COS 146-3 - Identifying refugia from drought in western Canadian forests

Wednesday, August 17, 2022

10:30 AM – 10:45 AM EDT

Location: 513D

Presenting Author(s)

DS

Diana Stralberg

Research Scientist
Canadian Forest Service, Natural Resources Canada
Edmonton, Alberta, Canada

Coauthor(s)

AR

Alec Robitaille

Canadian Forest Service, Natural Resources Canada, Canada
GC

Guillermo Castilla

Canadian Forest Service, Natural Resources Canada, Canada
JC

Jennifer Cartwright

U.S. Geological Survey, United States
MM

Mike Michaelian

Canadian Forest Service, Natural Resources Canada, Canada
TH

Ted Hogg

Canadian Forest Service, Natural Resources Canada, Canada

Background/Question/Methods

As the global climate warms, drought-related tree mortality has increased for many species, especially in parts of western North America where the climatic water balance can be negative. Tree ring data have revealed negative effects of drought on growth rates for several tree species. However, the local effects of drought can vary considerably, and some sites may be buffered from negative effects of drought by topographic and/or edaphic conditions—thereby serving as drought refugia. To evaluate these patterns we used Google Earth Engine and its archives of satellite imagery and climate data to map spatial patterns in historical drought sensitivity over a 37-year period (1985-2021) across western Canada. Using LANDSAT vegetation indices and DAYMET climate data, we modified an index of drought sensitivity (Cartwright et al. 2020) based on the difference in July vegetation greenness between statistically defined drought and baseline periods. We compared performance of the drought sensitivity index across vegetation indices (NDVI, EVI, NBR), antecedent periods (3 months, 12 months, and 5 years), and ecoregions. To evaluate potential drivers of refugia, we performed ecoregionally stratified regression analyses relating drought sensitivity to environmental covariates representing multi-scale landcover, topography, and hydrologic setting.

Results/Conclusions

We found that indices based on NDVI yielded higher absolute values (i.e., better discriminated between drought and baseline conditions) than those based on either EVI or NBR. In general, longer antecedent periods were more important in arid, interior regions, while shorter antecedent periods were more important in wetter, coastal regions. Spatial patterns in the index reflected continental-scale patterns of climatic moisture limitation, as well as local heterogeneity. Preliminary results from regression analyses suggest the importance of topographic position and aspect throughout, but also indicate regionally specific controls on drought that varied according to climate and hydrologic setting. Our planned next step is to evaluate this remotely sensed index using a drought sensitivity index based on annual growth increments from a national tree-ring database developed by the Canadian Forest Service and research partners. The fine spatial resolution (30 m) and broad spatial extent of the mapped index should increase its value as a planning tool for land managers. Potential applications include forest management plans, regional land-use plans and protected area management plans.