On the use of a snow aridity index to predict remotely sensed forest productivity in the presence of bark beetle disturbance

Abstract

We used multiple sources of remotely sensed and ground based information to evaluate the spatiotemporal variability of snowpack accumulation, potential evapotranspiration (PET), and Normalized Difference Vegetation Index (NDVI) throughout the Southern Rocky Mountain ecoregion, USA. Relationships between these variables were used to establish baseline values of expected forest productivity given water and energy inputs. Although both the snow water equivalent (SWE) and a snow aridity index (SAI), which used SWE to normalize PET, were significant predictors of the long-term (1989-2012) NDVI, SAI explained 11% more NDVI variability than SWE. Deviations from these relationships were subsequently explored in the context of widespread forest mortality due to bark beetles. Over the entire study area, NDVI was lower per unit SAI in beetle-disturbed compared to undisturbed areas during snow-related drought; however, both SAI and NDVI were spatially heterogeneous within this domain. As a result, we selected three focus areas inside the larger study area within which to isolate the relative impacts of SAI and disturbance on NDVI using multivariate linear regression. These models explained 66%-85% of the NDVI and further suggested that both SAI and disturbance effects were significant, although the disturbance effect was generally greater. These results establish the utility of SAI as a measure of moisture limitation in snow-dominated systems and demonstrate a reduction in forest productivity due to bark beetle disturbance that is particularly evident during drought conditions resultant from low snow accumulation during the winter.

Snowmelt‐Driven Trade‐Offs Between Early and Late Season Productivity Negatively Impact Forest Carbon Uptake During Drought

“…The significant early/late season effect on total GPP (Figure b) may thus result from spatial integration of heterogeneous meteorological and/or ecosystem conditions; that is, some areas within the ecoregion may not have experienced sufficient drought severity to activate the cumulative GPP response during the widespread drought years of 2002 and 2012. Longer distances from the best fit linear trendline in 2007 and 2011 may have been due to interannual lags in snow water storage or spatial variability within the ecoregion (Knowles et al, ).…”

“…At the Niwot Ridge, CO (US‐NR1; Blanken et al, ), and Valles Caldera, NM (US‐VCM; Litvak, ), AmeriFlux sites within the study domain, potential evapotranspiration (PET) was calculated from meteorological forcing data using the Penman approach (Shuttleworth, ). A snow aridity index (Knowles et al, ) was subsequently evaluated as PET/SWE where annual maximum SWE was measured at the Niwot SNOTEL site (US‐NR1) or averaged from measurements at the Quemazon and Vacas Locas SNOTEL sites (US‐VCM). An analogous calculation was performed at the ecoregion scale where snow aridity index was equal to PE/SWE.…”

“…Notwithstanding, seasonal changes in forest productivity associated with the combination of these perturbations have not been characterized. Snow aridity takes into account both snow water equivalent (SWE) and PE and is therefore a useful composite index for understanding drought (Knowles et al, ). In particular, the 2012 growing season in the Southern Rocky Mountains provides an ideal analog with which to evaluate the sensitivity of forest productivity to the combined effects of reduced SWE and increased PE (i.e., snow aridity) as it included the third lowest snowpack and warmest spring on record (Wolf et al, ).…”

The sensitivity of snow ephemerality to warming climate across an arid to montane vegetation gradient

“…Therefore, as water inputs shift earlier in the year from earlier melt and winter rain, carbon uptake by montane forests early in the year may not offset reduced late season uptake (Knowles, Molotch, Trujillo, & Litvak, 2018). In addition to less net carbon uptake during earlier snow water inputs, carbon allocation at the tree scale has important controls for resisting forest disturbance like drought (Hart, Veblen, Eisenhart, Jarvis, & Kulakowski, 2014;Knowles, Lestak, & Molotch, 2017;Williams et al, 2013) and impacts net partitioning to transpiration (Garcia, Tague, & Choate, 2016). These differences in ecophysiological adaptations to earlier water inputs are likely to vary substantially based on species and traits, with conifer forests showing higher ability to utilize earlier water inputs (Kelly & Goulden, 2016).…”

“…The stressful effects of snowmelt-mediated changes in soil water availability on vegetation are likely to be magnified by increased summer temperatures and higher vapour pressure deficit. Numerous studies have found that the combination of low winter precipitation and hot summer droughts is correlated with widespread forest mortality events from drought, fire, and insects in the Western United States (e.g., Hart et al, 2014;Knowles et al, 2017;Westerling, Hidalgo, Cayan, & Swetnam, 2006;Williams et al, 2013). These climateinduced vegetation stressors are expected to alter semiarid vegetation distribution (Chambers & Pellant, 2008;Guardiola-Claramonte et al, 2011), by requiring upslope migration (Kelly & Goulden, 2008) or movement into areas of microrefugia (McLaughlin et al, 2017).…”