Monitoring Soy Crops in Brazil with Satellite Imagery in Google Earth Engine

Project: Monitoring 2013/2014 Soy Crop Health in Brazil Cerrado Biome Using Google Earth Engine Methods

Software: Google Earth Engine - Click here to view my JavaScript code and web map on Google Earth Engine for this project.

Objective: I have attended various Google Earth Engine trainings and want to apply the skills that I have learned to different agricultural applications. For this project, I am tracking my progress with learning how to use Google Earth Engine to monitor crop health. Yes, I could do most of this analysis using a desktop GIS software, but I want to improve my Google Earth Engine skills. Therefore, I will be updating this page over time as I learn more about Google Earth Engine and its capabilities.


I created a JavaScript file to examine crop health for the 2013/2014 soy crop area and Cerrado biome in Brazil.

Climate change events put farmlands at increased risk for crop failures due to droughts, floods, and pest outbreaks. To maintain global food security, farmers need an efficient way to monitor their crop areas to maximize crop yields using as little resources as possible.

Various methods have been developed to monitor crop health around the world with remote sensing techniques. However, there are limited methodologies that utilize the tools available on Google Earth Engine to examine crop growth. Google Earth Engine has the capability to batch process large amounts of data at once while also providing a plethora of free satellite imagery that can be accessed at on one platform. The goal of this ongoing project is for me to practice using Google Earth Engine and explore the many possibilities it offers as a tool to study crop health. Brazil is one of the top producers of soy in the world, so I wanted to start here first to study soy phenology.

Study Area: 2013/2014 Soy Area (green pixels) and Cerrado biome (black outline with gray fill)

I used the 2013/2014 soy extent raster as the basis for identifying soy fields in the satellite imagery I added to my analysis.

First, I imported the Landsat 8 imagery. I created an Image Collection that clipped to the study area of interest and filtered for the following soy calendar time periods: planting (October 2013 - December 2013) and flowering (January 2014 - March 2014). I created a cloud mask to select the most cloud free imagery in the given time periods. Then I tried various methods and Landsat 8 and MODIS band combinations to look at the soy crop health during the flowering period.

Landsat 8 - NDVI (January 2014- March 2014)
MODIS EVI 250m 16 day (January 2014- March 2014)
Soy Extent + Landsat 8 - False Color to highlight Agriculture fields (B6, B5, B2) (January 2014- March 2014)
To provide more contextual information about the 2013/2014 soy growing season, I also looked at rainfall and temperature datasets.
Soy Extent + TRMM 3B43: Monthly Precipitation Estimates (January 2014- March 2014)
Soy Extent + MOD08_M3.006 Terra Atmosphere Monthly Global Product - Surface Temperature (January 2014- March 2014)

Future work includes sharing time series charts plotting monthly NDVI, EVI, precipitation, and temperature values for the 2013/2014 soy growing season. I also want to share the masks I created to highlight areas of crop stress or growth based on the climate data pixel values. In addition, I want to compare my results to previous 2013/2014 Brazilian soy crop outlook reports.