4 ms·

I've been wondering this since last time I saw your app, but I wasn't sure how to ask it: If I understand correctly, you use barometric pressure reported by pe

by xur17 14y ago

I've been wondering this since last time I saw your app, but I wasn't sure how to ask it: If I understand correctly, you use barometric pressure reported by peoples' phones to attempt to predict weather. This makes sense to me. How do you handle changes in barometric pressure due to changes in altitude. If I work in a tall office building, wouldn't the app see a change in barometric pressure when I go up the elevator? Is this small enough that it doesn't matter, or do you handle it in some way? Anyway, the app looks great - I think I'll install it again to play around with. Keep up the good work.

cryptoz 14y ago
Thanks for your comment! > How do you handle changes in barometric pressure due to changes in altitude. If I work in a tall office building, wouldn't the app see a change in barometric pressure when I go up the elevator? Is this small enough that it doesn't matter, or do you handle it in some way? These are excellent questions. The data is extremely noisy due to exactly the reasons you suggest. At near-sea-level, an 8 meter altitude change results in a 1mbar change. So the effect is very significant and causes noise. There are some simple methods to weed out the noise: some changes are obviously too fast to be meteorological in nature, so those can be discarded. Cumulonimbus itself is not working to solve this problem right now, but Cliff Mass and his lab are. We've livestreaming our data to them and they are currently working on quality control algorithms. They are starting by looking for phones that aren't moving, using the GPS and accuracy information. From there they will work on assimilation and then begin forecasting. Quality control and altitude correction is indeed important for our data to be properly used in meteorological systems, but currently we're seeing a simpler way to do quick analysis. The density of the network is getting pretty good in cities (though we're seriously lacking in more rural areas and we'll have to pick that up), which means that single data-point altitude changes are very obvious compared to the rest of the nearby points.
- xur17 14y ago
  Thanks for the fast response! It's pretty amazing to see the map, and see how many people are contributing data to this. In terms of the data that is submitted: 1) Is my location data submitted, and if so, how accurate is it? 2) Is this data anonymized in some way (are all contributions linked to a single person - because this would basically give you my location data)? 3) The public sharing option is a little bit confusing - does this share with PressureNET, researchers, and forecasters in addition to the public?
  - cryptoz 14y ago
    1) Yes, your location is submitted. We just started collecting location accuracy information with this release a couple of days ago. So I'm not sure yet exactly how accurate this is, but we'll be learning about that in the coming days and weeks. 2) For the map view in the app and for our public site, yes the data is anonymized. However, we do keep a unique ID in the database. This is required in order to calibrate the sensors, filter out noise and have good quality control. 3) I forgot to add better descriptions / Help screen to this update. Sorry about the confusion. The Public option is not in use yet, but will be soon when we finish our livestream API. The idea is that yes, the data will then be available to us at Cumulonimbus, to researchers and forecasters, and also to the public at large. For the Public API, we do plan to anonymize the data but we haven't decided exactly how yet (hence it's not available yet). Thanks for your questions!

by xur17 14y ago

cryptoz 14y ago

Thanks for your comment! > How do you handle changes in barometric pressure due to changes in altitude. If I work in a tall office building, wouldn't the app see a change in barometric pressure when I go up the elevator? Is this small enough that it doesn't matter, or do you handle it in some way? These are excellent questions. The data is extremely noisy due to exactly the reasons you suggest. At near-sea-level, an 8 meter altitude change results in a 1mbar change. So the effect is very significant and causes noise. There are some simple methods to weed out the noise: some changes are obviously too fast to be meteorological in nature, so those can be discarded. Cumulonimbus itself is not working to solve this problem right now, but Cliff Mass and his lab are. We've livestreaming our data to them and they are currently working on quality control algorithms. They are starting by looking for phones that aren't moving, using the GPS and accuracy information. From there they will work on assimilation and then begin forecasting. Quality control and altitude correction is indeed important for our data to be properly used in meteorological systems, but currently we're seeing a simpler way to do quick analysis. The density of the network is getting pretty good in cities (though we're seriously lacking in more rural areas and we'll have to pick that up), which means that single data-point altitude changes are very obvious compared to the rest of the nearby points.

xur17 14y ago
Thanks for the fast response! It's pretty amazing to see the map, and see how many people are contributing data to this. In terms of the data that is submitted: 1) Is my location data submitted, and if so, how accurate is it? 2) Is this data anonymized in some way (are all contributions linked to a single person - because this would basically give you my location data)? 3) The public sharing option is a little bit confusing - does this share with PressureNET, researchers, and forecasters in addition to the public?
- cryptoz 14y ago
  1) Yes, your location is submitted. We just started collecting location accuracy information with this release a couple of days ago. So I'm not sure yet exactly how accurate this is, but we'll be learning about that in the coming days and weeks. 2) For the map view in the app and for our public site, yes the data is anonymized. However, we do keep a unique ID in the database. This is required in order to calibrate the sensors, filter out noise and have good quality control. 3) I forgot to add better descriptions / Help screen to this update. Sorry about the confusion. The Public option is not in use yet, but will be soon when we finish our livestream API. The idea is that yes, the data will then be available to us at Cumulonimbus, to researchers and forecasters, and also to the public at large. For the Public API, we do plan to anonymize the data but we haven't decided exactly how yet (hence it's not available yet). Thanks for your questions!