A review of Google Earth Engine application in mapping aquaculture ponds

Abstract

Google Earth Engine (GEE) can effectively monitor aquaculture ponds, but it is underutilized. This paper aims to review the application of GEE in mapping aquaculture ponds around the world using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) method. A total of 16 journal articles have been identified since 2019 from the Scopus and Science Direct databases. Most of the studies were conducted in China and United States using the Sentinel-2, Sentinel-1 and Landsat 8 images. Random Forest and Decision Tree are commonly used machine learning classifiers in GEE-based aquaculture ponds mapping studies. In general, some studies reported that GEE can extract the spatial distribution of aquaculture ponds with great overall accuracies, which are more than 90%. Difficult to detect small ponds and misclassification due to similar spectral reflectance are among the limitations reported in previous studies. Future research directions include incorporation of more aquaculture pond extraction techniques and different types of satellite images in GEE.