Optimal Grid-Based Filtering for Crop Phenology Estimation with Sentinel-1 SAR Data

Abstract

In the last decade, suboptimal Bayesian filtering (BF) techniques, such as Extended Kalman Filtering (EKF) and Particle Filtering (PF), have led to great interest for crop phenology monitoring with Synthetic Aperture Radar (SAR) data. In this study, a novel approach, based on the Grid-Based Filter (GBF), is proposed to estimate crop phenology. Here, phenological scales, which consist of a finite number of discrete stages, represent the one-dimensional state space, and hence GBF provides the optimal phenology estimates. Accordingly, contrarily to literature studies based on EKF and PF, no constraints are imposed on the models and the statistical distributions involved. The prediction model is defined by the transition matrix, while Kernel Density Estimation (KDE) is employed to define the observation model. The approach is applied on dense time series of dual-polarization Sentinel-1 (S1) SAR images, collected in four different years, to estimate the BBCH stages of rice crops. Results show that 0.94 ≤ R2 ≤ 0.98, 5.37 ≤ RMSE ≤ 7.9 and 20 ≤ MAE ≤ 33.

Editorial for the Special Issue “Advances of Remote Sensing in the Analysis of the Spatial and Temporal Variability of Land Surface”

“…In Mascolo et al [10], a novel approach, based on the optimal Bayesian Filter, the Grid Based Filter (GBF), was proposed to estimate phenological stages of agricultural crops with SAR data. The optimal GBF is properly employed by considering crop phenology as a discrete variable with a finite number of stages, in accordance with the numerical scales commonly employed (e.g., Biologische Bundesanstalt, Bundessortenamt and CHemische-BBCH).…”

Exploring the effects of crop growth differences on radar vegetation index response and crop height estimation using dynamic monitoring model

“…While obtaining phenological scales through field surveys at the field scale is relatively straightforward, obtaining regular phenological information at the regional scale presents more challenges. The research idea of incorporating a temporal dimension to address the issue of estimating crop growing status is widely adopted in phenology estimation (Mascolo et al, 2021;McNairn et al, 2018;Silva-Perez et al, 2022;Yang et al, 2021), enabling the monitoring of crop growing status over multiple years. While other remote sensing studies are also employing this technique, further research is necessary to understand how to consider the dynamic context in plant height estimation.…”

Crop Phenology Estimation in Rice Fields Using Sentinel-1 GRD SAR Data and Machine Learning-Aided Particle Filtering Approach

“…To achieve continuous estimation of crop lifecycles, the principles of dynamic system theory have been introduced into phenology estimation research. Based on this theory, Kalman Filtering (KF) and Extended Kalman Filtering (EKF) have gradually been used in SAR data for rice crop phenology studies (Mascolo et al,2021). However, due to KF's applicability to linear problems and EKF's suitability for linearizing nonlinear problems (Mascolo et al,2021).…”

“…Based on this theory, Kalman Filtering (KF) and Extended Kalman Filtering (EKF) have gradually been used in SAR data for rice crop phenology studies (Mascolo et al,2021). However, due to KF's applicability to linear problems and EKF's suitability for linearizing nonlinear problems (Mascolo et al,2021). Particle Filtering (PF) is not constrained by model limitations, has begun to be applied in crop phenology estimation.…”