Last week severe storms devastated large areas of the Veneto region, North East of Italy.
Local and national communities, administrations and Civil Protection bodies immediately cooperated to face the multitude of issues caused by those storms at different levels of management.
At the European level, the Copernicus EMS (Emergency Management Service – Mapping) (one of the services of the Copernicus programme) was activated by the Italian National Civil Protection Department.
The official communication by the Twitter account of Copernicus EMS, following a series of tweets on the topic, communicated the creation of a new activation “EMSR332: Flood in Veneto, Italy“:
The aim of this short article is not to present in detail the work of Copernicus EMS and the content of activation EMSR332, but it is just a quick contribution to the integration of information produced by the activation with a few feedbacks on possible follow up actions.
One of the main outputs of the Copernicus EMS work is the creation of a series of vector files, representing basic information on hydrography, transportation, land use, built-up areas and, in particular, a layer named observer_event, which contains a first evaluation of the flooded areas, based on some satellite images (in this case from the satellite Sentinel 1) collected a few days after the event.
The issue of separated maps
The Copernicus EMS work was separated in 9 areas: Bassano del Grappa  · Cortina d’Ampezzo  · Feltre  · Padova  · Pordenone  · Rovereto  · Trento  · Venezia  · Verona .
For each of them, various base layers where created, including the observed_event layer, and are downloadable in separated compressed zip files from http://emergency.copernicus.eu/mapping/list-of-components/EMSR332
In this way, the information is scattered in different places and it is not easy to “see it” in a single map.
A python notebook to download and merge all the maps
For the purpose of aggregating all the observed_event layers in a unique one, I reused and slightly adapted some code already developed after the Italian earthquake in 2016:
- the python script written by Paolo Frizzera: https://github.com/emergenzeHack/terremotocentro_geodata/tree/gh-pages/CopernicusEMS/scripts, which actually does the merge
- and a Jupyter notebook written by Andrea Borruso and myself, which adds the downloading of the files and some descriptive information: https://github.com/emergenzeHack/terremotocentro_geodata/blob/gh-pages/CopernicusEMS/EMSR190/CopernicusEMS_autoDownMerge.ipynb
Here you can find the data and Jupyter notebook specifically related to the Copernicus EMS 332 activation: https://github.com/alesarrett/CopernicusEMS/tree/master/EMSR332
Inside you can find the Jupyter notebook, all the vector files produced by Copernicus EMS and specifically the merged observed_event shapefile which you can also see as blue areas in the figure below and access in an interactive navigable map here.
With small adjustments, the code can be easily adapted to repeat the same operations (download and merge of vector layers) for any other Copernicus EMS activation.
It would be great to have such a functionality (the integration of layers produced by the different maps in an activation) directly inside the CopernicusEMS portal, maybe also with an interactive web GIS interface and reusable interoperable services.
As visible in the figure, almost all flooded areas have been identified on the plain, while almost no flood event has been identified in the mountain. This is probably not a real “error”, because in the mountain damages where actually caused mainly by severe wind storms. Anyway, some differences, probably due to different approaches to the analysis of the images are visible on the gap between maps 2 and 4, where a lot of areas have been identified as flooded, and the map 9, where no areas have been identified on the western border.
How to map wind-damaged forests?
The main question to me is how Copernicus EMS could help in identifying and mapping forests impacted by the winds that uprooted trees in several areas that are not appearing in this analysis (that is, correctly, related “only” to floods).
This is probably possible only through more detailed images: is it foreseen that other satellites or airborne images are collected in the near future to have such a more detailed analysis in the mountain areas?
Are there already examples (in Copernicus or elsewhere) of mapping activities specifically related to forests damaged by strong wind events?
As already suggested in my initial tweet, I think that the OpenStreetMap community could help in this, if more detailed images are available in the future.
Given the importance of forest in these areas and the extensions of damages, these questions will probably become important in the near future, when the efforts and costs for “cleaning” the forest from collapsed trees will have to be planned and monitored during months and maybe years…