Suchergebnisse

We consider the potential and limitations of widely accepted remote sensing algorithms for detecting forest areas damaged by fires, which allow the monitoring systems to automatically form data about the fire areas and the areas where death of tree stands is subsequently recorded. The resulting size of the detected areas has a measurement error, which is typically determined on the basis of a onetime survey for a large territory and one forest fire season. Depending on the geographical features of the territory, forest fires have specific spatiotemporal and qualitative characteristics, and are accompanied by specific heterogeneous damage to forests, which affect the accuracy of remote detection of a burnt area or firedamaged forest. Hence the use of the unified largearea error estimate for locallevel surveys could lead to inaccurate results. The analysis of space images of forest fund lands in the Tver region for the period 2007-2022 demonstrated the need to establish regional values of the measurement error for fireimpacted areas. By comparing the medium spatial resolution data with the data of high spatial resolution we identify a regional bias, which is significant given the relatively small size of the detected areas. The study demonstrates the expediency of establishing regional error values for measurements of pyrogenic impact on forest areas. By implementing the suggested changes we could improve the accuracy of remotely-sensed estimates of fireimpacted areas and the amount of associated damage to forests.