It has been a few days since last report, and today I'm very please to annouce that the graph-generating part of the new system is nearly complete, and you can see a result below:


Please click to see the sample with its original size. Three keypoints have to be note:

  1. Some argue that the forecast time is too short (compare to the 180 hours used now). Over the past two years, we have conduct two monitoring programs about the accuracy vs. forecast hours. The result shows that the accuracy for forecast past 72 hours has draw down significantly (for many test sites, but few inland sites -- that's because the atmospheric enviroment of inland sites are normally quite steable, so it's possible to make long-time forecast with relatively high accuracy), so I decide to discard the forecast past 72 hours for Panel 3 (now it got a new name -- APanel). For those who prefer long-time forecasts, Weatherchart or Weatherman (two other products of 7Timer!) is other choices, but the development of Weatherman is not included in this developing circle.
  2. For some reasons, the "darkness" function does _not_ consider that artificial light pollution, different from what I had note. I'm sorry about this...
  3. The sample module works fine, but only one problem that I have nothing can do about it -- as the "graph" is generated in realtime, it may take several seconds to complete the creating, typically 8-15 seconds if the users aren't many. Not 40+ seconds as shows above -- I was testing how slow is it under heavy connections! But actually I've no idea how will it performs when it's really put into use... let's see.

We still to wait for some time until the new version can be put into test -- there are still many things yet to be done. Stay tuned for updates.