By Natali Moss
It is noted that the developed program is incomplete and cannot be considered perfect. However, it offers a neutral view of the intensity of hostilities in Ukraine. The surveillance system is based on FIRMS data that has been developed to track forest fires. Journalists have created a machine learning algorithm that evaluates the location of the fire and analyzes whether it is connected with war.
The algorithm has 100 models, and if 95 of them agrees that the fire is abnormal compared to the pre -war years, the algorithm of the machine will consider a fire related to combat. However, the algorithm depends on the weather, because if the front line is cloudy, the satellite will not fix the fire. Therefore, the authors once again emphasize that they cannot provide 100% accurate data. Over the past 7 days, more than 20,000 fires have been recorded.
A significant number is likely to be the result of artillery, while others are rockets and drones. The indicators are almost record, the authors say. The lack of ammunition plays a role in this, which limits the ability of the Armed Forces to suppress Russian attacks. However, the Ukrainian army should not be underestimated. The analysis shows that the defense forces are in return. We will remind, on March 1 the media wrote that the occupiers came on Chinese golf machines instead of BMP.
USA
Українська
English
Français
Deutsch
Italiano
Polski
Čeština
Español
Slovensku
