AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


Monitoring Compliance with the Common Agricultural Policy

Translated title

Monitorering af Overholdelse af den Fælles Europæiske Landbrugspolitik

Author

Term

4. semester

Publication year

2017

Submitted on

Pages

71

Abstract

Dette speciale udvikler et praktisk system til at overvåge, om landmænd overholder de stedbundne krav i EU’s fælles landbrugspolitik (CAP). Systemet samler gratis satellitbilleder fra Landsat, Copernicus (bl.a. Sentinel) og ASTER i en online applikation (https://monitor.trig.dk), der peger på områder med mulig manglende overholdelse. Arbejdet gennemgår først de rumlige CAP-krav og relevante metoder til at overvåge dem. Derefter opbygges en dataløsning, der bruger offentligt tilgængelige portaler til at håndtere billeder og metadata, og applikationen hostes i skyen. For at udvide applikationen til at følge et udsnit af CAP-reglerne testes to tilgange: (1) vegetationsindekser (simple beregninger af lys, planter reflekterer) kombineret med k-means-klyngedannelse (gruppering af lignende billedpunkter) og (2) maskinlæring med random forest (en samling beslutningstræer). I alle tests anvendes Sentinel‑2B-data sammen med markdata fra Danish AgriFish Agency. Resultatet er en online applikation og et varslingssystem til overvågning af landbrugsarealer samt en todelt arbejdsgang, der udpeger mulige regelbrud ved at se på statistiske afvigere, afgrødeklassifikation og heterogenitet (variation inden for marker). Afslutningsvis anbefaler specialet at etablere et åbent, europæisk datasæt med fler-sæson ground truth-prøver af afgrøder og mindre justeringer af den nuværende overvågningsproces. Projektområdet omfatter øerne Lolland og Falster syd for Sjælland.

This thesis builds a practical system to check whether farms meet location-based rules in the EU’s Common Agricultural Policy (CAP). It combines free satellite imagery from Landsat, Copernicus (including Sentinel) and ASTER in an online application (https://monitor.trig.dk) that highlights areas of potential non-compliance. The work first reviews the CAP’s spatial requirements and relevant monitoring methods. It then develops a data solution that uses public portals to manage imagery and metadata, with the application hosted in the cloud. To extend the app to monitor a subset of CAP regulations, two approaches are tested: (1) vegetation indices (simple calculations from plant-reflected light) with k-means clustering (grouping similar pixels) and (2) a machine learning approach using random forest (an ensemble of decision trees). All tests use Sentinel‑2B data together with field data from the Danish AgriFish Agency. The outcome is an online application and alert system for monitoring farmland and a two-step workflow that flags possible rule breaches by examining statistical outliers, crop classification, and heterogeneity (variation within fields). Finally, the thesis recommends creating an open, EU-wide dataset of multi-season crop ground-truth samples and making minor adjustments to the current monitoring workflow. The project area covers the Danish islands of Lolland and Falster, south of Zealand.

[This abstract was generated with the help of AI]