Skip to main content

Project

I spy with my little eye: automated detection of agricultural parcels and crops using satellite images



Different views of agricultural landscapes from space
© ESA Copernicus Open Access Hub: https://scihub.copernicus.eu/dhus/#/home
Different views of agricultural landscapes from space

Assessing the potentials of remote sensing data in the fields of land use, agricultural economics and biodiversity

German-wide and plot-specific land use data is so far not public available. This can change due to the new data from satellite imagery from the Copernicus program (Sentinel), which is free and taken at a high frequency.

Background and Objective

So far, the resolution of freely available satellite remote sensing data in Germany and in the EU was too low and thus only of limited use for the description and analysis of land use dynamics. Since 2015, the European Earth observation programme “Copernicus” provides high resolution satellite remote sensing data regarding area and time dimensions. This new quality of data offers the opportunity to display up-to-date information on the land use and land use structures in Germany with a high degree of spatial accuracy for each single plot of land. Thus, this satellite remote sensing data could help to improve the explanation of the relationship between land use intensity and its regional distribution and to better answer questions regarding the protection of abiotic resources as well as the distribution and development of biodiversity.

The main aim is to extend the existing estimation approach for the Thünen-Atlas on agricultural use (www.thuenen.de/thuenen-atlas) with remote sensing information on cropping pattern and to develop appropriate inference statistics. Remote sensing data, provided by JKI, will be used in the statistical estimation approach developed for the Thünen-Atlas to improve the fit of the downscaling. Additionally, it should be assessed, how other data sources (data collection on soil use, IACS) could be complemented or substituted.

 

 

Target Group

 

The following objectives should be achieved:

  1. Updating the “Thünen-Atlas on agricultural use” as well as the inclusion and analysis of field specific remote sensing data
  2. Development of methods and their applications to assess the quality of the estimation models (inference statistics)
  3. Map on agricultural plots for Germany

Approach

The following steps will be developed:

  • Building a test environment for the use of Sentinel 1 and Sentinel 2 data
  • Development of an approach to recover the optimal parameters for segmenting remote sensing data into parcels using multi-temporal S1/S2 stacks
  • Development of an approach to process German-wide segmentation of parcels with the help of cloud services such as available in CODE-DE
  • Development of methods to detect fields crop from satellite images
  • Validation of methods using ground proof data from agricultural farms

Results

  1. An automated processing routine has been developed for the german-wide delineation of agricultural fields from Sentinel-1 and Sentinel-2 images.
  2. The routine can be used every year for the delineation of agricultural fields in Germany.
  3. Using this routine, all agricultural fields in Germany from 2017 to 2021 have been delineated.
  4. By intersecting the field outlines with satellite-based mapping of the main land use classes, it was possible to produce nationwide maps of land use in the agricultural landscape for Germany.

The developed routine will be further enhanced in follow-up projects. The data provide an important basis for ongoing projects, e.g. MonViA, KlimaFern or HotSpots Erosion.

Links and Downloads

Involved external Thünen-Partners

  • Martin-Luther-Universität Halle-Wittenberg
    (Halle (Saale), Deutschland)

Duration

9.2017 - 8.2022

More Information

Project status: finished

Publications to the project

  1. 0

    Tetteh GO, Schwieder M, Erasmi S, Conrad C, Gocht A (2023) Comparison of an optimised multiresolution segmentation approach with deep neural networks for delineating agricultural fields from Sentinel-2 images. J Photogramm Remote Sensing Geoinf Sci 91(4):295-312, DOI:10.1007/s41064-023-00247-x

    https://literatur.thuenen.de/digbib_extern/dn066421.pdf

  2. 1

    Tetteh GO, Gocht A, Erasmi S, Schwieder M, Conrad C (2021) Evaluation of sentinel-1 and sentinel-2 feature sets for delineating agricultural fields in heterogeneous landscapes. IEEE Access 9:116702-116719, DOI:10.1109/ACCESS.2021.3105903

    https://literatur.thuenen.de/digbib_extern/dn063902.pdf

  3. 2

    Schlund M, Erasmi S (2020) Sentinel-1 time series data for monitoring the phenology of winter wheat. Remote Sens Environ 246:111814, DOI:10.1016/j.rse.2020.111814

  4. 3

    Tetteh GO, Gocht A, Schwieder M, Erasmi S, Conrad C (2020) Unsupervised parameterization for optimal segmentation of agricultural parcels from satellite images in different agricultural landscapes. Remote Sensing 12(18):3096, DOI:10.3390/rs12183096

    https://literatur.thuenen.de/digbib_extern/dn062673.pdf

Scroll to top