Science question 4
Big data, non-conventional data, and their uses
A cutting-edge research field especially pertinent in cities is about the non-conventional data. Urban climate studies, in most cases, are restricted by lack of available synoptic meteorological stations for inclusion in these studies. Meteorological stations are placed in general in the countryside, often in airport fields, in order to follow the WMO instrumentation rules. This is because such stations are aimed to observe the weather at the synoptic scale, and should then be less perturbed by local features, such as dense urbanization. Even if a very few numbers of meteorological stations exist in the agglomeration of Paris, they are located in parks, and therefore do not really represent the urban weather. Satellite imagery provide surface temperature information, but for Paris at 3km of for geostationary satellite and 1km two times a day for polar satellites. Furthermore, the interpretation of these images is difficult because of the extremely strong small-scale heterogeneity in one km² of urban surface. Therefore, traditional observing ways do not allow to observe the urban climate, even less to evaluate models at 100m of resolution.
Non-conventional observations are then a pertinent way to gather information on the urban micro-climate. Indeed, such information is provided by people voluntary, and given the density of persons within cities, there is potentially a high density of such observation. For example, netatmo stations are personal weather stations that can be connected to the internet. Over Paris, there are more than 10,000 stations netatmo stations. Even if the quality of the measurement is not good (either because of the station, without radiation shield, or the precise location of it), there exists some new works that extract UHI quantification from this data (Napoly et al 2018, see figure below). The quality controlled data will be provided (for at least one summer).
Paris UHI quantified using netatmo non-conventional stations (from Napoly et al 2018). Spatial distribution of air temperature difference in Paris, France based on quality-controlled citizen weather station data, June 21, 2017 at 00:00 h (UTC).
Many scientific questions linked to the use of non-conventional data could be explored on the site of Paris. How to use these to quantify the UHI variability? How to perform the fusion between several sources of non-conventional data, or with conventional ones? Can they be used for model evaluation? Can such data be assimilated in NWP systems?
Use of Artificial Intelligence is also an evolving technology that should be studied.
Other aspects that could be linked to non-conventional data, is the use of crowdsourcing and open databases (such as Open Street Map) in order to be able to describe relatively finely any city in the world. The very fine data on Paris and Marseille would serve as benchmark for the developed methodologies.
