0.31 • C decade −1 from 1979 to 2005) [1]. Precipitation and the frequency of extreme precipitation show a rate of 11.3% in the NTM [16] amid a longer-term drying trend [17, 18] . Other areas in CA generally show a slight decrease in average annual precipitation [4, 19] . Additionally, the region has been experiencing distinct intense oases expansion since the 1950s [20] [21] [22] . Oases have expanded more than 400% in the past 60 years (from 121.0 × 10 4 ha in 1949 to 512.5 × 10 4 ha in 2010). A series of ecological problems have appeared as a result, including soil salinization, oasis degradation, and desertification [23] [24] [25] [26] . Horton [27-29] found that regional climate change was largely independent or potentially related to land cover change processes. The abnormal regional temperature and precipitation changes in the NTM may be due to the rapid oases expansion. Therefore, understanding the mechanisms of oasis effects and quantitatively investigating the climate effects of oases expansion on the regional climate are important for ensuring the sustainable development and ecological stability of oases, and will also provide useful information for regional climate change assessments [30] . Numerical simulation using regional climate models (RCMs) is the most effective method to explore both oasis effects and climatic effects of the oases expansion in the complex mountain-basin systems of CA because RCMs can account for climatic mechanisms not included in field measurements [31, 32] and Global Circulation Models (GCMs) [33, 34] . GCMs are unable to adequately resolve many important meso-microscale processes, like wind patterns and precipitation due to orographic effects based on large-scale convective parameterization schemes, and simpler land surface processes [35] . The Weather Research and Forecasting model (WRF) is an RCM that has been widely used to simulate regional climatic patterns, particularly over the past 10 years [36, 37] . Because the default terrestrial datasets in RCMs are generally derived from Advanced Very High Resolution Radiometer (AVHRR) data from 1992-1993, the ability of RCMs to accurately simulate weather and climate conditions is limited by the use of these outdated terrestrial datasets [38, 39] . Integrating actual terrestrial datasets from satellite products or observation in the model simulations is a novel way to overcome these limitations. Many numerical simulations have used MODerate resolution Imaging Spectroradiometer (MODIS) products, including land use (LU), albedo, leaf area index (LAI), and green vegetation fraction (VF), to improve the boundary layer meteorology simulation and to
doi:10.3390/rs9121273 fatcat:yj7o2dcxovbn5pgru2oxat53qm