Home » September 2010 Southwest Climate Outlook » Temperature Verification
Forecast Verification
Published September 23, 2010

Temperature Verification

 (October 2010–March 2011)
Data Source(s): Forecast Evaluation Tool

For a thorough description of the interpretation of these maps, see the feature article, “Evaluating forecasts with the RPSS,” in the April 2009 issue of the Southwest Climate Outlook.

Comparisons of observed temperatures for October–December to forecasts issued in September for the same period suggest that forecasts in all but the southwest region of Arizona and northern New Mexico have been more accurate than an equal chances forecast; in these two regions forecasts  have been similar to an equal chances forecast (Figure 15a). Forecast skill—a measure of the accuracy of the forecast—is highest in southeast Arizona and southwest New Mexico for the October–December and November–January periods (Figures 15a–b). Forecast skill for the three- month lead time diminishes and few regions have been more accurate than simply using an equal chances forecast (Figure 15c). Forecast skill for the January–March period have highest in southern Arizona and New Mexico (Figure 15d). While bluish hues denote more accurate forecasts, caution is advised to users of the seasonal forecasts for regions with reddish colors.

Temperature Verification
Notes:

These maps evaluate the historical performance of the one- to four-month long-lead forecasts made by NOAA’s Climate Prediction Center (CPC). The maps convey the historical accuracy of the CPC forecasts in relation to the reference forecast, which assigns a 33 percent chance to the three CPC categories, “above,” “below,” and “neutral.”  These categories indicate whether conditions are predicted to be similar to the warmest, coolest, or normal temperatures for 1971 to 2000. The maps are generated from the Forecast Evaluation Tool, which was developed by The University of Arizona in partnership with NOAA, NASA, NSF, and the University of California-Irvine.

The maps display the Ranked Probability Skill Score (RPSS). The more the forecasts and actual weather match, the bluer the color. A bluish or reddish RPSS indicates the forecast is more accurate or less accurate, respectively, than assigning a 33 percent chance to each of the three CPC categories.

The RPSS is calculated by comparing all the forecasts made since December 1994 for particular seasons and specified lead times to the actual weather of the season.

Related Links: