Evaluating forest wildfire effect on tree increment patterns for boreonemoral forests in Sweden: A pilot study using remote sensing

Sammanfattning: The release of anthropogenic greenhouse gases (GHGs) has substantially increased the global mean surface air temperature. Increases in global mean surface air temperature will lead to warmer and drier conditions, promoting more frequent, long-lasting, intense forest wildfires. The usage of remote sensing (RS) can aid in quantifying forest characteristics and large-scale changes in forest ecosystems. RS can detect wildfires, assess the damage level of burnt forests, and enhance the evaluation of forest regeneration after a fire event. Differenced normalised burn ratio (dNBR), Normalised differential vegetation index (NDVI), and differenced normalised difference vegetation index (dNDVI) have been proven to assess forest fire disturbance and forest health. However, many of these techniques have yet to be validated by field sampling in Swedish boreonemoral forest systems. The study aimed to investigate and evaluate the existing RS methodology for fire disturbance and forest health in a group of Swedish boreonemoral forests. This was done by using the proposed RS methodology and dendrochronology assessment. Estimating burn severity (dNBR) and forest health (dNDVI) on boreonemoral forests show good potential as the fire disturbance signal and health of the forest are captured using Sentinel-2 images. This study concluded that using the presented RS methodology for visualisation (dNBR and dNDVI) is viable as it helps users visualise the effects and severity of boreal forest wildfires and vegetation recovery. Using dNBR as a tool to estimate burn severity patterns has been proven possible but unreliable regarding the relationship between high burn severity and decreased tree increment patterns. NDVI temporal changes have been shown to explain some of the changes to Pine increment patterns but are restricted to 1–2-year trends. However, NDVI might be reliable for evaluating temporal growth increment patterns in Swedish boreonemoral forests. Due to the few sites, this cannot be confirmed or denied. Both presented RS methods are robust but need modifying as variabilities in reflectance can be uncertain. The usage of the used RS methodology shows potential for further studies, as improvements can be made from this study to validate the presented method and assessment better.