I hope your summer is progressing well, and have enjoyed or will enjoy some deserved down time. We recently returned from a trip to the Smokey Mountains, where we witnessed a very powerful thunderstorm that took down a few trees. Now we are in a hot spell, and had some thunderstorms go by last night. One had just passed right at sunset, and was beautiful. I captured a panorama: http://twitpic.com/5sc8bw/full The only enhancement I made to the photo was to stretch the color saturation to bring out the colors more, but they were all there in the sky to see.

The June aggregations have completed, as well as the first-half 2011 aggregations (which can be viewed by selecting "All" for the month). I've made some enhancements to the audit checks for the observations that are derived from the hourly data, as I noticed there were a few classes of rare events that weren't getting caught. About 3% of observations are getting flagged as invalid, up from about 2%.

I also went through via the Trends product to look for anomalies in historical aggregation that weren't caught previously. It looked pretty clean with the exception of a issue with calculating BSS between May 2010 and July 2010, and BSS in April 2008 for Canada. These have been recalculated and are now correct. I wanted to conduct this check prior to restarting the historical precipitation reaggregation project and ensure that there are no remaining issues in the data.

Work continues on a few major projects. Collection started on two new providers on July 1, so data will be available starting next month. WeatherBug is the first, and was selected due to "popular appeal" and potential rather than forecast innovations (they use the NDFD, like WeatherUnderground). But, like WeatherUnderground, they are starting to use their observation network data and may provide custom forecasts in the future. The second is Foreca, which is starting to slowly make a name for itself, and generates forecasts worldwide.

One other small change was made to the unskilled Persistence forecast. Previously, an observation of "thunderstorms in the area" was considered a precipitation forecast, even if no precipitation was reported at the station. This has been changed, and now a precipitation forecast for Persistence is only issued if precipitation was reported at the observation station.

Finally, speaking of Persistence forecasts, I added "weather changeability" data to ForecastAdvisor, see for example http://forecastadvisor.com/change/high/ . Data also appears on each city's page as well. You can see pretty readily which areas of the country are more changeable than others. Us as weather geeks already know this, but beyond the "If you don't like the weather, wait a few hours" quips, most people don't have a quantitative understanding of how much the weather does change each day. Hopefully this gives them another discussion point.