New deep-learning based approaches for forest landscape modeling

<p>In times of rapid global change, the ability to faithfully predict the development of vegetation on larger scales is of key relevance to society. However, ecosystem models that incorporate enough process understanding for being applicable under future and non-analog conditions are often restricted to finer spatial scales due to data and computational constraints. Recent breakthroughs in machine learning, particularly in the field of deep learning, allow bridging this scale mismatch by providing new means for analyzing data, e.g., in remote sensing, but also new modelling approaches. We here present a novel approach for Scaling Vegetation Dynamics (SVD) which uses a deep neural network for predicting large-scale vegetation development. In a first step, the network learns its representation of vegetation dynamics as a function of current vegetation state and environmental drivers from process-based models and empirical data. The trained model is then used within of a dynamic simulation on large spatial scales. In this contribution we introduce the conceptual approach of SVD and show results for example applications in Europe and the US. More broadly we discuss aspects of applying deep learning in the context of ecological modeling.</p>

Forest adaptation and transition in the eastern united states under climate change : a forest landscape modeling method

Forest management is rapidly shifting in focus to address the adaptive capacity of forests under uncertain future climates. Managers and researchers often utilize models to proactively develop strategies for forest adaptation management and in order for these models to provide useful results they must realistically represent a multitude of complex processes. Here we detail a linked-model methodology for predicting the response of forests to climate change over large heterogeneous landscapes under a range of adaptation management scenarios. We used a forest ecosystem process model to simulate forests across the eastern United States under a range of future climate scenarios and found that ecotones between major forest types or natural community types may be the most vulnerable to large declines in biomass due to climate change. We then show that the implementation of a probability-based method for estimating individual tree fire mortality can realistically reproduce conditions observed in field inventory data. Finally, we test the effectiveness of different climate forest adaptation strategies at maintaining or increasing the presence and geographic distribution of species on a heterogeneous landscape under climate change.