Estimating Recent Climatic Change via Historical Air-Temperature Frequency Distributions

年和月平均气温是用来估计近期气候变化的标准参数。传统的对平均气温的关注忽视了可以从每日气温历史数据中获得的额外概率信息。最近的气候变化可以通过中心趋势、方差和其他形状度量的各种变化在频率分布中表现出来。该项目将开发一个全面的程序，用于同时检测气温频率分布的所有部分的趋势。该项目将使用48个毗邻州1,062个站点的空间广泛的每日气温数据，以开发使用时变百分位数和其他参数，以创新的方式估计最近的气候变化。将制定灵活的程序，估计移动窗口内的百分位数和参数，并将评估这些程序的稳健性。在温度数据中检测到的变化将与湿度和云量的同时变化以及气温日循环的形状变化有关。监测这些协同作用将提供气候变化的视角，这是对中心趋势的传统关注的实质性延伸。该项目将开发方法来检测当人们只检查平均气温时不明显的趋势。例如，气温的趋势可能显示平均气温上升，而频率分布的下部显示显著变暖，而上部则没有变化。这些变化不仅表明气候变暖，而且还表明极端温暖的空气温度变化不大，变异性减少。作为一般方法，这项研究将开发新的方法来检测频率分布的变化，这些方法可以应用于以相对较高的时间分辨率观察到的任何环境变量。考虑到生态、经济和社会系统往往对变异性和极端环境事件的变化更加敏感，本研究具有广泛的适用性。所确定的变化将对评估气候变化的经济、社会和生态层面具有价值。特别是对时变百分位数的考察，为环境变化研究提供了一种灵活而稳健的方法。

Annual and monthly mean air temperatures are the standard parameters used to estimate recent climatic change. The traditional focus on mean air temperature has ignored the additional probabilistic information that can be derived from historical data on daily temperatures. Recent climatic change can be manifest in frequency distributions through a variety of changes to central tendency, variance, and other shape measures. This project will develop a comprehensive procedure for the simultaneous detection of trends in all portions of air-temperature frequency distributions. The project will use spatially extensive daily air-temperature data for 1,062 stations over the 48 contiguous states to develop the use of time-varying percentiles and other parameters that can estimate recent climatic change in innovative ways. Flexible procedures that estimate percentiles and parameters within moving windows will be developed, and the robustness of these procedures will be evaluated. Changes detected in the temperature data will be related to simultaneous changes in humidity and cloud cover as well as change in the shape of the diurnal cycle of air temperature. Monitoring these synergistic interactions will provide a climate-change perspective that is a substantial extension of the traditional focus on central tendency.This project will develop methods to detect trends that are not apparent when one examines only mean temperatures. For example, trends in air temperature may show an increase in the mean while the lower portion of the frequency distribution shows substantial warming and the upper portion shows no change. These changes would indicate not only a warming climate but also one where there is reduced variability with little change in extremely warm air temperatures. As a general approach, this research will develop new methods for detecting changes in frequency distributions that can be applied to any environmental variable that is observed at relatively high temporal resolution. Considering that ecological, economic, and social systems tend to be more sensitive to changes in variability and extreme environmental events, this research has wide applicability. The changes identified will be of value for evaluating the economic, social, ecological dimensions of climate change. The examination of time-varying percentiles, in particular, provides a flexible and robust approach for environmental change research.