Title

Crop Identification Based on Remote Sensing Data using Machine Learning Approaches for Fresno County, California

Publication Date

1-1-2021

Document Type

Conference Proceeding

Publication Title

Proceedings - IEEE 7th International Conference on Big Data Computing Service and Applications, BigDataService 2021

DOI

10.1109/BigDataService52369.2021.00019

First Page

115

Last Page

124

Abstract

With the advancement of machine learning algorithms, different models are being used in agriculture monitoring and food industries. Due to the significant growth in sensor technologies, a great deal of data has been made available in the public domain. The potential of machine learning models in satellite remote sensing data has earned a significant amount of attention. Agriculture plays a vital role in the United States economy with California being the foremost state in agriculture production. The availability of remote sensing images through open-source platforms, such as Google Earth Engine, has enabled researchers to work on data-driven agriculture mapping. For our research, we chose Fresno County in California as the area of interest, as it has more than a hundred types of crops grown each year as well as an abundance of ground available for surveying and collecting data. Landsat 7 satellite image data from Google Earth Engine has been used to calculate various vegetation indexes such as normalized difference vegetation index (NDVI), normalized difference water index (NDWI), enhanced vegetation index (EVI) and leaf area index (LAI). Cropland Data Layer (CDL), which is the product of National Statistics Service (NASS), has been used for classifying crop types and ground survey data for Fresno County has been used for testing the model. In this paper, we implemented traditional machine learning models like Random Forest, Decision Tree and SVM. We achieved the highest classification accuracy of 95% with a voting classifier ensemble.

Keywords

Agriculture, Cropland Data Layer, EVI, Fresno, Google Earth Engine, LAI, Machine Learning, National Statistics Service, NDVI, NDWI

Department

Applied Data Science; Computer Engineering

COinS