Development of a New Phenology Algorithm for Fine Mapping of Cropping Intensity in Complex Planting Areas Using Sentinel-2 and Google Earth Engine

Abstract

Cropping intensity is a key indicator for evaluating grain production and intensive use of cropland. Timely and accurately monitoring of cropping intensity is of great significance for ensuring national food security and improving the level of national land management. In this study, we used all Sentinel-2 images on the Google Earth Engine cloud platform, and constructed an improved peak point detection method to extract the cropping intensity of a heterogeneous planting area combined with crop phenology. The crop growth cycle profiles were extracted from the multi-temporal normalized difference vegetation index (NDVI) and land surface water index (LSWI) datasets. Results show that by 2020, the area of single cropping, double cropping, and triple cropping in the Henan Province are 52,236.9 km2, 74,334.1 km2, and 1927.1 km2, respectively; the corresponding producer accuracies are 86.12%, 93.72%, and 91.41%, respectively; the corresponding user accuracies are 88.99%, 92.29%, and 71.26%, respectively. The overall accuracy is 90.95%, and the Kappa coefficient is 0.81. Using the sown area in the statistical yearbook data of cities in the Henan Province to verify the extraction results of this paper, the R2 is 0.9717, and the root mean square error is 1715.9 km2. This study shows that using all the Sentinel-2 data, the phenology algorithm, and cloud computing technology has great potential in producing a high spatio-temporal resolution dataset for crop remote sensing monitoring and agricultural policymaking in complex planting areas.

phenofit: An R package for extracting vegetation phenology from time series remote sensing

“…The triple‐season region is distributed sporadically in northern Henan, where vegetables are planted. Those ratios of single, double and triple growing season regions, as well as their spatial distribution, concur with results from Guo et al (2021), being 40.7%, 57.8% and 1.5% in the same order and same region. Based on Sentinel‐2 high‐resolution imagery, Guo et al (2021) used a more elaborate algorithm to divide growing seasons, which combines NDVI and a land surface water index to better distinguish growing season patterns.…”

“…Those ratios of single, double and triple growing season regions, as well as their spatial distribution, concur with results from Guo et al (2021), being 40.7%, 57.8% and 1.5% in the same order and same region. Based on Sentinel‐2 high‐resolution imagery, Guo et al (2021) used a more elaborate algorithm to divide growing seasons, which combines NDVI and a land surface water index to better distinguish growing season patterns. The similar results confirm that $$$ phenofit $$$ can divide growing season stably and reliably.…”

“…As phenology extraction is more challenging in regions with multiple growing seasons, we selected Henan province (Figure S3), one of the biggest farming provinces in China. In this region, 59.3% of the farming area has more than one growing seasons each calendar year (Guo et al, 2021). (Figure 8).…”

Cropping Patterns of Annual Crops: A Remote Sensing Review

“…Studies by Liu et al [139] and Yan et al [140] used the MODIS VI products to map sequential cropping in China. Other studies based in China were Son et al [141] and Guo et al [142], but they used Sentinel-2 EVI and NDVI multi-temporal data for mapping sequential cropping. In India, SPOT MVC NDVI multitemporal was also used by de Bie et al [143] and Manjunath et al [112] to map single and double cropping.…”

Mapping Crop Distribution Patterns and Changes in China from 2000 to 2015 by Fusing Remote-Sensing, Statistics, and Knowledge-Based Crop Phenology

“…On the other hand, a plot in the triple cropping pattern could only occur in the area where the To detect the cropping seasons for each pixel, we firstly used a harmonic analysis of time series (HANTS) [47] to remove interference signals and restructure the time series curve of EVI. We excluded data after November to avoid fake peaks for winter wheat before it hibernates [47,48]. However, there still might be some fake peaks in the reconstructed EVI time series curve, such as the low peak of winter wheat before it hibernates.…”