Master Thesis: Can you see the wheat?
Date:
as of now
Background:
Detecting the crop area and the type of crop can be a highly valuable information for fertilizer management, water allocation, and crop yield trade. Accurate and reliable information with regard to the changes in crop area of a particular crop type from one agricultural year to another is still a challenging task to perform. However, over 40 years of satellite data have provided enough information to overcome this challenge. It should be noted that by only looking at the satellite images, one might not be able to separate agricultural fields from other land covers in short time, also known as crop classification; therefore, machine learning and deep learning algorithms may come in handy. In the crop classification field of study, these methods are generally categorized into supervised and unsupervised approaches.
Your tasks:
During this study, you are expected to gather a fair amount of information on the different widely used methods of supervised and unsupervised algorithms of crop classification. We will choose the two of the best together, based on the ground truth data we have at hand. You will be working on the data from Bayerische Landesanstalt für Landwirtschaft (LfL) with different stripes of various agricultural crops and/or Technische Universität München. Our goal is to use freely accessible multispectral satellite data, namely, Sentinel-2 and Landsat (8 and 9) for wheat field detection. In this project, you will be in contact with the chair of Precision Agriculture as well as our LfL collaborators.
Requirements:
- Good of knowledge of Google Earth Engine (JavaScript or Python)
- QGIS or ArcGIS
- A huge amount of motivation
Point of contact:
- Ali Mokhtari: ali.mokhtari(at)tum.de
- Prof. Dr. rer. nat. Kang Yu: kang.yu(at)tum.de
- Dürnast 9, 85354 Freising