Learning from Research

Soil property map

Soil property maps will be constructed based on data from various sources to generate spatially explicit indicators of the soil properties. Data will include Earth Observation data and products, existing soil databases and data collected by portable spectrometers (Micro Electro Mechanical Systems -MEMS).

Page contents

Output Description

The DIONE project will adopt a soil monitoring approach based on three tiers. 

First Tier (data acquisition): This stage aims to assimilate all the data used, which may come from: 

This data will include a fusion from various sources using either archive or open-access EO data of both multispectral sensors (Landsat-8 and Sentinel-2) and Synthetic Aperture Radar (SAR) sensors (Sentinel-1). Geospatial information from the SoilGrids global soil repository will also be used. The SoilGrids is a Global Soil Information system released by the International Soil Reference Information Centre (ISRIC) – World Soil Information.

DIONE will make use of the European Soil Data Centre (ESDAC) Land Use and Cover Area frame Survey (LUCAS) and the GEO-CRADLE Soil Spectral Library (GSSL). DIONE also will explore the potential of utilising the Global Soil Laboratory Network (GLOSOLAN), allowing for newly harmonised soil data sets.

These devices are low-cost portable and handled spectrometers in the visible and near-infrared (VNIR) range. They will contribute to the effective monitoring of soil. Because of their mobile use, MEMS can improve farming productivity by bringing real-time information of soil status through a wireless connection with the sensors at a dramatically reduced cost compared to laboratory soil analysis. In collaboration with novel machine learning techniques, MEMS will be able to estimate with efficient accuracy levels of the soil's physical and chemical status, including Soil Organic carbon (SOC), particle size distribution (sand, silt, clay), electrical conductivity, pH, total nitrogen, and others.

Second Tier (data modelling): This stage aims to process the spectral point data collected from existing databases and portable devices using machine learning algorithms and the EO image data using data mining techniques. 

Third Tier (knowledge): This stage presents the results in an organised soil property map. 

DIONE will estimate soil properties like pH, electrical conductivity, textural composition, other physical and chemical properties, and the Soil Organic Carbon. This data will be used to construct spatially explicit indicators of the soil properties as raster data (e.g. GeoTIFF) which will be made available through a database management system (DBMS). The DBMS will allow soil data and soil maps to be used by another of DIONE’s products named the environmental performance tool of DIONE.

Relevance for monitoring and evaluation of the CAP

The soil property map based on soil data analysis and various ancillary sources is the best and most cost-effective alternative when analytical soil maps do not exist. 

Soil maps are indispensable monitoring tools and can support the estimation of many environmental impact indicators. For example, in monitoring, soil maps allow the Managing Authority to specify the spatial extent of measures by considering various conditionalities in the form of soil Good Agricultural and Environmental Conditions - GAECs 4, 5 and 6. In evaluation, soil maps provide data to estimate the two soil impact indicators, i.e., soil erosion (I.13) and soil organic carbon (I.12). Soil maps also support the assessment of other indicators such as water abstraction and water quality. Of course, soil properties change very slowly, and this change can be measured or become evident after a period that exceeds the period of an RDP. However, if, for example, the policy is successful in establishing cover crops on the fields most prone to soil erosion, this is evidence that the policy confronts soil erosion even though this may not be measurable in the seven years of an RDP’s life. 

The tool claims that acquiring and operating a portable spectral sensor cannot be a barrier to adoption because it is low cost and easy to learn. The transferability of the tool to other regions and Member States depends on the availability of ancillary data and the ease and time with which machine learning algorithms can be trained in new data. Machine learning tools are used to transform the raw data collected through the in-situ soil scanning system to appropriate soil properties, including SOC, clay, pH and CaCO3. In addition, algorithms control the combination of point measurements with EO imagery towards delivering spatially detailed maps using a spiked bottom-up approach that needs calibration with local data. Thus the tool will require a period of testing and calibration before it can be functional. 

Relevance of the output per CAP Objectives

  • Specific Objective 4 - Climate change action
  • Specific Objective 5 - Environmental care

Additional output information

Data collection systems used:

  • Copernicus 
  • LUCAS Soil or relevant soil inventories 
  • National land use surveys 
  • Ad-hoc data collection 

Type of output:

  • New / improved data for M&E
  • Visualisation tools

Associated evaluation approaches:

  • Desk research
  • Data analysis
  • Impact evaluation ex post
  • Impact evaluation ongoing

Spatial scale:

  • Parcel
  • Sub-regional / local

Project information

Dione Logo

An integrated EO-based toolbox for modernising CAP area-based compliance checks and assessing respective environmental impact 

DIONE will devise a prototype toolbox in an operational environment that will use automated technologies to support CAP monitoring. 

The overall aim of the toolbox is to ensure more frequent, accurate and inexpensive compliance checks and support the assessment of environmental impacts. 

The DIONE toolbox will: 

  • Demonstrate the capabilities of the Copernicus DIAS cloud platform as a state-of-the-art (SotA) cloud infrastructure for building country-scale or even continent-scale EO-based monitoring systems. These systems will support the automated monitoring of green direct payments, aligned with the paying agencies’ requirements and recognised best practices at a Technology Readiness Level 7 (TRL7), implying a system prototype demonstration in an operational environment. 
  • Improve the resolution of free and open Sentinel data and combine them with high-resolution drone and commercial data to consider smaller EFA types of increased environmental impact. 
  • Develop and demonstrate ground-based geo-tagged photos that ensure photos with increased positioning accuracy and suitable orientation that are tamper-proof and securely transmitted to the Paying Agency’s compliance monitoring tool, which will complement EO data. 
  • Implement a low-cost system based on spectral sensors that can measure and assess soil quality (organic carbon, level of erosion), quantifying the current level of land degradation in the referenced land parcel. 
  • Integrate the results of the Sen4CAP project concerning the freely available produced crop-type maps, the improved resolution maps identifying non-productive EFAs and complementary information sources into DIONE’s compliance monitoring tool. 
  • Design and implement an environmental performance tool, which will be integrated with Paying Agencies’ monitoring tools through a machine learning-based inferencing system at the regional or national scale. 

Project’s timeframe: 2020 – 2022

Contacts of project holder: Institute of Communication and Computer Systems (info@dione-project.eu) 

Website: DIONE: https://dione-project.eu/ 

CORDIS database: https://cordis.europa.eu/project/id/870378 

Territorial coverage: Cyprus, Greece, Lithuania, Slovenia