University of Twente Student Theses

Login

Assesing Hailstorm Damages in Crops Using Multi-Temporal Remote Sensing Data and Machine Learning Solutions

Pareek, Rishi (2023) Assesing Hailstorm Damages in Crops Using Multi-Temporal Remote Sensing Data and Machine Learning Solutions.

[img] PDF
11MB
Abstract:This research aimed to create a Machine Learning (ML) model using multi-temporal Remote Sensing (RS) data to quantify the crop damage caused by hailstorms, which addresses the inefficiencies of labour-intensive field surveys. Using both Pixel Based Classification (PBC) and Object-Based Classification (OBC) techniques, the model was trained using in-situ data from, Rajasthan India. Key feature inputs for the model are derived from Sentinel-1 (S1) and Sentinel-2 (S2) datasets and include 15 features from S1 and 30 features from S2, including 19 vegetation indices. The study findings indicate that the PBC approach using combined S1 and S2 data resulted in accurate damage class identification with F1 scores of 0.97, 0.93, and 0.94 for low, moderate, and high damage categories. Conversely, using a single dataset, the OBC method with S1 data demonstrated the highest accuracy with F1 scores of 0.96, 0.90, and 0.85 across damage categories. This research addresses the significant societal issues linked with current field-based assessments, presenting a proof of concept for an RS-based crop damage assessment approach. This ML model potentially serves as a robust geoprocessing tool, substantially reducing field survey dependence, hastening the evaluation process, and enhancing transparency in damage assessment categorization, thus benefiting the farming community.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Subject:38 earth sciences, 48 agricultural science
Programme:Spatial Engineering MSc (60962)
Link to this item:https://purl.utwente.nl/essays/96455
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page