NR will work on developing methodologies for analyzing and modeling environmental gradients and properties in space and time, which systematically utilize the strength of remote sensing, specifically Sentinel data. We will focus on:

Develop data fusion strategies

A key component of the project is to connect Sentinel data with other geospatial datasets such as aerial images, terrain models and indices derived from them as well as thematic maps by means of data fusion. We will systematically explore the potential of spatial structures (“image objects”) acquired from a segmentation of relevant (higher resolution) data layers in the data fusion process. The activities will include improved segmentation, co-registration of multi-source data, and object based image analysis.

Develop models for the selected environmental gradients

In this task we will develop methods for estimating the selected environmental gradients and properties at a certain location and point in time.

Develop and improve methodology for detecting environmental changes

The high spatial and temporal resolution of Sentinel-1 and -2 provides opportunities to analyze dynamics of changes of nature at a scale that was previously not possible with satellites. In this task we aim at developing methods for detecting changes in gradients beyond shifts of the underlying nature types, which most likely have a delayed response on changing environmental conditions. Automatic methods for change detection will be evaluated, in particular for detection of abrupt changes due to intervention, but also gradual changes due to climate change. For analysis of climate changes, we will emphasize the study of the spatial-temporal snow cover distribution.

Uncertainty estimates of the models

Estimating the uncertainty of the environmental gradients and models is of vital importance for practical use