A Robust But Straightforward Phenology-Based Ginger Mapping Algorithm by Using Unique Phenology Features, Time-Series Sentinel-2 Images, and Google Earth Engine

Abstract: Ginger (Zingiber officinale), a commonly used spice, dietary supplement, and medical plant, is expanding rapidly in China due to its increasing demands and price in the international market. However, the existing information on its area and spatial distribution can only be derived from statistical data, and spatially explicit maps of ginger planting areas are unavailable. We proposed a simple but robust ginger planting area mapping algorithm using the time-series Sentinel-2 images and Google Earth Engine (GEE) platform. Changyi city, the hotspot city with the highest ginger production in China, was selected as our study area. Two ginger indices (GIs) were generated based on the four unique phenology features: 1) high reflectance at the start of the season due to white agricultural mulching films, 2) a slower green-up feature in the early season, 3) low reflectance in summer due to black sunshade nets cover, 4) a unique green feature at the end of the growing season. The accuracy assessment based on the ground truth data showed fairly reasonable overall accuracies of 96.7%–97.3%, while the Matthews correlation coefficients (MCC) were 0.871–0.924. We found a rapid ginger expansion from 63.0 km2 in 2019 to 101.5 km2 in 2021 in the study area, by 61% from 2019 to 2021. Moreover, about 78% of ginger land in the study area had crop rotation every two years to boost yield and prevent bacterial diseases in the soil. Our study identified four critical features of ginger and demonstrated the potential of the phenology-based approach, Sentinel-2 data, and GEE platform for ginger mapping. This study is expected to apply to regional or national ginger mapping to provide valuable information for the ginger planting plan and guide the international ginger trade