Discount Outlier Data and Questionable Personal Weather Stations
I just went to the main page to see the current conditions and forecast for my local area (Durham NC, 27713). As you can see attached--the header of the page is reporting data based on a temperature reading of 101.7F when none of the other nearby stations is registering anything above 92.3. It seems obvious that something is wrong at that site.
When choosing which reading to use for the headline forecast, your system should make an effort to look at surrounding stations and should not use a value that is out of line with other nearby reports.
In the long run, you should track how often each personal weather station provides results that are out of line with other stations in the area and should discount results from stations that are consistently incorrect.
Further, you should provide an easy way for a user to (a) block a particular personal weather station from ever showing up in a report and (b) exclude all personal weather stations from the report.
I find it annoying that the default station is some obscure and at times inaccurate. I look for out of place dew points. No way its correct that you have 64dp amid a city with all others are 75dp. How about the airport as default? And if want the local floral club let me change it.
same here, and lately, the weather history shows for a location, while closer physically, shows wildly different info than the previously matched site.(current station is several thousand feet higher in altitude on a nearby mountaintop Sunlight Mountain) and shows temps 15~20 degrees variance from actual local conditions, you used to pick a further site that was closer climatically, Rifle Airport I am in Glenwood Springs CO
I agree. There should also be a way to report outliers. There is an outlier in my area that often throws off the current conditions and forecast analysis.