Application of a computationally efficient method to approximate gap model results with a probabilistic approach

Abstract

Abstract.To be able to simulate climate change effects on forest dynamics over the whole of Switzerland, we adapted the second-generation DGVM (dynamic global vegetation model) LPJ-GUESS (Lund-Potsdam-Jena General Ecosystem Simulator) to the Alpine environment. We modified model functions, tuned model parameters, and implemented new tree species to represent the potential natural vegetation of Alpine landscapes. Furthermore, we increased the computational efficiency of the model to enable area-covering simulations in a fine resolution (1 km) sufficient for the complex topography of the Alps, which resulted in more than 32 000 simulation grid cells. To this aim, we applied the recently developed method GAPPARD (approximating GAP model results with a Probabilistic Approach to account for stand Replacing Disturbances) (Scherstjanoi et al., 2013) to LPJ-GUESS. GAPPARD derives mean output values from a combination of simulation runs without disturbances and a patch age distribution defined by the disturbance frequency. With this computationally efficient method, which increased the model's speed by approximately the factor 8, we were able to faster detect the shortcomings of LPJ-GUESS functions and parameters. We used the adapted LPJ-GUESS together with GAPPARD to assess the influence of one climate change scenario on dynamics of tree species composition and biomass throughout the 21st century in Switzerland. To allow for comparison with the original model, we additionally simulated forest dynamics along a north-south transect through Switzerland. The results from this transect confirmed the high value of the GAPPARD method despite some limitations towards extreme climatic events. It allowed for the first time to obtain area-wide, detailed high-resolution LPJ-GUESS simulation results for a large part of the Alpine region.

FORests and HYdrology under Climate Change in Switzerland v1.0: a spatially distributed model combining hydrology and forest dynamics

“…Couplings between models that explicitly simulate forest dynamics and hydrological models have so far mostly been the object of experimental studies. The results from these experiments show that coupling these processes may substantially alter model results and behavior (Sutmöller et al, 2011;Schattan et al, 2013). This gives an opportunity to increase the confidence in simulated impacts of climate change on forested ecosystems.…”

“…For example, Sutmöller et al (2011) used the physically based distributed hydrological model WaSiM-ETH (Schulla, 2015) to perform hydrological simulations with vegetation parameters derived from an individual-based forest model driven by different forest management scenarios, which influence forest structure and species composition. Also, Schattan et al (2013) applied the semi-conceptual hydrological model PREVAH (Gurtz et al, 1999) with vegetation parameters obtained from the forest landscape model TreeMig (Lischke et al, 2006). In the original versions of both these hydrological models, vegetation parameters were parameterized as a function of season and land cover class only.…”

Taking off the training wheels: the properties of a dynamic vegetation model without climate envelopes, CLM4.5(ED)

“…We calculate the R 2 and root mean square error (RMSE) of the spatial distribution of each metric. We acknowledge that there exists a choice of metrics (maximum vs. minimum vs. range, and spatial vs. temporal correspondence), but also note that subjectivity in the definition of objective functions is generic to high-dimensional model output (Abramowitz et al, 2008;Randerson et al, 2009;Blyth et al, 2011;Abramowitz, 2012;Kelley et al, 2013;Luo et al, 2012;Schwalm et al, 2013;Anav et al, 2013).…”

Linking tree physiological constraints with predictions of carbon and water fluxes at an old‐growth coniferous forest

“…Cohort‐based models such as GAPPARD (Scherstjanoi et al. ) and TREEMIG (Nabel et al. , Zurbriggen et al.…”

“…A second-generation dynamic global vegetation model (DGVM; Fisher et al 2010Fisher et al , 2015Fisher et al , 2018 has the ability to simulate both sub-daily biophysical processes and longer-term demographic processes, thereby enabling exploration of the interaction of the shorter-and longer-term processes in response to climate drivers. Cohortbased models such as GAPPARD (Scherstjanoi et al 2014) and TREEMIG can better represent the heterogeneity of plants than big-leaf models and have lower computational cost than gap and individual-based models (Dietze and Latimer 2011, Christoffersen et al 2016, Fischer et al 2016. In this study, we applied a cohort-based model, the Ecosystem Demography Model 2 (ED2; Medvigy et al 2009), at the Wind River Experimental Forest, an old-growth forest in Washington with a wealth of observation data and studies ).…”