Crop Area Mapping in Southern and Central Malawi With Google Earth Engine

Seth Peterson,Greg Husak

doi:10.3389/fclim.2021.693653

Abstract

Agriculture in sub-Saharan Africa consists primarily of smallholder farms of rainfed crops. Historically, satellite data were too coarse to account for the heterogeneity in these landscapes. Sentinel-2 data have improved spectral resolution and much higher spatial resolution (10 m) than previously available satellites with global coverage, such as Landsat or MODIS, making mapping smallholder farms possible. Spectral mixture analysis was used to convert the Sentinel-2 signal into fractions of green vegetation, non-photosynthetic vegetation, soil, and shade endmembers. Very high spatial resolution imagery in Google Earth Pro was used to identify locations of crop and natural vegetation classes, with over 20,000 reference points interpreted. The high temporal resolution of Sentinel-2 (5 days repeat) allows for classification of landcover based on the phenological signal, with natural areas having smoothly varying amounts of photosynthetic vegetation annually, while cropped areas show more abrupt changes, and also the presence of bare soil due to agricultural activity at some point during the year. We summarized the endmember values using monthly medians, extracted values for the reference data points, randomly split them into training and test data sets, and input the training data into the random forests algorithm in Google Earth Engine to map crop area. We divided southern and central Malawi into tiles, and found crop/no crop classification accuracies on the test data for each tile to be between 87 and 93%. The 10 m map of crop area was aggregated to the district level and showed an R2 of 0.74 with ground-based statistics from the Malawi government and 0.79 with a remotely sensed product developed by the USGS.

Highlights

Climate variability—combined with a lack of resources, social and political instability, pest outbreaks, and other contributing factors—have led to food-insecurity events throughout sub-Saharan Africa, compromising the lives and livelihoods of the most vulnerable populations (Devereux, 2009; Samasse et al, 2018; Funk, 2021)
The high spatial resolution allows for fewer mixed pixels in these landscapes characterized by smallholder farms, which result in mosaics of fields, forests, and pastures, often heterogeneously mixed at even the 30 m Landsat scale, and certainly at the 500 m MODIS scale (Ozdogan and Woodcock, 2006; Samasse et al, 2018; Jin et al, 2019; Misra et al, 2020)
non-photosynthetic vegetation (NPV) for the natural area has an inverse trend to green vegetation (GV), the new green leaves are dominant at the beginning of the season so NPV is low, the leaves senesce, and woody shrub material and senesced ground cover become more exposed to the satellite, leading to an increase in NPV

Summary

Introduction

Climate variability—combined with a lack of resources, social and political instability, pest outbreaks, and other contributing factors—have led to food-insecurity events throughout sub-Saharan Africa, compromising the lives and livelihoods of the most vulnerable populations (Devereux, 2009; Samasse et al, 2018; Funk, 2021). Homegrown food production is a function of crop area and crop yield, but these components are difficult to assess because agricultural statistics in sub-Saharan Africa are known to be inaccurate due to poor organization and data analysis (Devereux, 2009; Carletto et al, 2015). They are quite coarse, being reported at the administrative unit level for ground-based statistics, and generally, 300–1,000 m pixels for satellite-based maps (Carletto et al, 2015; Samasse et al, 2018). The use of GEE reduces these requirements because: (1) the data sets are already loaded into GEE; they do not have to be ordered, downloaded, and stored locally, and (2) processing can be done on Google’s server cloud, effectively bringing supercomputing to the average user, for free (Gorelick et al, 2017)

Methods

Results

Conclusion