Downstream service / application development for national statistics and reporting
This action aims to develop and test a downstream application / service to support the annual national or sub-national monitoring of land use and land cover that is required for (statistical) reporting purposes by different European public authorities such as the German Federal Statistical Office (Statistisches Bundesamt, Destatis). To date, base datasets for statistics and reporting duties are collected by public authorities with hardly any support of earth observation (EO) techniques. In principle, there would be considerable advantages that objective EO-based land use / cover maps could contribute, but there are several aspects that still hinder their application for statistical reporting: While classifiers and pre-processing of EO data have become highly automated, one of the most time consuming aspects in EO-based mapping is the collection of suitable training data that meet the needs of EO classification methods. This time-consuming sample collection has to be repeated when maps have to be updated, or when maps with slightly different thematic foci (regarding their legends) have to be created for other reporting obligations.
Thus, this action aims to develop an application / service that supports public agencies to produce EO-based land use/cover maps for their needs in a more automated manner. Focus is on developing a repeatable approach for collecting generic, high-quality and representative training data and for supervised mapping. The action will further demonstrate in a test case how such training data and approaches can be used for national mapping. The approach will exploit Copernicus Sentinel data, in combination with existing in-situ data and further available geoinformation. Building on an active user dialogue, the application / service will be developed according to user requirements. In a first step, the action is oriented towards developing an application / service for a German public agency (Statistisches Bundesamt). However, this action is not necessarily restricted to only German authorities but user requirement and uptake actions will aim at involving stakeholders with similar needs from other EU countries and countries bordering the EU in order to stimulate international adoption and cross-border cooperation.
Within this action, a concept of a generic, semi-automated classification approach including the collection of suitable training data for land cover assessments will be developed and applied for a national mapping test case in Germany. The development and test case application will be designed in close collaboration with German public agencies, in particular Destatis, and by also considering user requirements of European stakeholders with similar needs. The action comprises (1) a detailed identification of user requirements (workshop and surveys), (2) the concept development for training data collection, and, based there upon, (3) a test case demonstration of a national mapping exercise, followed by (4) a user uptake workshop.
User requirements will be identified based on written surveys, a user requirement workshop and through close communication with German public agencies such as Destatis. The automated training data selection concept will make value out of existing authoritative data (e.g. topographic reference data (Digitales Landschaftsmodel, DLM, digital elevation models, LUCAS data), further in-situ and geo-data, and EO data (primarily Copernicus Sentinel data). The idea behind integrating multiple input data for identification of training samples is to build this process upon various factors like geographic location, exposure, terrain slope besides the spectral signature of surface. The information content of these datasets will be checked, combined, and evaluated in order to finally select training samples according to principles of spatial distribution, completeness, representativeness and homogeneity. The sample set will then be used for demonstrating a national mapping exercise for Germany using Copernicus Sentinel-1, Sentinel-2 and/or Sentinel-3 data. Regarding the output in form of thematic map content, one out of various possible themes or land cover types will be selected based on the collected user requirements. It could e.g. be a further development and (thematic) refinement of one of the existing Copernicus high resolution layers for Germany that is adapted to national statistical reporting requirements or that can contribute to the monitoring of sub-indicators of sustainable development goals. This action includes user requirement and user uptake workshops at national and cross-border level in order to work towards an internationally transferable application development, and in order to stimulate international exchange and cooperation between user communities.
Outputs and Results
- Systematic collection of user requirement through workshop and user requirement surveys
- Development of an application/service for training data creation
- Test case demonstration of national mapping to support reporting duties of public agencies
- Stimulated international exchange and user uptake through international user uptake workshop