A NEW LOOK FOR FORECASTWATCH.COM

In order to support more desktop environments, and to provide a cleaner, easier to understand look-and-feel, ForecastWatch.com received a new look this weekend. Besides the cosmetic changes, there is a new News section, and the high and low temperature statistics are now on their own pages. We hope that the new look makes it easier for you to navigate the site, and to view and understand the weather forecast accuracy data we provide.

FORECASTWATCH.COM AUGUST DATA AUDITED AND LOADED

High and low temperature errors for August have been audited and loaded into the Drilldown and Compare products. The Year-To-Date numbers have also been updated. Much of the Midwest and Central U.S. continued to experience below normal temperatures in August, as they did in July, and that is reflected in the error statistics, which tend to show over- forecasting (forecasting a temperature higher than actual) in those areas.

You can see the trend visually, by noticing the large red areas on the

U.S. map, which indicate forecasts higher than actuals.

To view the August data, as well as year-to-date and historical numbers, log on to http://www.forecastwatch.com

SEPTEMBER RELEASE ADDS ADDITIONAL LOW FORECASTS

Besides the new look, the September release of ForecastWatch.com offers additional accuracy. Previously, on any given day, a high temperature error statistic needed to exist for a low temperature error statistic to be recorded. This is no longer the case, and this has resulted in 50,000 new low temperature scores between February and August. Not much when you consider there are over six million scores in the database, but it does improve the accuracy of the statistics slightly.

Additionally, the RMS and Overall RMS statistics are now properly populated. RMS was one of the first statistics calculated when ForecastWatch.com started two years ago. Due to error or habit, what had been referred to as RMS error is really MS error, or mean squared error. The only difference between RMS and MS error is a square root (RMS is the square root of MS). However, what was being displayed was MS error. That is now corrected, and RMS error is really RMS error now.

Let us know what you think at ideas@forecastwatch.com . We welcome your ideas and suggestions to make our services better and more useful.