Predicting Species and Structural Diversity of Temperate Forests with Satellite Remote Sensing and Deep Learning

Abstract

Anthropogenically-driven climate change, land-use changes, and related biodiversity losses are threatening the capability of forests to provide a variety of valuable ecosystem services. The magnitude and diversity of these services are governed by tree species richness and structural complexity as essential regulators of forest biodiversity. Sound conservation and sustainable management strategies rely on information from biodiversity indicators that is conventionally derived by field-based, periodical inventory campaigns. However, these data are usually site-specific and not spatially explicit, hampering their use for large-scale monitoring applications. Therefore, the main objective of our study was to build a robust method for spatially explicit modeling of biodiversity variables across temperate forest types using open-access satellite data and deep learning models. Field data were obtained from the Biodiversity Exploratories, a research infrastructure platform that supports ecological research in Germany. A total of 150 forest plots were sampled between 2014 and 2018, covering a broad range of environmental and forest management gradients across Germany. From field data, we derived key indicators of tree species diversity (Shannon Wiener Index) and structural heterogeneity (standard deviation of tree diameter) as proxies of forest biodiversity. Deep neural networks were used to predict the selected biodiversity variables based on Sentinel-1 and Sentinel-2 images from 2017. Predictions of tree diameter variation achieved good accuracy (r2 = 0.51) using Sentinel-1 winter-based backscatter data. The best models of species diversity used a set of Sentinel-1 and Sentinel-2 features but achieved lower accuracies (r2 = 0.25). Our results demonstrate the potential of deep learning and satellite remote sensing to predict forest parameters across a broad range of environmental and management gradients at the landscape scale, in contrast to most studies that focus on very homogeneous settings. These highly generalizable and spatially continuous models can be used for monitoring ecosystem status and functions, contributing to sustainable management practices, and answering complex ecological questions.

PREDICTING VEGETATION ATTRIBUTES WITH NEURAL NETWORKS AND SENTINEL-1 & 2

“…Leaf area index can be used to study the patterns that govern the global leaf economics spectrum, and field data can also be collected fairly easily, for forests as well as for grasslands. For other parameters, such as stand density, an r 2 of 0.47 and RRMSE of 0.33 were achieved when fusing Sentinel-1 and Sentinel-2 predictors (see Hoffmann et al, (2022) for further details). Other forest structural parameters such as gap frequency of DBH_std cannot be mapped at the finest spatial resolution of the satellite, since they vary with area in a non-linear way.…”

“…A synthetic cloudless time series of Sentinel-2 imagery was created using FORCE processing software (Frantz, 2019). For Sentinel-1, we created median multitemporal backscatter composites for summer and winter season in Google Earth Engine (Hoffmann et al, 2022). The study period is ranges from 2017-2020 for grasslands, and 2014-2018 for forests.…”

Multi-Sensor Satellite Analysis of Drought Impact on Vegetation in Central Germany: Insights from MODIS and Sentinel-2 Time Series Analysis

“…The main land cover types in the study area are forest and grassland with some patches of croplands. Forests cover 83% of the area and usually grow on limestone bedrock (Hoffmann et al, 2022). Forest ecosystems can be characterized as temperate beech forests.…”

“…Due to Sentinel-2's up to 10-m spatial resolution (Drusch et al, 2012) and ve-day observation period, these data are highly suitable for monitoring various land covers including forests, croplands, and grasslands (Henebry, 2019). It has been demonstrated that the Sentinel-2 time series is particularly useful for capturing intra-eld vegetation dynamics by mapping biophysical parameters in natural (Gri ths, Nendel, Pickert, & Hostert, 2020; Wang et al, 2019) and managed ecosystems (L. Gao et al, 2020;Hoffmann, Muro, & Dubovyk, 2022). Sentinel-2 has the potential to map the effects of droughts accurately and consistently on a variety of ecosystems, thanks to its enhanced spatial and temporal resolution capabilities.…”

Spatiotemporal Variation and Driving Factors of Water Supply Services in the Three Gorges Reservoir Area of China Based on Supply-Demand Balance

“…However, the results were from the lack of spatial concepts, single evaluation, and preliminary dynamic studies [20]. In recent years, with the development of remote sensing and GIS technology, some researchers have started to conduct ecosystem service studies with the help of spatial analysis models, which have become a breakthrough in solving the above problems [21]. The Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model is the most widely used.…”