Changing snowpack dynamics: Phase predictions and forest implications
MS: University of Colorado Boulder, 2016.
We use a 29-year observational precipitation phase dataset with ~71 million precipitation records to explore the influence of three physically relevant variables (air temperature, relative humidity, surface pressure) on controlling precipitation phase (rain or snow) over the land surface. This results in the most extensive study to date of global precipitation phase controls. We find that precipitation events that fall in low relative humidity environments have greater snow frequencies at higher temperatures (Temp > 0°C) compared with events that fall in high relative humidity environments. Similarly, we observe greater snow frequencies for precipitation events that fall at low surface pressures compared with events that fall at high surface pressures. We develop a binary logistic regression model using air temperature, relative humidity and pressure as predictor variables for precipitation phase (snow or not snow). This global precipitation phase model is trained on one half of the observational dataset and displays an accuracy of ~88% when tested on the remaining observations, which is a stark improvement from the traditional temperature-only based model that has a prediction accuracy of ~71%. Our model shows that air temperature is the most influential predictor variable, followed by relative humidity and lastly surface pressure.
Chapter II explores the impact of changing snowpack dynamics on subalpine forest carbon balance trends. Previous work demonstrates conflicting evidence regarding the influence of snowmelt timing on forest net ecosystem exchange (NEE). Using 15 years of eddy-covariance measurements in Colorado, we show that years with earlier snowmelt exhibit less net carbon uptake during the snow ablation period. Earlier snowmelt was correlated with lower air temperatures during the snow ablation period (R2=0.72, P<0.001), with lower air temperatures resulting in reduced rates of daytime NEE (R2=0.87, P<0.001). Hence, climate warming, counter-intuitively, leads to colder atmospheric temperatures during the snow ablation period and concomitantly reduced rates of net carbon uptake. We developed a multiple linear regression using snow ablation period mean air temperature and peak snow water equivalent (SWE) magnitude as predictors of ablation period NEE (R2=0.79, P<0.001). Our model predicts that earlier melt and decreased SWE can cause a 45% reduction in mid-century ablation-period net carbon uptake, which is a period of high potential for forest productivity.