Scientists' warning on extreme wildfire risks to water supply

Abstract 2020 is the year of wildfire records. California experienced its three largest fires early in its fire season. The Pantanal, the largest wetland on the planet, burned over 20% of its surface. More than 18 million hectares of forest and bushland burned during the 2019–2020 fire season in Australia, killing 33 people, destroying nearly 2500 homes, and endangering many endemic species. The direct cost of damages is being counted in dozens of billion dollars, but the indirect costs on water‐related ecosystem services and benefits could be equally expensive, with impacts lasting for decades. In Australia, the extreme precipitation (“200 mm day −1 in several location”) that interrupted the catastrophic wildfire season triggered a series of watershed effects from headwaters to areas downstream. The increased runoff and erosion from burned areas disrupted water supplies in several locations. These post‐fire watershed hazards via source water contamination, flash floods, and mudslides can represent substantial, systemic long‐term risks to drinking water production, aquatic life, and socio‐economic activity. Scenarios similar to the recent event in Australia are now predicted to unfold in the Western USA. This is a new reality that societies will have to live with as uncharted fire activity, water crises, and widespread human footprint collide all‐around of the world. Therefore, we advocate for a more proactive approach to wildfire‐watershed risk governance in an effort to advance and protect water security. We also argue that there is no easy solution to reducing this risk and that investments in both green (i.e., natural) and grey (i.e., built) infrastructure will be necessary. Further, we propose strategies to combine modern data analytics with existing tools for use by water and land managers worldwide to leverage several decades worth of data and knowledge on post‐fire hydrology.

extreme precipitation ("200 mm day À1 in several location") that interrupted the catastrophic wildfire season triggered a series of watershed effects from headwaters to areas downstream. The increased runoff and erosion from burned areas disrupted water supplies in several locations. These post-fire watershed hazards via source water contamination, flash floods, and mudslides can represent substantial, systemic long-term risks to drinking water production, aquatic life, and socio-economic activity. Scenarios similar to the recent event in Australia are now predicted to unfold in the Western USA. This is a new reality that societies will have to live with as uncharted fire activity, water crises, and widespread human footprint collide all-around of the world. Therefore, we advocate for a more proactive approach to wildfire-watershed risk governance in an effort to advance and protect water security. We also argue that there is no easy solution to reducing this risk and that investments in both green (i.e., natural) and grey (i.e., built) infrastructure will be necessary. Further, we propose strategies to combine modern data analytics with existing tools for use by water and land managers worldwide to leverage several decades worth of data and knowledge on post-fire hydrology.

K E Y W O R D S
climate change, extreme events, fire regime restoration, forest ecosystem services, risk governance, socio-hydrology, water security, watershed protection

| WILDFIRE RISKS TO WATER SECURITY
The 2019-2020 wildfire season in Australia was unprecedented in recorded history (Boer et al., 2020), burning several catchments supplying drinking water to the 5.5 million Sydney's inhabitants (Box 1) and threatening many aquatic species with extinction (Pittock, 2020). Similarly, the 2020 wildfires in the Western US, the greatest recorded annual area burned in the country (4.2 million hectares 1 ), caused dozens of millions in damages to water distribution systems (Walton, 2020) and triggered widespread debris-flow warnings. 2 In both countries, heavy rainfall eventually stopped fire spread, but triggered extensive runoff, erosion, and mass movements degrading source water quality and availability, potentially for decades to come (Hanscombe, 2020;Hohner et al., 2019;Niemeyer et al., 2020;Robinne, 2020).
Australia and Western US extreme wildfire events, among many others, were directly linked to persistent drought and record-breaking temperatures (Harris & Lucas, 2019;van Oldenborgh et al., 2020). As the climate becomes increasingly hotter and drier and human activities continue to expand, threats to water security will become more prevalent (AghaKouchak et al., 2018;Hallema et al., 2018;Robinne et al., 2018).
The growing overlap of extreme hydroclimatic events and expanding human activities makes water crises more likely (Franco, 2020).
Climate warming is leading to greater fire danger, including in regions where fire was previously uncommon (Flannigan et al., 2013;Higuera & Abatzoglou, 2021;Shukla et al., 2019). Indeed, warmer temperatures often lead to drier fuels in which fires can ignite sooner, spread further, and burn more intensely (Flannigan et al., 2016). In rainforests and wetlands, where wet conditions usually slow or stop the spread of fires, unusually dry conditions will lead to destructive fires impacting surface waters with sediment, carbon compounds, and toxic metals (Abraham et al., 2017;Granath et al., 2016). Hence, wildfire-watershed risks (WWR) represent a global challenge that must be addressed through proactive forest and water governance, starting with identification of areas at risk. We must then strategically apply innovative risk reduction strategies to address long-term, largescale impacts from catastrophic wildfires in source watersheds (Abadi et al., 2016;Kinoshita et al., 2016).
Building on the second World Scientists' Warning to Humanity (Ripple et al., 2017), we argue that scientists, NGOs, water providers, watershed managers, fire managers, policy-makers, and citizens share the responsibility to collect, share, and use knowledge of WWR to develop sustainable environmental policies. In this commentary, we explain the systemic nature of these risks, illustrating the need for regionally adaptive and proactive WWR governance. We also briefly review existing alternatives to WWR management, and we provide examples of forward-thinking governance schemes in at-risk locations.
Box 1 2019-2020 fire extent, severity, and post-fire erosion in municipal watersheds of New South Wales and Australian Capital Territory (Figure 1).
New South Wales was the epicentre of the 2019-2020 extreme wildfire season, experiencing 445 fires, burning $5.8-million hectares. The yearly average from 1989 to 2019 was 300 fires burning $180 000 hectares, with a maximum area burned of 1.7 million hectares (computed from NSW historical fire database, see Data S1). Compared to the long-term average, in 2019-2020 there were $1.5-times more fires, which burned 32-times greater area.
Most concerning, the wildfires affected source watersheds that supply drinking water for 5.5-million people, including 25 catchments supplying Sydney and six catchments supplying Australia's capital city Canberra. In total, 46 of 78 (59%) water supply catchments burned to varying extents (14 over 50%) (see Data S1). Several townships had to restrict water consumption through boil water advisories and noconsumption advisories. These restrictions were due, in part, to direct damage to water treatment and distribution infrastructures and to power grid damage leading to shutdown of water treatment facilities. As a result, the Australian Government pledged to commit $88.1-million AUD to create a national disaster research centre (Australian Government, Department of Industry, Science, Energy, and Resource, 2020).

| CATASTROPHIC WILDFIRES AS SOCIO-HYDROLOGICAL EXTREMES
Anthropogenic development has long affected the occurrence and magnitude of wildfires, droughts, and floods. These are not "true" natural hazards anymore: in many regions, most wildfires are human-caused, often fed by excessive fuel availability resulting from past fire exclusion efforts . Intensive human water use has concurrently increased the magnitude of droughts in water-scarce regions, such as California (AghaKouchak et al., 2015), and many megacities struggle to meet growing water demand (McDonald et al., 2014). Comparatively, communities with budgetary constraints, endemic poverty, gender issues, and systemic racism are likely to be disproportionately impacted by water supply impairment (Cross, 2001;Davies et al., 2018). Hence, urban water supply is increasingly vulnerable to disruption caused by wildfires (Balch et al., 2020;Keys et al., 2019) WWR is inherently systemic (Deere et al., 2017): fire is a source of socio-hydrological extreme, whereby the dynamic interactions and dependencies between upstream source water and downstream water demand can be disrupted due to (a) exceptional wildfire magnitude, (b) vulnerability of water supply infrastructure, and (c) lack of risk governance (Di Baldassarre et al., 2018). Seeing fire as a sociohydrological extreme allows spatial-temporal modelling of negative wildfire impacts on water resources and the influence of WWR reduction efforts within the larger challenges of watershed management.
Water resource managers are able to test various disaster risk scenarios and adjust to envisioned consequences of future wildfires-ex ante-and to the actual consequences of past wildfires-ex post (Linton & Budds, 2014). Therefore, water security and forest management are intrinsically linked; differing coping capacities of sociohydrosystems around the world will, however, lead to different WWR governance regimes (Di Baldassarre et al., 2013;Fischer et al., 2016;Kumar, 2015;Srinivasan et al., 2012).

Box 2 Global wildfire-watershed risk hotspots
Wildfire-watershed risk hotspots are locations where water supplies and communities are susceptible to wildfire effects, such as waterworks damages (e.g., water supply infrastructure and reservoirs), loss of ecosystem services

| Wildfire-watershed risk governance
Stakeholders hold different degrees of risk knowledge, perception, and tolerance to risk (Klinke & Renn, 2012); WWR governance can account for these varying risk cultures. Effective WWR governance must also account for the compound effects of catastrophic wildfires, water supply vulnerabilities, and ecological, social, and economic stresses that can cascade towards socio-hydrosystem collapse Community and water assets that are vulnerable to wildfire often display different risk profiles due, for instance, to different wildfire regimes. As such, post-fire outcomes for the US and Australia do not directly apply to Chile, India, or Canada (Nunes et al., 2018). Socioeconomic development, land-use history, public health, and relationships to water also often diverge (Linton & Budds, 2014; Miller F I G U R E 1 Estimated burn severity and estimated erosion in New South Wales and Australia Capital Territory. Panel (a): Wildfire severity values derived from very-high resolution Sentinel-2 satellite imagery using the fire extent and severity mapping algorithm from the Department of Planning, industry and environment (see Data S1). Panel (b): Estimated hillslope erosion values for the month of February 2020 calculated using the revised universal soil loss equation (RUSLE) model (see Data S1). Estimated erosion rates were higher for high burn severity. Municipal watersheds supply drinking water to Sydney and Canberra, among other communities. Data from: https://www.environment.nsw.gov.au/, https://www.seed.nsw.gov.au/, and https://data.nsw.gov.au/data/dataset et al., 2017; Murphy et al., 2020). Given the social-ecological complexity involved, advancing WWR governance requires synthesis of knowledge among hydrologists, wildfire scientists, citizens, water providers, land managers, and various levels of governments responsible for watershed planning (Clark et al., 2016;Ostrom, 1996;Wheater & Gober, 2015;Parkes et al., 2010).
Understanding and embedding contextual factors is one major challenge in the development of locally-relevant decision-support tools for WWR governance (Blair & Buytaert, 2016;Hallema et al., 2019;Paté-Cornell, 2012;Ruckelshaus et al., 2015). Parameterizing and adapting such tools rely heavily on data availability (Fischer et al., 2016;Kumar, 2015). Reliable hydrological data from fire-impacted areas is hard to obtain even in economically developed countries, due to the difficulties inherent to predicting future fire activity and limited available research funds to deploy and maintain hydrological monitoring networks. Where data exist, there may be restrictive access policies and data inconsistencies. Risk governance, based on a collaborative approach to knowledge production, will help gather missing information towards the reduction of socio-hydrosystems' vulnerability, leading to more relevant and accurate WWR-reduction tools as a result (Canning et al., 2020;Hallema et al., 2018;Lowndes et al., 2017;Thompson et al., 2019;Wheater & Gober, 2015).
Even so, it may never be possible to exactly predict WWR (Gannon et al., 2019;O. D. Jones et al., 2014). Therefore, investing in risk prevention and literacy, as well as stakeholder preparedness through transparent communication, will help devise locally appropriate responses towards risk reduction (Boisramé et al., 2019;Kinoshita et al., 2016;McWethy et al., 2019). Working this way will help reach consensus towards the definition of regional risk profiles warranting tailored watershed policies for successful risk management; such initiatives will also facilitate the social acceptability of risk and of the actions aiming at its reduction (Blair & Buytaert, 2016;Hamilton et al., 2019;Wheater & Gober, 2015).

| BEYOND FIREFIGHTING
Appropriate forest management maintains natural water storage and increase drought resistance, while reducing the negative impacts of unwanted fires in source watersheds (Boisramé et al., 2019;van Wagtendonk, 2007). Active forest management, including mechanical thinning and prescribed burning, can be effec-  savannas into dense closed-canopy forests, ecosystem water use increased significantly due to increased evapotranspiration (Boisramé et al., 2019;Roche et al., 2018), leading to reduced streamflow. Safely using naturally-occurring wildfires to restore overstocked source watersheds can increase water security by reducing the likelihood of F I G U R E 3 Risk governance in the wildfire-watershed value chain. Wildfire-watershed risks are recognized through the identification of interactions between upstream wildfire hazard (i.e., likelihood of a wildfire event of a given, potentially harmful, magnitude), watershed vulnerability, and downstream water security. After identification of water security vulnerabilities and their social and economic consequences, effective wildfire-watershed risk governance will offer a set of options to deal with existing at-risk situations. Rapid, slow, and prolonged onset drivers refer to the speed and depth at which changes in fire and forest management can occur: Rapid onset drivers can be acted upon quickly and have immediate effects (e.g., biomass reduction), while slow onset drivers are deeply ingrained and affect fire activity on the long term, even after changes have been made (e.g., fire exclusion policies). Icons made by Freepik and Eucalip extreme fire events while simultaneously increasing streamflow and subsurface water storage (Boisramé et al., 2019;Roche et al., 2018); this option must, however, be carefully reviewed, as water demand from post-fire regrowth can offset gains in water availability (Brookhouse et al., 2013;Niemeyer et al., 2020).

Degradation of wetlands and their biodiversity worldwide has led
to a greater occurrence of wildfires, particularly in peat landscapes (Turetsky et al., 2015;Wilkinson et al., 2018). Restoring wetlands, biodiversity, and natural fire regimes where the ecosystems have coevolved with fire will help maintain the water table closer to the surface, keeping water in the landscape, preventing extreme fires, regulating water flows, and enhancing water quality (Fairfax & Whittle, 2020;Granath et al., 2016;Wilkinson et al., 2018).
Upfront investments in wildfire prevention, biomass reduction, and watershed restoration are more successful and cost-effective than firefighting and post-fire slope stabilization (North, Stephens, et al., 2015). Simultaneously improving watershed health and water supply infrastructures will reduce vulnerability (Box 3). Likewise, payments for ecosystem services schemes, whereby downstream communities financially support the protection of forested headwaters, are gaining traction (UNECE and FAO, 2018). There is, however, no easy solution: WWR mitigation must rely on a combination of grey (e.g., infrastructure retrofitting) and green infrastructures (e.g., watershed restoration) adapted to current and projected risk levels.

Box 3 Building on existing WWR governance
The following locations where WWR has been identified are proactively dealing with wildfires and their consequences.
Melbourne, Victoria, Australia: Melbourne relies on forested catchments for 80% of its water supply, in which Eucalyptus forests are highly flammable. The terrain is steep and prone to high hillslope erosion rates post-fire, leading to sediment loads >100 times greater than normal. Water entering treatment facilities is unfiltered, therefore small changes in colour and turbidity significantly impact disinfection efficacy and drinking water quality. Modelling indicates a large wildfire in the Upper Yarra Reservoir could result in water being untreatable for a year or more. Although desalinated water is available, it is expensive and cannot meet demand should the major water supply catchment go offline due to fire-caused contamination. Thus, Melbourne Water has invested millions in research programs to inform fuel reduction, firefighting efforts, and post-fire response (Canning et al., 2020).
Sydney, New South Wales, Australia: The extensive 2019-2020 wildfires burned 35% of Sydney's largest water supply catchment, Warragamba. In 2019, the Greenwattle Creek fire was followed by intense rainfall exceeding 200 mm in one day, resulting in substantial ash and sediment transfer into the water storage, Lake Burragorang.
Hence, Sydney Water led risk mitigation efforts-ash, sediment, and contaminant transfer modelling into the lake; sediment plume monitoring; water quality monitoring. The vertically flexible design of the offtake at the dam wall enabled water to be extracted from outside of the plume zone (Canning et al., 2020).