85th AMS Annual Meeting

Tuesday, 11 January 2005: 12:00 PM
Trial by fire - developing, delivering, and defending experimental climate forecasts during three drought years in Colorado
Klaus Wolter, NOAA/ERL/CDC, Boulder, CO
New and improved climate divisions (see separate talk) can be used for the prediction of climate anomalies in the interior southwestern U.S. The author has been engaged in this task for several years. Originally, this came about through public outreach in conjunction with the 1997-98 El Niņo and subsequent La Niņa years. It was found that the ENSO "signal" was (1) seasonally variable in much of the domain of interest (in particular, Colorado), and (2) not well represented in traditional climate division data.

During the early part of the millenium, in particular the winter of 2001-2, the utility of ENSO-related climate information diminished along with vanishing La Niņa conditions. Long-term trends are quite pronounced and usable for temperature fields, but much less so for precipitation forecasts. At the same time, a moderate drought situation had been building since late 1999. This formed the background for a systematic exploration of sources of interannual climate predictability in this region, being fully aware that Climate Prediction Center forecasts had a poor track record in the interior U.S.

This presentation gives a brief survey of the employed forecast technique (stepwise multiple regressions in "ensemble" mode) as well as different predictors that were found useful. Experimental climate (pecipitation) forecasts were first issued in late 2001, and are updated monthly on the internet. In addition to ENSO-related sources of predictability, other predictors include tropical sea surface temperature and sea level pressure anomalies, northern hemispheric teleconnection patterns, antecedent precipitation anomalies, and more exotic influences such as the state of the quasi-biennial oscillation and sunspot cycle.

Forecast skill during the last three years is analyzed and compared against benchmark forecasts from CPC. It shows large spatial and temporal variation. Particularly poor performance during the spring and summer of 2002 may be related to the exceptional wildfire activity level of that period, as well as extreme antecedent soil moisture conditions. Much better performance in late 2002 and early 2003 appears to be linked to moderate El Niņo conditions.

This presentation will include a brief summary of my seven-year experiences dealing with both private media and public agencies from Denver Water to the National Interagency Fire Center in Boise. The lion's share of these interactions was with the Colorado Water Availability Task Force that has been meeting on an almost monthly basis during the recent drought years.

Supplementary URL: