Defra - Department for Environment, Food and Rural Affairs.

Science Search

Science and Research Projects

Return to Science Search homepage   Return to Project List

Remote Sensing of Agricultural Land Parcels: Identification of Crops - SD0471

Description
There is a need to have detailed knowledge of agricultural land use at the field level to assist in research and subsequent policy formation. Current data on crops grown in England is only available at a farm level and this is often too coarse to allow detailed analysis. This project aims to develop an automated rule-based image classification model to use state of the art methods to map fine-scale agricultural land use using Earth Observation (EO) and other data in an integrated framework. The rule-base model allows ecologically relevant information to be included in the image classification and will use expertise from Fera agronomists, and EO data to classify crops. The model will also use field parcel polygons from MasterMap or the Customer Land Database (CLAD) to specify the boundaries of fields. The model will be trained and tested using a range of images across England and historical field level data available from the Integrated Administration and Control System (IACS, only available at the detailed level from 1993 – 2004).

The outputs will include a spatially explicit field level map of cropping in case study areas of England (historical and present day) as well as details of the associated uncertainty. Results will be disseminated via the projects final report, through peer reviewed papers and will be presented to Defra and the wider Defra network.
Objective
Aim
This project aims to develop an automated rule-based image classification model to use state of the art methods to map fine-scale agricultural land use using Earth Observation (EO) and other data in an integrated framework. This will produce a data product that describes the crop type of agricultural fields in England at a given time.

Objectives
1) Establish data requirements and statistical methodology to interpret imagery. Suitable data sources will be identified and acquired and the techniques used to overcome issues such as cloud cover and mixed field membership will be established.
2) Design rule sets in remote sensing software to interpret imagery and produce historic map series for IACS decade using techniques developed in objective 1. eCognition software will be used to develop rules to classify imagery to individual crop type and will be supplied to Defra as an appendix to the final report. The historic IACS data will be used to train and validate this classification.
3) Produce a current crop map of a case study area in England using techniques developed in objectives 1 & 2. The rule sets developed with the historic data will be applied to current imagery.
4) Conduct a cost-effectiveness analysis of using this methodology. Fera’s economist will make recommendations for the most cost-effective techniques and efforts.
5) Data product available via government data portals. These data will be in a suitable format that can be accessed by Defra staff.
6) For further value, the project outcomes will be disseminated to all relevant stakeholders. Existing knowledge and expertise will be shared across all partners of the project to develop a collaborative cross-organisation remote sensing team. This will involve staff at Fera, UKSA and the Environment Agency.
Time-Scale and Cost
From: 2013

To: 2014

Cost: £193,586
Contractor / Funded Organisations
F E R A (FERA)
Keywords
Environmental policy