Learning from the Exploratories to make prediction beyond them: AI-based mapping and explanation of grassland biodiversity and ecosystem functions for entire landscape units (BEyond)

预测、理解和监测时空格局是生态学研究的一个重大挑战。该项目的目的是从生物多样性探索站前所未有的数据集中学习，以预测探索站以外的生物多样性和生态系统功能模式-其整个景观单元。我们选择了一个间接的建模方法，包括地形，土壤，气候，土地利用和景观结构作为潜在的驱动因素，除了遥感数据。机器学习为预测映射提供了很好的机会，因为它能够学习驱动因素和生物多样性变量之间的非线性和复杂关系。然而，最近的研究表明，当训练模型被应用于深入研究领域之外的预测时，存在相当大的局限性。空间过拟合和学习科学错误的关系可能会显着导致有限的模型可移植性。缺乏模型的可解释性进一步阻碍了生态研究的进步。为了克服这些限制，我们将开发和应用新的时空预测映射方法，重点是增加模型的可移植性，再加上可解释的人工智能方法。我们希望能够开发出科学合理的模型，提供探索站以外的选定生物多样性变量的时空连续地图。我们进一步超越预测，并期望对前“黑盒”模型的见解提供关于模式、驱动因素和相互作用的新发现。

Prediction, understanding and monitoring of spatio-temporal patterns is a major challenge in ecological research. The aim of this project is to learn from the unprecedented dataset of the Biodiversity Exploratories to predict patterns of biodiversity and ecosystem functioning beyond the Exploratories - for their entire landscape units. We choose an indirect modelling approach by including terrain, soil, climate, landuse and landscape structure as potential drivers in addition to remotely sensed data. Machine learning offers great opportunities for predictive mapping, due to the ability to learn non-linear and complex relationships between drivers and biodiversity variables. However, recent research indicates considerable limitations when trained models are applied to make predictions beyond intensively studied areas. Both, spatial overfitting and the learning of scientifically wrong relationships may significantly lead to a limited model transferability. A lack of model interpretability further prevents advancements for ecological research. To overcome these limitations we will develop and apply new methods for spatio-temporal predictive mapping that focus on increasing the model transferability coupled with methods of explainable artificial intelligence. We expect to be able to develop scientifically sound models that provide spatio-temporal continuous maps of selected biodiversity variables beyond the Exploratories. We further go beyond predictions and expect that insights into the former “black-box” models provide novel findings on patterns, drivers and interactions.