EO-Based Ecosystem Services Module for Organic Farming Monitoring

The EO-Based Ecosystem Services module leverages satellite imagery and derived markers to monitor organic farming compliance and assess general parcel conditions. This module integrates with the DIAS/CDSE (Copernicus Data Space Ecosystem) and Amazon Web Services (AWS) Open Data storage, providing users with extensive satellite data archives and advanced machine learning models for area monitoring.

Key Components

Image Provider:

  • Uses Sentinel Hub to efficiently retrieve Satellite Imagery

  • Allows on-the-cloud processing of satellite imagery (i.e. computation of vegetation indices such as NDVI)

  • Enables access to custom data stored in cloud storage buckets

  • Enables retrieval of temporal statistics over parcels to generate signals that are used as input for the markers.

Marker Provider Module:

  • Markers are insights about what is happening on the parcel derived from satellite (or other aerial) imagery.

  • The module delivers computed markers for parcels within a specific area.

  • Inputs include parcel identifiers, marker types, time range.

  • Outputs are delivered in JSON format – they can either be classifications (i.e. the confidence of the model that the parcel is organic), events (i.e. the time period over which greening was detected on the parcel) or observation-based (i.e. the observations on which bare soil was detected)

  • Utilizes the image provider module to source satellite imagery derived signals.

Interfaces

Image Provider:

  • The image provider is based on Sentinel Hub which allows retrieval of georeferenced raster data via it’s API or through OGC compliant services

  • The service offers various APIs (and more):

    • OGC API: Sentinel Hub allows retrieval of data seamlessly and effortlessly through the standard OGC services and enalbes access in standard GIS application with support for powerful WMS features.
    • Processing API: The Process API enables users to perform advanced processing tasks on satellite data. It supports the creation of custom image processing chains using a wide range of algorithms, including filtering, band math, index computation, and more.
    • Statistical API: Calculates statistics for a satellite image without downloading it. Calculate histograms, percentile calculations, and more.

Marker Provider Module:

  • Marker provider is connected to Theros components through a Rest API.

  • Marker information for a specific parcel can be retrieved by calling the service with the marker and scope identifier (where scope defines the area of interest and the time of interest).

  • Marker provider returns results in JSON format

Services

  • Imagery provider: Given input parameters, return corresponding satellite / aerial / raster imagery for the given area.

  • Marker provider: Retrieve marker information for a specific marker type and parcel.

The following markers are currently computed and made available via the Theros Marker provider:

  • Homogeneity marker: The homogeneity marker serves to identify polygons on which several crops are being grown. The polygon for which the markers are computed should constitute only a single crop. This can help the certificate agency identify claims that are not consistent with the state on the ground.

  • Greening and harvest marker: The greening and harvest marker identifies greening and harvest phases in the FOI’s NDVI time series and combines them into a single period that signifies the life cycle of a crop. It can be a very useful tool for detecting presence of catch crops or green cover and nitrogen-fixing crops, i.e the management practices under the EFA obligation of which main objective it to safeguard and improve biodiversity on farms. The greening and harvest marker can give answers to the following questions: When was the main crop sown and when harvested? After the harvest of the main crop, when did the second crop start developing? Until when was the second crop present on the field, i.e. when was it harvested?

  • Bare soil marker: The bare-soil marker identifies all observations with exposed bare soil due to field being recently plowed, or due to non-photosynthetic vegetation cover. The latter can be a consequence of a harvest or vegetation drying up on the field. This can help identify periods where the field was bare and allow the agency to check whether this is consistent with what is beeing claimed.

  • Crop marker: The crop marker can be used to determine consistency of a agricultural parcel (FOI) with its declared crop.

  • Organic / non-organic marker: Similar methodology as is used for the crop marker is also used for the organic / non organic marker. It checks whether a parcels temporal profile is consistent with a parcel that is grown as organic or not.

  • Similarity and Euclidian Distance Marker: Similarity and Euclidean distance markers belong to the group of crop classification markers. Output of both markers can be used to check validity of a FOI’s claim. In their essence both markers compare signal time series of a FOI to signal time series of. Similarity and Euclidean distance markers perform statistical comparison and don’t require any model training, which is their biggest advantage.

Usage and applications

This module significantly enhances the monitoring capabilities for organic farming practices. It enables broad area monitoring of activities that are done on farms that are claimed to be organic. It aids in the precise identification of parcel conditions, using advanced statistical analysis and machine learning models that use satellite imagery as input.

Suitable for stakeholders in agricultural sectors who require accurate and up-to-date information on land use and compliance with organic farming standards over large areas quickly and efficiently.  It helps with flagging suspicious activity and allows targeted spot visits to the parcels identified as suspicious.