Monitoring of crop growth stages using Sentinel-1 synthetic aperture radar data

Abstract

Importance of the work: Given the ever-increasing scientific and practical interest in the implementation of satellite data from synthetic aperture radar (SAR) sensors in agriculture, it may be useful for successful crop monitoring to investigate the effects on the total backscatter signal of the biophysical characteristics of the vegetation, such as crop density, stem height and leaf arrangement.
Objectives: To monitor rice, maize, cassava and sugarcane for physical growth characteristics to develop an efficient production process based on capability evaluation of Sentinel-1 products using the mean change of backscatter of polarization, such as VV, VH, VV/VH and the radar vegetation index (RVI).
Materials & Methods: Sigma nought values were customized from the start of cultivation until harvest due to their sensitivity to soil moisture, leaf canopy and stem elongation. Time-series data were used from Sentinel-1 between January 2019 and August 2020. The level-1 ground range detection product resolutions were 20 m × 22 m (range by azimuth) with 10 m × 10 m pixel spacing. Sampling (100 points per crop) was used to establish the research area field boundary. Coverage was in Nakhon Ratchasima (15°N,102°E) and Khonkaen (16°N,102°E) provinces, Thailand.
Results: Statistical analysis using the t test identified the same crop growth stages of cassava, sugarcane and maize using VH polarization. The VV polarization was optimized for rice and cassava, while VV/VH was suitable for rice and maize. RVI could be applied to rice, maize and sugarcane. Finally, VV/VH for rice and VV/VH and RVI for maize could identify the different growth stages, while cassava and sugarcane had no identifiable polarization.
Main finding: The average sigma nought (in decibels) values of polarization from SAR sensors are appropriate to monitor and detect the physical growth of some economic crops in Thailand.