区域气候变化统计降尺度研究进展

统计降尺度方法(the Statistical Downscaling Methods, SDM)是为合理预测区域尺度的气候变化情景而提出的新型研究方法。统计降尺度法利用多年大气环流的观测资料建立大尺度气候要素和区域气候要素之间的统计关系,并用独立的观测资料检验这种关系的合理性。把这种关系应用于大气环流模式(Global atmospheric general circulation models, GCMs)中输出大尺度气候信息,来预估区域未来的气候变化情景(如气温和降水)。同时,10a来降尺度方法在生态过程模拟以及气候变化与生态预报关系拟合研究方面也取得一定进展。对统计降尺度方法概念的内涵和外延、基本原理和操作步骤的创新研究方面进行了综述,归纳了该方法在模拟区域气候变化中的应用进展、研究热点及发展趋势,介绍了降尺度在生态预报中的相关应用,为相关研究提供参考。

Recent advances on regional climate change by statistical downscaling methods

The statistical downscaling methods (SDMs) are reasonable forecast tools which were recently proposed for climate change scenarios at a regional scale. The SDM are used to establish large-scale climate factors, the regional statistical relationship between the climate factors using years of observations of atmospheric circulation, and to test the suitability of this relationship using independent observations. Furthermore, the relationship is applied to atmospheric general circulation models (Global atmospheric general circulation models, GCMs) to produce large-scale climate information and predict future regional climate change scenarios (such as temperature and precipitation). In recent years, a variety of sophisticated statistical downscaling methods have been developed to meet the needs of domestic and international climate prediction programs and to provide effective support for regional ecosystem management. However, the choice of different statistical methods in a particular region must be highly targeted and effective. It is widely accepted that reasonable and consistent meteorological input data is a critical factor for modeling regional climate change at different levels. Previous studies showed that the GCMs provide the means of estimating climate change in the future by providing a time series of climatic variables. However, there are two main problems if GCM outputs are directly used for regional impact studies: 1) they are biased with respect to observations of present climate, and 2) the spatial scale is frequently too coarse. Therefore, dynamic downscaling methods employ regional climate models (RCMs), using the output of GCMs as forcing and boundary conditions. The RCMs provide sufficient information for ecological and hydrological modeling of the impact of expected climate change at different levels. Over the last decade, quite a few studies on the SDMs have been conducted and applied in ecological prediction. The recently proposed theory "coupling human and natural systems" provides us with the possibility of incorporating the SDM in ecological prediction. In this paper, we designed a flow chart to illustrate the method of ecological forecasting. Due to the complexity of ecological processes, we collected a series of factors including social and natural factors which influence the changes in ecosystem and regional climate. The model proposed here will undoubtedly validite and emphasize the importance of the accuracy of the results on ecological forecasting. In the past 10 years, some scholars have developed new techniques or methods to forecast ecological phenomena, such as biological invasion, agro-ecological safety, forest carbon sinks, biological diversity change, the ecological carrying capacity and other forecasts. Significant breakthroughs have been achieved in the SDMs of the interactions between climate change and ecological forecasts. Recent advances on connotation and extension of the concept, basic principles and operating procedures of SDMs were summarized. This paper summarizes the applicable progress, hot issues and developmental trends of the SDMs in simulating regional climate change. The review provides some initial approaches for the construction of meteorological and ecological prediction models and serves as a reference for related research.