Abstract
Accurate and timely estimation of crop yield at a small scale is of great significance to food security and harvest management. Recent studies have proven remote sensing is an efficient method for yield estimation and machine learning, especially deep learning, can infer a good prediction by integrating multisource datasets such as satellite data, climate data, soil data, and so on. However, there are some bottleneck challenges to improve accuracy. First, the popular remote sensing data used for yield prediction fall into two major groups—time-series data and constant data. Surprisingly little attention has been devoted to deep learning networks which can integrate the two kinds of data effectively; second, both temporal and spatial features play a role in affecting the yields. But most of the existing approaches employed either convolutional neural network (CNN) or recurrent neural network (RNN). CNN cannot learn temporal patterns, while RNN barely can learn spatial characteristics. This work proposed a novel multilevel deep learning model coupling RNN and CNN to extract both spatial and temporal features. The inputs include both time-series remote sensing data, soil property data, and the model outputs yield. We experimented with the model in U.S. Corn Belt states, and used it to predict corn yield from 2013 to 2016 at the county-level. The results approve the effectiveness and advantages of the proposed approach over the other methods. In the future, the model will be used on other crops such as soybean and winter wheat to assist agricultural decision-making.
Highlights
T HE U.S is the largest corn producer in the world
Kim et al [24] compared several artificial intelligence (AI) models for crop yield modeling, these models include multivariate adaptive regression splines, support vector machine, random forest (RF), extremely randomized trees, artificial neural networks (ANNs), and deep neural network (DNN); the results suggested that the DNN with multiple hidden layers are more powerful to reveal the fundamental nonlinear relationship between predictors and yields
You et al [28] employed convolutional neural network (CNN) and long short-term memory (LSTM) for soybean yield prediction in the U.S, the results proved that CNN and LSTM can outperform traditional methods such as ridge regression and decision tree, the performance of CNN is better than LSTM
Summary
T HE U.S is the largest corn producer in the world. Corn accounts for more than 95% of total grain production in the U.S The production of corn plays a noteworthy role in the economy. The yield varies every year and is affected by natural disasters and socioeconomic reasons. To secure our food security, preventive adaptive policies should be made ahead of time to secure our food security and economy. Manuscript received June 3, 2020; revised July 24, 2020 and August 12, 2020; accepted August 14, 2020. Date of publication August 25, 2020; date of current version September 16, 2020.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
