Topic: Does WeatherUnderground Use Reported Data to Improve Predictions?

[LeoC4]

I've submitted data from my AcuRite 5-in-1 to WeatherUnderground for several years.  I realize that forecasting temperatures is a shot in the dark, but am continually amazed that the predicted low (during the cooler months) is generally 5 to 10 degrees F higher than I actually experience.

My weather station is not located in a pocket; it's in a wide open area of reasonably flat ground within a 3 mile radius.  I've checked the AcuRite sensor against a different thermometer; both are within a degree of each other.

Does WeatherUnderground use reported data to improve their predictions or am I just wasting time supplying them with my data?

[WSWeather]

It is a pretty good bet any data you supply is not incorporated into the forecasts on WU.  They aggregate PWS data and sell it to their corporate users for lots of cash every month, but any forecast you get from them is almost always for conditions at the nearest airport.  You might as well use NOAA...they're more accurate anyway.

[sky_watcher]

It is a pretty good bet any data you supply is not incorporated into the forecasts on WU.  They aggregate PWS data and sell it to their corporate users for lots of cash every month, but any forecast you get from them is almost always for conditions at the nearest airport.  You might as well use NOAA...they're more accurate anyway.

Who are these corporate customers who pay lots of cash for data that isn't being used by WU in their activities? What do they use it for?

[WSWeather]

A few examples:
https://www.informationweek.com/big-data/big-data-analytics/ibms-deep-thunder-shows-how-weather-is-big-business/d/d-id/1325946?
https://www.ibm.com/products/weather-targeting
https://doc.media.weather.com/products/v3.1/home.php

[sky_watcher]

A few examples:
https://www.informationweek.com/big-data/big-data-analytics/ibms-deep-thunder-shows-how-weather-is-big-business/d/d-id/1325946?
https://www.ibm.com/products/weather-targeting
https://doc.media.weather.com/products/v3.1/home.php

If I understand what you are saying:  IBM owns WU. IBM sells weather products. WU provides people a place to look at an IBM weather product but it does not use the data it gathers but IBM uses that data in its weather products.

Seems rather implausible to me as it would require two different forecasting processes, but the world is a strange place at times.

[ocala]

Do a comparison. See what forecasting service consistently provides the best forecast for your location. I have actually found WU to be slightly more accurate for low temps in my area but forecasting low temps is tricky. On a dead calm night with no wind radiational cooling takes place and you'll see lower readings. But with those same conditions a slight breeze can kick up for just a short time raising temps a few degrees before starting to fall again as the breeze subsides. What would have been a 20 degree night turns into a 25 degree night because of that little breeze. Also some unexpected cloud cover could move in also raising temps. There are a few variables that can change temps that just can't be forecasted. 

[WSWeather]

Seems rather implausible to me as it would require two different forecasting processes, but the world is a strange place at times.

I didn't say that WU didn't use IBM data...I said your PWS data is probably not incorporated into the forecast you see.

[sky_watcher]

Seems rather implausible to me as it would require two different forecasting processes, but the world is a strange place at times.

I didn't say that WU didn't use IBM data...I said your PWS data is probably not incorporated into the forecast you see.

I guess we are using "incorporate" differently, because to me any item used in a data-set used for a forecast is incorporated in that forecast. Looking at your comment, I assume you meant that the forecast is a broad-area forecast, not one specifically for the PWS locale.

My misunderstanding.