Loading...
Characterizing hydrologic response and recovery patterns of disturbed ecosystems from the burn scar to continental scale
Collar, Natalie McLennan
Collar, Natalie McLennan
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2022
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
2023-03-30
Abstract
Forested watersheds comprise much of the source water area in the United States. Downstream, particularly in the arid and over-appropriated west, the runoff generated in the treed and often snow-dominated headwaters is captured, moved, and consumed according to intricate water transfer agreements that rely on predictions of yield and consumptive losses. It is problematic, then, that numerous pressures are expanding the uncertainty around current and future supply estimates. Protracted drought, insect infestation, and wildfire - each capable of modifying hydrologic processes in forested catchments - are escalating in frequency and intensity in step with increasing climate variability and human development patterns. The current density and flammability of fuel loads on America’s public lands, driven by a century of fuel suppression and climate aridification, exacerbate the problem.
Recent extreme fire behavior and lengthening fire seasons have pushed wildfire hydrology into the spotlight among the research community and western water providers. Concurrently, advances in remote sensing and cloud computing capabilities are enabling larger-scale studies of fire behavior trends and source water hydrologic effects in general. However, most studies are still conducted at the plot to basin-scale. Due to the high spatial variability of landscape response to forest disturbance, knowledge gained by direct observation of one location may have limited transferability elsewhere. Thus, predicting the specific effects and hazards posed by fire to an exact location remains difficult: transcendent landscape response patterns, if any, are largely undefined. The pre-existing studies that do evaluate post-fire watershed hydrology at regional scales have done so primarily by way of the streamflow signal; regional-scale studies of the effects of combustion upstream of the basin outlet, where precipitation inputs are first being partitioned into the primary components of the water balance, are currently lacking in the literature.
This dissertation uses remotely sensed and other observational datasets, as well as statistical models, to explore burn scar to regional-scale source water pyrologic patterns with an emphasis on actual evapotranspiration (ETa). Of the primary hydrologic fluxes, ETa is particularly vulnerable to fire-induced modification because its rate is a function of both surface vegetation and sub-surface hydraulic soil properties, each capable of being altered by wildfire. Further, because ETa comprises the largest outgoing flux in terrestrial hydrology, it has been posited that ETa reduction following fire-induced vegetation removal is a primary driver of the amplified runoff response frequently observed following conflagration.
The first chapter (chapter two) is a conterminous United States-scale (CONUS) spatiotemporal study of fire-induced ETa response and recovery patterns. ETa estimates from the United States Geological Survey’s (USGS) 1-km operational Simplified Surface Energy Balance (SSEBop) product were used to characterize post-fire ETa and evaporation ratio (ETa/Precipitation, denoted as “ET/P” herein) shifts at 5,500 unique fires across the CONUS. The largest changes in post-fire ET/P were observed in the southwestern CONUS where first-year ratios were reduced by 50 to 90% and pre-fire ratios were rarely recovered by post-fire year five. Regional and intra-fire ET/P response variability was also highest in the western CONUS where climatic, topographic, and ecologic gradients are steep. Post-fire ET/P modifications were small to negligible in the east-southeast CONUS, and 18% of all 1-km pixels analyzed exhibited small to moderate increases in post-fire year one ET/P, including at nearly every pixel impacted by controlled burns. A comparison of burned and unburned pixel pairs confirmed the role of fire in the shifts but also indicated a high degree of background variability in the ETa and precipitation data. Linear models showed that higher burn severities were consistently correlated with greater post-fire ET/P reductions, while relationships between post-fire ET/P shifts and numerous other landscape attributes (e.g., pre-fire vegetation type) varied spatially in both direction and magnitude.
Chapter three is a pixel-scale evaluation of 13 burn scars scattered throughout the American West. At 30-m spatial resolution, ETa estimates were calculated with the SSEBop model and used to assess the precursors to and effects of fire-induced ETa shifts on pixel-scale hydrology and hydrologic connectivity. Agreement of the 30-m estimates with the 1-km operational SSEBop product was also evaluated and shown to be relatively good but variable by location. Results indicated that fire significantly reduced ETa at every burn scar except in humid Oregon and at the only location comprised of grassland, versus forest or shrub/scrub cover, in the pre-fire period. Further, ETa was still significantly reduced relative to pre-fire ETa conditions by post-fire years six through ten at most burn scars. Runoff/recharge generation zones and regolith ETa zones shifted after fire events in a similar manner to the changes imposed by multi-year drought, and most forested land immediately converted to grass, then shrub/scrub cover, over variable lengths of succession. Non-linear models explained greater than 90% of the post-fire year one ET/P response variability when pixels were grouped by EPA Level III ecoregion with a similar set of topographic, meteorologic, and pyrologic predictor variables used in chapter two. This represented a substantial increase in explanatory power.
Last, chapter four investigates linkages between fire-induced ETa shifts and streamflow signals. Making this leap was paramount for understanding the relevance of the previous chapters’ findings to downstream water availability. Ten of the 13 burn scars analyzed in chapter three were selected with criteria including the presence of burns in more than 5% of drainage areas in a single water year, availability of twenty years of pre and post-fire daily stream gage data, and others. It was observed that significant fire-induced ETa reductions were only distinguishable in basin-scale aggregated datasets when at least 73% of the basin burned. A novel statistical method for isolating landscape from climate-driven streamflow shifts detected significant baseflow and quickflow modification in basins with as little as 6% burned drainage area and showed that the effects of fire on the streamflow signal were seasonally variable. However, shifts only persisted beyond the fifth post-fire year where more than three-quarters of the basin was fire-impacted. Notably, the volume of surplus water from ETa loss reduction was sufficient to account for fire-induced increases in discharge. But where fire-induced streamflow shifts were not significantly cross-correlated with the ETa anomaly, other fire-impacted landscape processes may have contributed to modified runoff generation and routing. Findings suggest that water providers with small, homogeneous source water collection areas are disproportionately vulnerable to fire-induced hydromodification. Results also illustrate the tendency of overarching climate signals to mask or artificially boost the apparent effect of landscape disturbance on streamflow at the basin outlet.
Associated Publications
Rights
Copyright of the original work is retained by the author.