Online Resource 2
Title: Relative effects of climate change and wildfires on stream temperatures: a simulation modeling approach in a RockyMountain watershed
Journal: Climatic Change
Authors: Lisa Holsinger1, Robert E. Keane1, Daniel J. Isaak2, Lisa Eby3, Michael K. Young4
(1)USDA Forest Service (USFS) Rocky Mountain Research Station (RMRS) Fire Sciences Laboratory, 5775 Hwy 10 West, Missoula, MT 59808 (,Corresponding Author)
(2)USFS RMRS Air, Water, and Aquatics Program-Boise Aquatic Sciences Laboratory, 322 East Front Street Suite 401, Boise, ID83702
(3)University of Montana, Department of Ecosystem and Conservation Sciences, College of Forestry and Conservation, Missoula, MT59812
(4)USFS RMRS Forestry Sciences Laboratory, 800 East Beckwith Avenue, Missoula, MT59801
We validated the stream temperature regression equation by comparing observations from 24 digital thermographs monitored from August 2010 to August 2011, to stream temperatures predicted by implementing the daily maximum equation in FireBGCv2 for this same period. Modeled predictions compared reasonably well to observed values (rs= 0.83, MAE = 2.59°C,RMSE = 3.42°C) but suggested a tendency for over-predicting stream temperature (Figure 1A). The validation dataset however may have been skewed towards warmer temperatures since thermographs were distributed across only a subset of the EFBR subwatersheds with a considerable concentration in warmer downstream reaches. Figure 1B-D shows relationships between maximum daily stream temperature and each of the dynamic predictor variables based on observed data (2001-2009) used to develop the stream temperature regression equation.
In addition, we conducted a one-way sensitivity analysis to evaluate the order of importance of model predictors to variation in stream temperature (Howard 1988; Clemen 1996). The base values in the stream temperature equation were the median values observed for each predictor variable in the historical climate without active fire management scenario for higher-elevation subwatersheds that represent cold water habitat for native fish and the lower and upper bound values for the dynamic predictor variables were the minimum and maximum values observed across all simulation scenarios. Among the dynamic variables, the air temperature*flow interaction term and air temperature had the most potential to increase stream temperatures given the model parameter sensitivities (Figure 2) and the amount of future changes in climate scenarios. But solar radiation changes associated with fire disturbance could warm streams by up to 2.4°C locally. We emphasize that estimates of solar radiation describe the amount of solar radiation penetrating the riparian canopy and reaching the stream surface, and changes observed in simulations reflect how fire occurrence altered canopy structure where burn severity depends on topography, vegetation, weather and fuel conditions at the location and time of fire. We also note that stream temperatures for climate change studies can be modeled using various approaches (e.g. empirical regression approach, equilibrium temperature modeling), which are detailed at length by others (Caissie 2006; Isaak 2011; Null et al. 2013). We chose a multiple regression modeling approach because it allowed simultaneous assessment of both spatial and temporal factors and was computationally efficient when linked to the complex mechanistic FireBGCv2 model. We did not attempt to model stream temperatures using more complex statistical approaches that could account for spatial autocorrelation among temperature sites (Isaak et al. 2010, Isaak et al. in press) because their increased computational requirements would have precluded linkage to FireBGCv2 for simulations. The dominant role at the network scale of changes in air temperature relative to solar radiation and streamflow is consistent with other findings for the Northern Rocky Mountains (Isaak et al. 2010). Moreover, our overall estimates of stream temperature increases associated with A2/B2 climate scenarios are consistent with other projections of 2°C - 5°C increases for the Pacific Northwest (Cristea and Burges 2010; Mantua et al. 2010).
Figure 1. Relationships of (A) FireBGCv2 predictions vs observed for maximum daily stream temperature based on validation data (2010-2011); and observed stream temperature to (B) observed air temperatures, (C) solar radiation, and (D) streamflow based on monitoring data (2001-2009) used to build stream temperature regression equation.
Figure 2. One-way sensitivity analysis tornado diagram showing the effect of predictor variables on maximum daily stream temperature regression equation. See Table 1 for description of predictor variable inputs.
Table 1. Lower bound, base and upper bound values for predictor variables used as input for one-way sensitivity analysis of maximum daily stream temperature regression equation.
Predictor / Low / Base / HighElevation (m)
Air temperature * Stream flow Interaction term
Air temperature (°C)
Solar radiation (SRAD) (W∙m-2)
Channel slope (drop/length)
Contributing Area (m2)
Stream Flow (m3∙sec-1) / 1,225
0
11.3
43.4
0.00
96,300
0.00 / 1,734
195
13.5
318
0.04
14,472,900
0.22 / 2,269
354.53
22.0
463.2
0.46
1,074,978,048
0.26
Table 2. Mean (± standard error) of dynamic predictors for maximum daily stream temperature observed for each climate scenario in FireBGCv2 simulations.
Predictor / H / B2 / A2Air temperature (°C)
Solar radiation (SRAD) (W∙m-2)
Stream Flow (m3∙sec-1) / 13.56 ± 0.01
316.22 ± 0.19
1.86 ± 0.01 / 15.29 ± 0.01
304.2 6± 0.22
2.35 ± 0.01 / 18.85 ± 0.02
325.53 ± 0.27
1.05 ± 0.01
References
Clemen RT (1996) Making Hard Decisions: An Introduction to Decision Analysis, 2nd Ed., Duxbury, Belmont, CA.
Cristea N, Burges S (2010) An assessment of the current and future thermal regimes of three streams located in the Wenatchee River basin, Washington State: some implications for regional river basin systems. Climatic Change 102:493–520
Howard RA (1988) Decision Analysis: Practice and Promise. ManageSci 34:679-695.
Isaak DJ, Luce CH, Rieman BE, Nagel DE, Peterson EE, Horan DL, Parkes S, Chandler GL (2010) Effects of climate change and wildfire on stream temperatures and salmonid thermal habitat in a mountain river network. Ecol Appl 20:1350–1371
Isaak DJ, Peterson E, Hoef JV, Wenger S, Falke J, Torgersen C, Sowder C, Steel A, Fortin MJ, Jordan C, Reusch A, Som N, Monestiez P (2014) Applications of spatial statistical network models to stream data. WIREs – Water 1.
Mantua, N., Tohver, I. and Hamlet, A., 2010. Climate change impacts on streamflow extremes and summertime stream temperature and their possible consequences for freshwater salmon habitat in WashingtonState. Climatic Change, 102:187-223.
Null, SE, Viers JH, Deas ML, TanakaSK, MountJF (2013) Stream temperature sensitivity to climate warming in California’s Sierra Nevada: impacts to coldwater habitat. Climactic Change 116:149-170