[论文解读] Recent applications of machine learning, remote sensing, and iot approaches in yield prediction: a critical review

Fatima Zahra Bassine, Terence Épule Épule+2 more|arXiv (Cornell University)|Jun 7, 2023

Smart Agriculture and AI被引用 8

一句话总结

对远程感知（RS）、机器学习（ML）、云计算和物联网（IoT）在预测作物产量和水资源管理中的应用进行综合性批评性综述，梳理当前方法、植被指数及实际建议。

ABSTRACT

The integration of remote sensing and machine learning in agriculture is transforming the industry by providing insights and predictions through data analysis. This combination leads to improved yield prediction and water management, resulting in increased efficiency, better yields, and more sustainable agricultural practices. Achieving the United Nations' Sustainable Development Goals, especially "zero hunger," requires the investigation of crop yield and precipitation gaps, which can be accomplished through, the usage of artificial intelligence (AI), machine learning (ML), remote sensing (RS), and the internet of things (IoT). By integrating these technologies, a robust agricultural mobile or web application can be developed, providing farmers and decision-makers with valuable information and tools for improving crop management and increasing efficiency. Several studies have investigated these new technologies and their potential for diverse tasks such as crop monitoring, yield prediction, irrigation management, etc. Through a critical review, this paper reviews relevant articles that have used RS, ML, cloud computing, and IoT in crop yield prediction. It reviews the current state-of-the-art in this field by critically evaluating different machine-learning approaches proposed in the literature for crop yield prediction and water management. It provides insights into how these methods can improve decision-making in agricultural production systems. This work will serve as a compendium for those interested in yield prediction in terms of primary literature but, most importantly, what approaches can be used for real-time and robust prediction.

研究动机与目标

• Survey and synthesize how RS, ML, cloud computing, and IoT are used in crop yield prediction and water management.
• Evaluate widely used satellites, vegetation indices, and ML algorithms to assess predictive performance and applicability.
• Identify challenges, best practices, and recommendations for real-time, robust yield forecasting in agriculture.

提出的方法

• Conduct a critical literature review of RS, ML, cloud computing, and IoT applications in yield prediction.
• Summarize satellite data sources (MODIS, Landsat, Sentinel-2, Planet) and their trade-offs.
• Review vegetation indices (NDVI, SAVI, MSAVI, NDWI, EVI, NDRE, CI) and their relevance to yield prediction.
• Compare classic ML algorithms (SVM, RF, ANN, DNN) for regression tasks in yield prediction and water management.
• Discuss ML evaluation metrics (MAE, MSE, RMSE, R-squared, MAPE) and classification metrics (Precision, Recall, F1, Accuracy, Kappa).
• Outline IoT and cloud-based smart farming integration and practical recommendations.

Figure 1: Population growth between 1950 - 2050 Source: Based on Data from Alexandratos and Bruinsma ( 2012 ) .

实验结果

研究问题

• RQ1 What satellite sources and vegetation indices are most used for crop yield prediction and water management?
• RQ2 Which machine learning algorithms are most effective for yield prediction in agricultural settings, and under what conditions?
• RQ3 What are the main advantages and limitations of RS-ML-IoT integrated approaches for real-time yield forecasting?
• RQ4 What practical recommendations and open challenges exist for deploying robust, real-time yield prediction systems in Agriculture 4.0?

主要发现

• Sentinel-2 provides a strong balance of spatial and temporal resolution and is valuable for precision agriculture using vegetation indices.
• Vegetation indices (e.g., NDVI, SAVI, MSAVI, NDWI, EVI, NDRE) effectively reflect plant health, water status, and stress, aiding yield prediction, though NDVI can saturate in dense canopies and Soil/atmospheric effects can bias results.
• ML algorithms (SVM, RF, ANN, DNN) show promise for yield and irrigation optimization, with supervised methods often more accurate but requiring labeled data; deep learning requires large datasets and substantial compute.
• ML-assisted irrigation and input management can reduce water usage while maintaining or improving yields; RS data combined with ML can enhance predictive performance when integrated with weather and biophysical model data.
• IoT devices enable real-time data collection (weather, soil moisture, sensors), supporting proactive management and robust, real-time yield prediction in Agriculture 4.0.
• Evaluation of models relies on standard regression metrics (MAE, MSE, RMSE, R-squared, MAPE) and classification metrics (Precision, Recall, F1, Accuracy, Kappa) to compare predictive performance.

Figure 2: Water requirements for food production ( $Km^{3}/Year$ ). Source: Authors’ conceptualization based on data from Boretti and Rosa ( 2019 ) .