Assessing post-fire water quality changes in reservoirs : Insights from a large dataset in Portugal

Abstract

Wildfires in the Mediterranean basin, especially in Portugal, have increased in extent and frequency over the last few years. One of the impacts of wildfires on humans and ecosystems is on the water quality of surface waters. Ashes and increased erosion rates might, for example, change oxygen levels and elevate the influx of sediments, nutrients, or other water quality-related components like metals and polycyclic aromatic hydrocarbons (PAHs), possibly affecting water supply. In this study, time series of eight water quality parameters: biological- and chemical oxygen demand (BOD and COD), electrical conductivity (EC), total phosphorous (TP), nitrate (NO₃⁻), total suspended sediments (TSS), dissolved oxygen (DO), and pH, were assessed via changepoint analysis to identify events of post-fire water contamination in over 60 Portuguese reservoirs. Further, possible fire, watershed, reservoir, and climate-related drivers were linked with the occurrence of these contamination events through logistic regression using generalized additive models. All measured parameters exhibited post-fire changes, with some being more frequently affected than others. The concentrations of TP, NO₃⁻, and TSS showed a noticeable increase following 9.6 %, 12.6 %, and 13.6 % of all wildfires, respectively. Most changes fell into the unusually large fire seasons of 2003–2005 and 2017. The most significant impacts could be seen in southern reservoirs after the fire seasons of 2003–2005. The burned area ratio of the watershed was identified as the main driver of post-fire water contamination, while reservoir and climate-related characteristics like water levels also played a significant role in some parameters. Increased levels of suspended sediments were identified as a potential threat to water supply, especially when large wildfires coincide with drought-induced low reservoir water levels. The identification of post-fire water contamination events and their drivers from large datasets can inform water managers about potential threats to water supply.