Finding Optimal Sensing Periods for Crop Classification
Sensing Time Period for Crop Identification
Published in: International Journal of Remote Sensing (Volume 43)
Publisher: Taylor & Francis
DOI: 10.1080/01431161.2021.1975846
Abstract: Satellite crop identification processes are increasingly being used on a large scale, both to verify the crop and to improve production. As it is necessary to study phenological data over a period of time across a large territory, a lot of storage space is needed to save the satellite images and a lot of calculation time to analyse all this information. Sensing periods are usually established based on subjective expert criteria or previous experience. However, this decision may cause several differences when discriminating crop patterns, besides not guaranteeing good precision. These processes would greatly improve if the appropriate time periods could be found systematically using the minimum number of satellite images in the shortest possible time. In this paper, we propose a new methodology to determine a suitable sensing period for crop identification using Sentinel 2 images, applying hill climbing algorithms to the training sets of neural network models. We have used the method successfully in the 2020 Common Agricultural Policy campaign in the Extremadura region, Spain. The article also describes the use of the method in a case on tobacco detection in this region.
This peer reviewed paper proposes a novel methodology to systematically determine the best time windows for satellite based crop classification: replacing subjective expert criteria with an algorithmic approach.
The Problem
Satellite crop identification requires multi-temporal analysis: you need images from different points in the growing season to distinguish crops by their phenological patterns. But which dates matter most? Traditionally, sensing periods were chosen based on expert intuition or prior experience, with no guarantee of optimality.
More images means more storage and more computation. Fewer images risks missing critical phenological signals. How do you find the minimum set of satellite acquisitions that maximizes classification accuracy?
The Approach
We applied hill climbing heuristic algorithms to the training pipelines of neural network models to search for optimal sensing periods. The method was tested using Sentinel-2 imagery from the 2020 Common Agricultural Policy (CAP) campaign in the Extremadura region, Spain.
The algorithm iteratively evaluates subsets of available acquisition dates, searching for the combination that produces the highest classification accuracy with the fewest images.
Real-World Impact
The methodology was successfully deployed to support the Government of Extremadura in their EU CAP aid payment verification process. By identifying the ideal sensing windows, the system significantly reduced both the data volume and processing time required for large-scale crop monitoring.
The paper also demonstrates the method's application to a specific case study on tobacco detection in the region.
Citation
Published in International Journal of Remote Sensing / Taylor & Francis, September 2021.