Statistics of Climate and Climate Impacts
Climate Scenario Generation/Statistical Downscaling
Richard Katz, in collaboration with Marc Parlange (Johns Hopkins U) and Claudia Tebaldi (GSP), continued work on the development of stochastic weather generators for locations in Southeast US, conditional on indices of large-scale atmospheric circulation (e.g., the Bermuda High). Such models will be used as part of a statistical downscaling approach to producing climate change scenarios for use in assessments of the economic impacts on the agricultural sector. This work forms part of the Southeast Agricultural Project described under the section on Integrating Climate Models and Impacts.
Back to Top
Stochastic Modeling of Precipitation
Katz, in collaboration with Xiaogu Zheng (National Institute of Water and Atmospheric Research, New Zealand) completed work on the development of a mixture model for daily precipitation amounts to account for the "overdispersion" of monthly/seasonal precipitation totals. It was established that the "hidden" index identified by this approach is directly related to observed measures of large-scale atmospheric circulation patterns. This research has important implications concerning how to improve the properties of stochastic weather generators used, for example, to produce scenarios of climate variability and change.
Back to Top
Methods for Assessing Economic Value of Weather and Climate Forecasts
Katz reviewed Bayesian decision-theoretic studies of the economic value of imperfect weather and climate forecasts. This review focuses on work of the late Allan Murphy (Oregon State U) and includes treatment of some controversies over the most appropriate approach to valuing forecasts. In a related activity, a web site that categorizes recent case studies of the value of weather and climate forecasts was developed.
Back to Top
Extreme Value Theory and Climate Change
Katz reviewed the use of the statistical theory of extremes in research on climate change and its impacts. Among other things, reasons for the inadequate treatment of extremes in the Intergovernmental Panel on Climate Change (IPCC) 1995 report are being identified. In a related activity, Katz completed work on the application of extreme value theory to examine the sensitivity of extreme high precipitation amounts to changes in the parameters of a stochastic model for daily precipitation amounts. In the example considered, for a physically plausible range of changes in these parameters, the effective return period is most sensitive to the scale parameter of the intensity distribution.
Back to Top
Retrospective of 1997-98 El Niņo for WMO/UN General Assembly
Glantz, in collaboration with Anthony Barnston and Yuxiang He (both of NOAA/Climate Prediction Center), prepared a manuscript entitled "Predictive Skill of Statistical and Dynamical Climate Models in Forecasts of SST during the 1997-98 El Niņo/La Niņa Episodes." This manuscript was prepared at the request of the World Meteorological Organization (WMO), in WMO's response to a Resolution of the UN General Assembly in November 1997. The primary goal of the manuscript (and the contribution to the UN retrospective assessment of the 1997-98 ENSO event) was to compare 15 dynamical and statistical models for their 1997-98 ENSO forecasts. The paper also details the improvements made over the last 15 years in the ability of scientists to forecast ENSO warm and cold events at longer lead times. Foremost among these improvements are increasingly improved data observing and assimilation systems, higher computer speed and storage capacity, and increased understanding of the tropical oceanic and atmospheric physics underlying the evolution of the ENSO phenomenon. Another objective of this study was to get an idea of the general skill of the models in a collective sense and to revisit the question of whether the numerical models have begun to outperform the statistical models; it was not an attempt to compare the relative accuracy of one model against another, but to evaluate the state of ENSO forecasting. Various qualitative aspects of the 15 model forecasts of the 1997-98 ENSO event were also assessed. See also CGD ASR for Kevin Trenberth's contribution to the UN Retrospective. The manuscript has been reviewed and accepted for publication in BAMS for February 1999. The results of their assessment were presented to an International ENSO Forum in Guayaquil, Ecuador, in November 1998.
Back to Top