Authors: Chishan Zhang*, University of Illinois Urbana-Champaign, Chunyuan Diao, University of Illinois Urbana-Champaign
Topics: Remote Sensing, Agricultural Geography
Keywords: deep learning, crop yield, remote sensing
Session Type: Virtual Guided Poster
Start / End Time: 1:30 PM / 2:45 PM
Room: Virtual 53
Presentation File: No File Uploaded
Crop yield prediction is helpful for both crop management and scientific research. It is a fundamental way to study the relationship between crop yield and management and environmental factors. New advances in remote sensing technology have made satellite imagery the best data to conduct this research. The rapid development of machine learning, especially deep learning (DL) technology, has continuously improved the accuracy of satellite-based yield production. In the study, we propose a hybrid DL model composed of convolutional neural networks (CNN) and long short-term memory (LSTM). Based on google earth engine, satellite data and environmental variables, including vapor pressure deficit, soil maps, temperature, precipitation, etc. are used to train the model. The results of experiments show that the hybrid CNN-LSTM model can extract spatio-temporal characteristics of the satellite data and thus achieves the best yield accuracy.