Climate Refugia: the physical hydrologic and disturbance basis Zachary Holden - US Forest Service Region 1, Missoula MT Marco P. Maneta – University of Montana Dept. Geosciences Alan Swanson – University of Montana Dept. of Geography John Abatzoglou –University of Idaho Jason Forthofer – US Forest Service, Missoula MT Solomon Dobrowski – University of Montana Dept. of Forestry Allen Warren – University of Montana Anna Klene – University of Monana Dept. of Geography Vegetation management in Complex Terrain Fine-scale gradients drive large variation in vegetation and fuel dynamics We lack fundamental tools and data needed to make informed Decisions about: What to plant where How fast it will grow How it will burn Penman-Montieth equation for evapotranspiration Integrates climate and energy into mechanistic variables Temperature Radiation Atmospheric Vapor Pressure (RH) Aerodynamic resistance (Wind) Each variable in the Penman-Monteith model varies with terrain Scaling Climate in Mountainous Terrain • Mountains create steep biophysical gradients • Every energy input to available moisture varies at fine scale in complex terrain • • • • • Radiation Minimum temperature Max. temperature atmospheric humidity Wind speed Holden and Jolly (2011) Scaling Climate in Mountainous Terrain • Mountains create steep biophysical gradients • Every energy input to available moisture varies at fine scale in complex terrain • • • • • Radiation Minimum temperature Max. temperature atmospheric humidity Wind speed Holden and Jolly (2011) Scaling Climate in Mountainous Terrain • Mountains create steep biophysical gradients • Every energy input to available moisture varies at fine scale in complex terrain • • • • • Radiation Min. temperature Max. temperature atmospheric humidity Wind speed Holden and Jolly (2011) Scaling Climate in Mountainous Terrain • Mountains create steep biophysical gradients • Every energy input to available moisture varies at fine scale in complex terrain • • • • • Radiation Min. temperature Max. temperature atmospheric humidity Wind speed Holden and Jolly (2011) Topographic variation in windspeed Slower wind speeds in valley bottoms Higher wind speeds on ridge tops Large effect on ET WindNinja Simulation for August 13, 2013 Soil water holding capacity Soil depth and physical properties make up the “bucket” that stores water making it available for plants STATSGO STATSGO data has complete US coverage – But it’s thought to poorly characterize soil variability SSURGO higher quality but large areas Of missing data in western US raw SSURGO Deeper soils in valley bottoms Deeper soils on Northfacing slopes Massive microclimate sampling with low-cost sensor networks 2000 sites in N. Rockies and Canada (2010-2013) 300 sites in WA/OR/CA in 2013-2014 | 12 Development of high resolution daily gridded air temperature data with distributed sensor networks For the US Northern Rockies 240 meter daily air temperature grids 1979-2013 daily Tmin and Tmax Holden et al. (in press) 13 Cold air drainage potential (CAD-P) • Difference between free air temperature and observed surface temperature • Modeled as a function of topography 14 Cold Air Drainage potential model Difference between free air temperature (NARR) and sensor observations Modeled as a function of terrain covariates 15 Daily cloud/shade corrected radiation Wind speed modeling with WindNinja • Historical daily wind climatology (240 meters) • Daily 1979-2013 simulations on 1 degree tiles Missoula, MT 1981-2010 Mean daily August Wind Speed Spatially complete maps of soil properties (gSSURGO) Imputation of missing SSURGO data using terrain and satellite data Deeper soils in areas of local accumulation Deeper soils on Shaded slopes shallow soils on steep slopes | 18 Vegetation and climate refugia • Vegetation mediates near-surface climate (radiation, air temperature, skin temperature, interception, wind speed) • Fires and disturbance accelerate climate-driven shifts in vegetation • What factors will mediate the effects or ‘severity’ of wildfires as warming continues? 2009 Kootenai Creek Fire, Montana MTBS fires (1984-2014) Modeling potential for low severity (non-stand replacing) wildfire effects Spatially independent sample of pixels from 535 Northern Rockies wildfires Pixels classified as low severity or “other” • linear model • Low Severity = GLM(DNBR ~ poly(solar) + poly(CAD) + wind + soil + ndvi*soil + ndvi*solar + deficit anomaly) Model fits GLM Random Forests Gradient Boosting AUC 0.70 0.71 0.72 Partial effects of model terms Interaction between soil depth and pre-fire vegetation • Probability of low severity: Deep soil mediates severity where pre-fire vegetation density is high Highest with low prefire greenness and shallow soils Lower severity on sparsely vegetated south facing slopes Lower severity on shaded but densely vegetated north facing (shaded) slopes Missoula, MT Valleys and north slopes are unique environments • high cold air drainage potential, low wind speeds, shading from sun and deeper soils Historical wildfire data suggest that fine-scale variation in energy and moisture mediate the effects of wildfires when they occur Revising the fire triangle terrain weather Veg/fuels A revised fire triangle weather topoclimate Veg/fuels terrain TOPOFIRE: a system for mapping terrain influences on climate for improved wildfire decision support Topofire.dbs.umt.edu TOPOFIRE datasets • • • • • • • • CONUS Daily 240 meter (1979-present): Solar radiation Minimum temperature Maximum temperature Dewpoint Vapor Pressure Deficit Snow Water Equivalent Soil Moisture Fuel Moisture Thank you Extra slides in case of questions Minimum temperature High resolution daily air temperature models for the US Northern Rockies Tmin = reanalysis lapse + CAD * pressure + humidity + MODIS VCF | 35 Maximum temperature 02/10/2015 Empirical model with physical basis: Tmax = reanalysis lapse + radiation * FASST soil moisture + MODIS VCF Tmax: captures differences in north and South slope temperatures Tmax: captures interaction between Surface moisture and insolation | 36 1981-2010 2035-2065 difference More low severity fire in areas with high CAD potential Small differences with wind speed No apparent differences with Soil depth??? Lower severity on southfacing slopes??