Remote sensing for nature monitoring

Summary

This report is the result of a review on the possibilities of remote sensing for applications in the nature domain, with emphasis on Natura 2000 habitat monitoring. In recent years, enormous progress has been made in the availability and processing of high-resolution satellite and drone images. This increases the potential application for answering all kinds of policy and nature management questions. We demonstrate that remote sensing can have much added value for the monitoring of habitat distribution and habitat quality across a wide range of nature areas. We also demonstrate that higher spatial resolution of remotely sensed imagery often results in better classification accuracies. Deep learning techniques are also becoming popular since they are able to consider the contextual information and not only the spectral information from the imagery in classifying or identifying objects (from habitats to individual plant species). However, the amount of training data can have a large impact on classification accuracies, much more than for more conventional classification methods. This, then, requires a large investment in the collection of in-situ (field) data as well. Another finding is that including LiDAR and hyperspectral data can significantly improve detailed habitat mapping. In summary, the resource of remote sensing data and techniques should be selected depending on the relevant nature types, research questions and nature targets at a specific local, regional or national scale. It requires more communication between remote sensing researchers and ecologists. If nature goals and remote sensing technologies are brought together at an early stage, many applications will be possible. For the Netherlands, the remote sensing community should focus especially on monitoring the structure and function of habitat types. Also, such large-scale and long-term remote sensing monitoring should become part of a national nature monitoring programme.