Climate change and wildland fire impacts on seasonal snow measurements
Cowherd, M. (2025). Climate change and wildland fire impacts on seasonal snow measurements. PhD Thesis. University of California: Berkeley. ISBN 9798288862724. 125 pp.
|  |
| Available in | Author |
|
Document type: Dissertation
|
| Abstract |
Snow is a vital water resource, contributing to supply, storage, and predictability for downstream uses. Our ability to track snow changes is linked to our ability to understand and represent the mechanisms behind snow accumulation and ablation as well as our ability to make useful observations. The relationship between measured snow properties and unmeasured snow properties is therefore a crucial gap between the actual water resource present and our ability to make optimal water management decisions. This work addresses the need to understand snow measurements and snowpack processes in the face of two prominent sources of change that challenge traditional snow measurement: climate change and wildland fire. First, I use an ensemble of Earth system models to show global shifts in the frequency, severity, and drivers of snow drought around the world. Warm snow droughts, in which a normal or high precipitation year still leads to a water storage deficit, represent an emerging threat to water management in many regions of the world. Second, I focus on snow droughts in the western United States using dynamically downscaled climate projections. This analysis of higher-resolution model outputs identifies spatial variability in snowpack response to climate change. This finding raises further questions on the use of snow observations at a specific location to predict values at a nearby location, as is traditionally done with snow pillow networks in the US and other countries. The strategic but static locations of snow pillows act as representatives of the state of snow water resources across the regions they are located. However, these locations may not react to climate change in the same manner as unmeasured locations, leading to a paradox: how can we know if the network is representative if we don’t measure outside of the network? Third, I model the behavior of the snow pillow networks in the western United States in a future climate projection to estimate how useful traditional measurements will remain in a warmer future. By comparing the skill of models ranging from linear regression to convolutional neural networks and with input data from sparse snow pillow proxies to hypothetical dense snow water equivalent observations, I show the potential for addressing future snow management challenges. In particular, I show that explicit two-dimensional spatial correlations are a key component of successful predictions under new climates. Finally, I present results from a field campaign in El Dorado County, California, showing that legacy snow course locations respond to fire in ways that are not representative of non-snow-course locations. Together, this work shows that the future of snowpack distributions and measurements will be nonstationary and increasingly driven by temperature rather than precipitation. Short-term change from wildland fire and long-term change from climate change demonstrate the need to adapt how we think about snowpack information, not just snowpack quantity, to disturbance. |
|
