| 24 Jun 2025
Temperatures of impervious surfaces in rural Montana
Abstract. Data described here demonstrate utility of bicycles as platforms for and carriers of modern small capable sensors. For two years (September 2021 to October 2023), I rode a bicycle carrying sensors that simultaneously measured GPS time and position, air temperature(s), surface temperature, downwelling visible and UV light, spectrally-resolved upwelling (reflected) light, plus air flow. I rode more than 170 times, covering a standard 15 km rural loop or along ~12 km paths to and from Bozeman, over a range of times and weather. To accommodate frequent snow and ice conditions, I walked the same bike carrying the same sensors more than 30 times back and forth along a quiet stretch of paved (mostly) snow-covered surface. Because loop and to/from Bozeman routes ran along an identical 3 km stretch of rural highway, that stretch represents one of the most-measured extents of impervious surface. Routes covered impervious paved surfaces punctuated by intervals of gravel or tree-shade or both. Sensors, adopted from consumer applications, produced reliable repeatable data. I achieved spatial resolutions of 4 to 5 meters and temperature resolutions of 0.5 °C; a typical ride of 45 minutes produced ~4000 clean data records. These data serve a wide variety of engineers and scientists exploring pavement temperatures, heat islands, surface run-off, etc. Users can access all data following guidance as follows:
| Time period | DOI | Image file DOI | Reference |
| Summer 2022–2023 (71 rides) | https://doi.org/10.5281/zenodo.15 053252 | https://doi.org/10.5281/zenodo.15 053336 | Carlson 2025b |
| Fall 2021–2023 (54 rides) | https://doi.org/10.5281/zenodo.15 053261 | https://doi.org/10.5281/zenodo.15 053390 | Carlson 2025c |
| Miscellaneous (53 files) | Sensor sheets, source files, pictures, etc. | https://doi.org/10.5281/zenodo.15 054004 | Carlson 2025d |
A set of screenshot images of combined sensor data time series, recorded for every ride, exists at separate Zenodo addresses; see table above and Data Description below. Users can familiarize themselves with these data by viewing a short (<5 minute) proof-of-concept video available at https://youtu.be/nMjBFbXxNWU.
This paper describes the utility of a bicycle as a measurement device for tracking pavement temperature. It goes into exhaustive length describing the equipment setup and the experimental method and provides access to the data. The author states that this equipment could be used to study pavement characteristics and temperatures with finer resolution than satellites.
The paper provides little discussion of how bicycles have been used for numerous local weather studies in the past, although he does provide a few references in his literature list, and does not describe how the bicycles were used in those studies. He does not describe other methods that could be used to do similar studies such as drones, and also does not mention how satellite imagery has become much more fine-scale over time and is expected to improve its resolution in the future. He describes the method of measuring air temperature compared to pavement temperature and notes that they do not correlate, but did not explain what factors might contribute to the difference. He did not attempt to compare the temperature discrepancy to the ambient weather conditions, including wind and humidity, that might be blowing the air from the land surrounding the pavement to the bicycle's path and contributing to the difference between the air and pavement temperatures.
The author spends a long time doing a correlation between the temperature at the nearest National Weather Service station but notes that the nearest station is about 24 km (14 miles) and 250 m (850 ft). In general, if a station is more than 10 miles or 100 feet different in elevation from the target location it is not considered a comparable record, so this seems like a useless enterprise, although in weather patterns that have uniform air masses, it might have some relevance. The correlations in the paper are discussed in general terms as "excellent" or "very positive" but there has been no attempt to quantify exactly how good they are so this does not seem germane to the main theme of the paper.
Another shortfall of the analysis is a description of how the use of a bicycle would make sense economically, since it is very labor and time-intensive and can cover only a small area at a time. This makes the use of a bicycle as a measuring device to be of extremely limited use. It would even more difficult if it were to be employed in a city or more urban area compared to the relatively traffic-free area of rural Montana. In those areas it might make more sense to use driverless cars like the ones Google uses for mapping, and that would also reduce the labor cost and improve safety since no humans would be involved.
I commend the author for an extremely detailed description of his equipment and his method of collecting data, because it could easily be reproduced from what information he has provided. It was clearly a labor of love. It is a very well-documented data set but of extremely limited use. Practically speaking, I cannot see that anyone would ever choose to use this method in any case that would require it since safer, less labor-intensive, and equally or more accurate methods are available using other types of technology. Comparisons of pavement temperature would be easier to do in a laboratory setting that provides an environment that could be controlled for wind, lighting, humidity, and other ambient atmospheric conditions, so it is hard to justify using this out in the open environment. If the author chooses to revise his manuscript, I would encourage him to include more descriptions of how bicycles have been used in other scientific studies, cut down the equipment and method sections considerably, remove the section on comparisons with distant NWS stations and provide some more general comments on the general pattern of agreement between NWS stations and his own measurements, and discuss what advantages using a bicycle has in comparison to other methods of collecting similar information.