Climate Analysis and Extended Range Seasonal Prediction

点击翻译按钮获取中文摘要

Abstract ATM-9417563 Gray, William M. Colorado State University Title: Climate Analysis and Extended Range Seasonal Prediction This project involves the analysis and further development of basic scientific concepts related to extended range quantitative climatic forecast schemes for Atlantic basin seasonal hurricane activity, African Sahel seasonal rainfall, ENSO variations, selective regional Asian monsoon rainfall, and the potential for seasonal predictability in the middle latitudes. Along with the basic science concerning underlying mechanisms, this work seeks accuracy at 9-14 month forecast lead times, sufficient to allow the development and implementation of meaningful policy and mitigating actions. The spectrum of potential climate precursor signals will be exhaustively explored. Potential predictability will be studied in the hindcast mode, principally from global data sets for the last 35 to 45 years (primarily since 1950), but with some extensions to much earlier data. Similarly, statistical regression methodology for combining predictor signals using both Least Absolute Deviations and Ordinary Least Squares procedures will be extensively explored. Although surprisingly skillful seasonal hindcast test results have been obtained, these results suggest that greater forecast skill is possible through further analyses with additional combinations of climate variables. Equal emphasis in this research is directed toward diagnostic work for greater physical understanding of the underlying atmospheric and oceanic processes responsible for long range precursor signals. This research is important because it seeks to improve our ability to predict climate variations of major economic and human safety importance.