28 July 2017 – As promised, our expert Alen Berta has made final analysis of the Split fire.
Unfortunately, on July 27, it was cloudy in the Split area and the footage could not be used, but as Sentinel 2B is in a test phase (twin brother of Sentinel 2A, which will shorten the recording time to only 5 days), we found the footage of that satellite from July 22, where there were very few clouds.
On the first footage of the area around Split there are a few smaller clouds, so the semi-automatic classification did not give perfect results (due to the cloud and its shadow), therefore, a part of it is not classified and in certain areas there are false readings due to the shadow of the cloud (e.g. in the northwestern part of the fire). Taking into account these limitations, by semi-automatic controlled classification (shown in blue), 1825 ha (18.25 km2) of the surface was burnt; which, by our free estimate, equals the surface of the built part of Split, Solin, Stobreč and Podstrana.
The accuracy of the classification when the sky is clear can be seen in the example of the fire on Mount Svilaj and on the fire northeast of Hrvac that occurred a couple of days before the one in Split. By analyzing this part of the satellite images we can conclude that 1032 ha (10.32 km2) was burnt.
Since these areas are mostly covered with degraded forest vegetation (shrubbery and maquis), there is no data on wood mass for them. This data is extremely important when drafting the National Greenhouse Gas Report (NIR) in order to know how much CO2 is emitted into the atmosphere. However, after the completion of one of our projects (link) we will be able to do it, i.e. to determine the wood mass of the burnt area only on the basis of analysis of such images.