To me, rainfall is a pleasing word. It just sounds relaxing.
“How much rain did we get?” seems more cold than “How much rainfall did we get?” But ChatGPT disagrees with me. It says the f in rainfall is “harsh”. Given that the letter f may have originated as a representation of a club, ChatGPT may have a point.
What do you think? Because I like to try shiny new toys, let’s try this Substack poll thingy…
We’re thinking about rainfall today because we ran across a new research article about tracking rainfall accumulation using cell phone signals. This novel application of consumer technology has the potential to help with the forecasting and monitoring of flash floods, where last-minute rainfall amounts can be critical information. In meteorology, this is sometimes called a nowcast - a forecast for about the next 0-3 hours.
Forecasters use various tools to measure rainfall accumulation. The most common are radar, satellites, and good old-fashioned rain gauges. However, none of them can monitor rain continuously with hyperlocal spatial data, which is the holy grail for flood forecasting - to know precisely how much rain has fallen/is falling on a very specific spot.
Calling Ma Bell
Cell phone towers constantly radiate energy all over the place all the time. When not causing zombie outbreaks, they allow us to call family and pass time while waiting for our microwaved burrito.
The same microwaves that cook those burritos are also used to transmit cellular signals— though at slightly different frequencies, thankfully.
Microwaves are quite energetic as electromagnetic waves go, and they tend to attenuate quickly, meaning they degrade in strength over distance. This gives them a relatively short range, which is why cell phone towers need to be much closer together than radio or TV towers. It also means that small objects, like raindrops, can interfere with their transmission. Longtime WWAT readers probably know where I’m going with this.
The idea of measuring rainfall using commercial microwave links (CML) dates back to the 1970s, but researchers h not truly explored its potential until the last 10–15 years. They needed accessible data from networks with good regional coverage, provided mainly by private companies.
There is considerable variation between cell phone towers in terms of the frequencies they use, their distance from each other, transmitting power, and more. Each of these variables affects the attenuation rate of their signal. To separate rainfall attenuation from these other causes, meteorologists need to properly calibrate these systems using rainfall data from other sources. Once properly calibrated, these systems can be used to estimate rainfall independently.
A research team led by Magdalena Pasierb at the Institute of Meteorology and Water Management - National Research Institute, Warsaw, Poland - recently published a study in the journal Meteorological Applications. They made CML-based rainfall estimates and compared them with measurements from manual rain gauges and estimates from commonly used radar and satellite-based techniques. The study examined 67 microwave links in southern Poland, ranging in length from less than 1 km to almost 20 km between the transmitter and receiver.
Detecting Rainfall
The first step was to identify rainfall in the CML data. They looked for signal loss between a transmitter and a receiver and compared it with loss observed on clear weather days, as well as loss measured by similar local transmission-receiver links, among other methods. A complication arose because each microwave link uses multiple frequencies, so they tested various ways to combine the varying attenuation recorded at each frequency.
Interestingly, they also had to account for rainfall on the antennas themselves. The amount of water collected on the antenna (and its contribution to attenuation) is influenced by both rainfall intensity and temperature.
Ground Truthing
Once they obtained an estimate from the CML, they needed to compare it with actual observations, a process known in meteorology as ground-truthing—checking assumptions against real-life data. Politicians tend to dislike this process.
They measured actual rainfall using 23 manual and 500 telemetric (remotely reporting) rain gauges. This data was combined with information from three Polish radars and one Czech radar to create rainfall estimates at thirty-minute intervals. Additionally, they used data from EUMETSAT Meteosat satellites to estimate rainfall using several established algorithms. They called this overall system RainGRS.
Each of these techniques has its strengths and weaknesses. Generally, radar is considered accurate for determining where it is raining, while rain gauges provide good estimates of how much rain has fallen. Satellite data serves as a backup when radar or rain gauge estimates are unavailable. Manual rain gauges are thought to be overall the most accurate1. So, they compared the CML-data to the RainGRS data and the manual rain gauges separately.
Results
The authors presented a comparison with one of the manual rain gauge sites in Opole, Poland. They found the CML method2 was quite consistent at detecting the beginning and end of rain events. But it was not quite so consistent at measuring the amount of rainfall. It underestimated rainfall amount in most events, but was relatively accurate in one of them (August 6-7).
They also created maps of all the rainfall measured over about a month in the summer of 2022. Here the CML data was relatively accurate. The areas where it missed rainfall (ex: the northern border) are areas where they were not monitoring any microwave links. A bigger dataset covering more towers would likely have detected that rainfall as well.
The authors believe that CML analysis shows promise but requires further development. Currently, it is not as accurate as radar-based estimates but is more accurate than satellite-based estimates. However, the errors are likely of a type that could be statistically corrected with more testing and larger datasets. For example, modeling the relationship between rain on the transmitters and the length of the microwave link could help eliminate that source of error. Additionally, CML analysis is less accurate for shorter links than longer ones, which limits its effectiveness in cities where towers are closer together, and the impact of rainfall rate on flooding is significant due to limited natural runoff.
Overall, this represents a promising advance in the development of a novel method for estimating rainfall that is independent of existing techniques. With further refinement, this system could become an important tool for meteorologists and hydrologists in forecasting flood events.
And Now for Something Completely Different
Last month I was at a workshop at the National Weather Service Training Center in Kansas City. An esteemed career meteorologist said that if he had to choose only one weather product to use in a forecast, it would be water vapor satellite imagery. Not radars, not surface maps, not visible cloud imagery. But water vapor. That surprised me because it’s not generally one of the sexiest products you see in weather forecasts. However, he showed some quite lovely water vapor maps, like this one:
Water vapor satellite imagery shows how much moisture is in the upper atmosphere (approximately 15,000 - 30,000 ft). The highest humidities are in the lighter regions, while dry regions are darker. This ability to track moisture is fundamental to many forecasting techniques.
There are many meteorological topics I’m just itching to write about. But as this is a newsletter that covers research, I must await a research article that references said topics. Water vapor imagery has been added to the list. Expect a deep dive, hopefully soon!
We acknowledge Dr. Milind Sharma for additional contributions.
ICYMI: If you love alternative uses of technology, check out our article about using weather radars to measure bird migration.
Nicely done!
Aaron, what is really important is whether those local area towers can detect density of waves of zombie attackers! That way I know whether to grab an umbrella or a bat before leaving my survival bunker!
IVT is the gold standard (I think) for measuring the precipitable strength of atmospheric rivers (quite common in my part of the country). I've long advocated for a 'Cat X' system for classifying those rivers and IVT is the closest thing we've got.