Modeling Climate Change Impacts on an Arctic Polygonal Tundra: 1. Rates of Permafrost Thaw Depend on Changes in Vegetation and Drainage

Abstract

Model projections of permafrost thaw during the next century diverge widely. Here we used ecosys to examine how climate change will affect permafrost thaw in a polygonal tundra at Barrow AK. The model was tested against diurnal and seasonal variation in energy exchange, soil heat flux, soil temperature (Ts), and active layer depth (ALD) measured during 2014 and 2015, and interannual variation in ALD measured from 1991 to 2015. During RCP 8.5 climate change from 2015 to 2085, increases in Ta and precipitation (P) to 6.2 °C and 27% above current values, and in atmospheric CO2 concentrations (Ca) to763 μmol mol−1, altered energy exchange by increasing leaf area index of dominant sedge relative to that of moss. Increased Ca and sedge leaf area index imposed greater stomatal control of transpiration and reduced soil heat fluxes, slowing soil warming, limiting increases in evapotranspiration, and thereby causing gradual soil wetting. Consequently, increases in surface Ts and ALD of 2.4–4.7 °C and 21–24 cm above current values were modeled after 70 years. ALD increase was slowed if model boundary conditions were altered to improve landscape drainage. These rates were smaller than those of earlier modeling studies, some of which did not account for changes in vegetation, but are closer to those derived from current studies of warming impacts in the region. Therefore, accounting for climate change effects on vegetation density and composition, and consequent effects on surface energy budgets, will cause slower increases in Ts and ALD to be modeled during climate change simulations.

Modeling Climate Change Impacts on an Arctic Polygonal Tundra: 2. Changes in CO2 and CH4 Exchange Depend on Rates of Permafrost Thaw as Affected by Changes in Vegetation and Drainage

“…Annual NPP modeled in LCP and FCP features (Figures b and c) and R h (Figures b and c) increased with mean annual air temperature ( MAT a ) and P . Increases in NPP and R h were greater in higher versus lower features (as shown for 2085 in Table ) due to greater increases in ALD of higher features (Grant et al, ) from increasing P (Grant et al, ). Increases in NPP exceeded those in R h modeled in all features during most years, causing net CO 2 uptake to rise as climate change progressed (Figure and Table ).…”

“…Artificial soil heating raised T s modeled and measured through the upper 50 cm of the soil profile by 4 °C (Figure 2b in Grant et al, ), driving more rapid evaporation (Figure 3a in Grant et al, ) and so reducing θ relative to that of unheated control (Figure 2c in Grant et al, ). In the model, higher T s raised below‐ground R a more than GPP, reducing NPP (Table ).…”

“…Effluxes of CH 4 measured at BEO during 2014 by Zona et al () and modeled during 2084 with 2014 weather in the baseline run (Figure a) remained small during winter but rose sharply with early summer warming (DOY 170–190; Figure b). Effluxes declined gradually with autumn cooling (after DOY 260), sustained by shrinking active layers that persisted until the end of the year (Figure 5a in Grant et al, ). Large spatial variation in CH 4 effluxes modeled among landscape features was partially apparent in differences between effluxes aggregated from LCP and FCP, with respectively larger and smaller proportions of lower features.…”

“…Annual net CO 2 exchange modelled in troughs, rims, and centers of the (a, c, and e) LCP and (b, d, and f) FCP at Barrow Experimental Observatory from 2016 to 2085 with meteorological data from 1981 to 2015 (baseline) and with these data incremented hourly with + T a + C a + P (see Table 2 in Grant et al, ). LCP = low‐centered polygon; FCP = flat‐centered polygon.…”

“…The trough and the center were connected through a 1 m breach in the rim, based on the observation of Dafflon et al () that LCP ridges are variable in height. The flat‐centered polygonal (FCP) landform was represented by features with the same dimensions, but the center was level with the rim (Figure 1b in Grant et al, ). The landform surfaces were thus 36% centers, 28% rims, and 36% troughs, similar to those derived from a high‐resolution digital elevation model by Kumar et al ().…”