Citizen Science: the mPING Project

If you are not a meteorologist, you may wonder what on earth this blog is about! mPING?

As we have discussed numerous times on this blog, meteorologists, just like other scientists, rely heavily on observations in order to do their job. It is observations of current weather that are input into the complex numerical models that output forecasts of future weather conditions days and weeks ahead. Additionally, observations serve an important role in verification of forecasts and warnings to know what is happening on the ground in real time. Lastly, we as forensic meteorologists use weather observations to reconstruct past weather events. Since it is impossible to have eyes everywhere, even in this day of satellite, radar, and automated surface observations, meteorologists continue to rely on citizens to submit weather observations. We highlighted many of these in a past blog post, but for today I would like to focus on mPING because recent events have brought this network into the spotlight.

Courtesy: NOAA/National Severe Storms Laboratory.

mPING, which is short for Meteorological Phenomena Identification Near the Ground, was designed to allow users of mobile phones to submit weather observations into a database at the National Severe Storms Laboratory (NSSL). The idea of leveraging the huge volume of cell phones to collect weather observations over widespread areas was relatively new back in 2012 when mPING was initially deployed. The motivation for the project stemmed from the nationwide upgrade of the radar network beginning in 2011 which brought dual-polarization technology and the capacity for the radar to determine precipitation types, a key component of winter weather forecasting. However, the radar beam is high above the surface, and so it is necessary to collect ‘ground truth’, or real-time surface-based observations, in order to determine if the radar algorithms are functioning correctly and represent what is actually going on at the surface.

One benefit of mPING is that the general public has both cell phones and enough general knowledge to determine most types of precipitation (snow, ice pellets, freezing rain, and rain, for example). The other ‘plus’ of mPING is that users can submit their reports anonymously, and even the developers and users of the database do not track the source of the observations. There is no registration required to use the app. This brings up a number of scientific questions, including how the data is quality-controlled. Generally speaking, crowdsourcing consistency among reports serves as an immediate flag of ‘suspect’ observations. With a very high volume of reports, it becomes relatively easy to identify those which don’t seem to ‘fit’ with other nearby observations.

Given the targeted users and goals of the app, its design was kept very simple and easy to use. There is some description of various precipitation types to assist users in correctly classifying their report off the menu. In order to keep the interface as simple and fast as possible (due to the high volume of reports), there are not photographs or comments permitted with reports. When viewed as a time-loop on a map, it is very easy to identify where precipitation is occurring and what type. The app was eventually expanded to allow for reports of other types of weather including floods, tornadoes, landslides, fog and dust storms.

As with any dataset, there are ‘pluses’ and ‘minuses’. One issue with mPING is the lack of training and data quality control, but the high volume of reports can be used to mitigate that to some degree, and the data source has been shown to be useful in a variety of forecast scenarios, including severe weather and winter weather. Population density is also an issue, as it is with many types of volunteer surface observations- reports are weighted most heavily toward where most people live (cities), and there still remains a paucity of reports in rural and outlying areas. Even with these drawbacks, the data has been very beneficial for many applications, including road maintenance, aviation operations, and of course public warnings and forecasts. In 2016, the capacity of the app was expanded to allow for reports to be submitted around the world.

On July 7, 2020, one of the vulnerabilities of this type of crowdsourced weather data became apparent, as mPING was inundated with large numbers of false weather reports. The Washington Post reported that the perpetrators committed ‘hateful acts’ in the process of this action. False reports were identified in New Mexico, Texas, Alabama, and even as far away as Hiroshima and Nagasaki (Japan), Tienanmen Square (China), and Auschwitz (Poland). The perpetrators were allegedly able to ‘spoof’ GPS data (manually override the device-identified location) to submit these false reports. As a result, mPING has been temporarily suspended and no users are allowed to submit these valuable reports.

Fortunately, this breach did not result in any compromise of user data, since no user data is collected in the app. However, it created a huge volume of additional work for federal workers, who had to identify and remove the erroneous reports so that the integrity of the database remained uncompromised. After the fix was complete, mPing went back online in late August and is now able to receive reports again.

Leave a Reply

Your email address will not be published.