Evaluation of Burn Severity for the Fires of 2020 in the Mountains of Córdoba : Integration of Field and Remote Sensing Data

Abstract

In 2020, the mountains of Córdoba, in central Argentina, registered the worse fire season of the last 31 years. To assess the effects of fire on the flora and soil surface, we integrated field and remote sensing (RS) data to propose a local burn severity classification, which was later used to map and quantify the areas affected with different levels of severity. Also, this local classification was then compared with other three classifications developed in other ecosystems. Field estimations were based on the GeoCBI index and RS data was represented by dNBR, derived from Sentinel 2 images. The model relating GeoCBI and dNBR explained 87 % of data variability. All fires showed heterogeneous levels of burn severity within their boundaries. Overall, from the 280,853 ha burnt in large fires, most area was affected with moderate (48.5 %), followed by low (27.6 %) and high (23.9 %) burn severity. Shrublands and grasslands were affected with moderate to low severity, while forests had moderate to high burn severity, reinforcing the idea that fires represent a threat to forest conservation. The comparison between the local and other severity classifications showed underestimation of burn severity in two cases, while the other provided similar maps and statistics. Nevertheless, the biological and ecological meaning of these categories should not be extrapolated. These results demonstrated the importance of developing local burn severity classifications and the need for testing the suitability of foreign burn severity classifications when they are going to be applied in a different ecosystem than the one where they were proposed.