Abstract
   Assessing fire impacts on a landscape scale can be a powerful tool to guide forest management in a changing climate. A key unknown, however, is how increasing wildfire severity is affecting forest structure in the context of management objectives, particularly at the patch scale. This research performs a supervised object-based classification with Support Vector machine learning utilizing multi-spectral data acquired from the 3-meter resolution Planet Labs satellite to identify live tree patches in the Dixie fire footprint in Lassen Volcanic National Park, CA. Patch size distributions pre- and post-fire were compared to target distributions detailed in the park’s Fire Management Plan (FMP). Live tree patch sizes ranged widely in size, from <0.1ha to >600ha and exhibited a statistically significant decrease in median between the pre-fire to the post-fire landscapes. Patch size distributions did not meet FMP targets, though our results indicate Dixie Fire effects did help achieve management targets in some areas. Further, the novel use of such high-resolution, regular acquisition data illustrates the potential to utilize a more robust, repeatable methodology by land managers assessing landscape-level fire effects.

Zoom

https://ucmerced.zoom.us/j/89636364918?pwd=M0JTZEl2YTRvQ21qcDR0RVphN2YxZz09