Title:	Fires in drained forested peatlands in Sweden
DNr:	NAISS 2025/22-1063
Project Type:	NAISS Small Compute
Principal Investigator:	Daniel Escobar Carbonari <daniel.escobar@natgeo.su.se>
Affiliation:	Stockholms universitet
Duration:	2025-08-11 – 2026-03-01
Classification:	10502
Keywords:

Abstract

Rewetting drained, forested peatlands is a central element of Sweden’s climate-mitigation portfolio because a higher water table curbs both ongoing peat oxidation and the risk of deep-burning wildfires that can instantly release centuries of stored carbon. Yet more than one million hectares of Swedish peatland were ditched for forestry, and restoration budgets remain limited. Targeting those funds wisely first requires a rigorous, landscape-scale test of whether drainage intensity truly elevates fire risk, and whether that effect varies among mineral soils, shallow peat and deep peat. This project will deliver the first nationwide statistical analysis of the ditch-fire relationship in Swedish forests. We will merge several state-of-the-art, high-resolution datasets: Sentinel-2 burn-scar composites for 2016-2025, high-confidence MODIS active-fire detections from the FIRMS product, the Swedish National Soil Map, lidar-derived ditch density and ditch depth, and daily Canadian Forest Fire Weather Index (FWI) components from Copernicus. The combined stack exceeds 150 GB and covers more than 500 million 10-metre forest pixels. Fire occurrence (burn-scar presence or active-fire hit) will be modelled as a function of soil class, centred drainage intensity and their interaction. Weighted logistic regression and spatially adaptive generalised additive models will be used to capture both linear and non-linear effects while controlling for spatial autocorrelation. Block-bootstrap resampling across 1 km tiles will quantify uncertainty and test the robustness of results to class imbalance. The outcome will be a set of peer-review-ready effect estimates, odds ratios and marginal‐response curves, that reveal how strongly ditching drives fire probability on each soil type, thereby providing the empirical foundation needed for cost-effective peatland-rewetting decisions.