Rice phenology estimation based on statistical models for time-series SAR data

The actual field survey data from the Rice Department of Thailand´s Ministry of Agriculture over a large area wastes a huge amount of resources. To solve this problem, this paper proposes a new approach to estimate rice phenology using SAR images derived from the RADARSAT-2 data. In this work, we divided the rice phenology into five stages, consisting of seedling, tillering, reproductive, ripening, and harvested stages and applied them to the hidden Markov model (HMM). We also examined three backscattering models, Log-normal, Rayleigh and Gamma distributions, and applied them to the Viterbi algorithm where the rice phenology states are estimated. We were able to estimate the rice phenology, with 80.35 %, 78.57, and 80.35% accuracy, respectively for each of the three backscattering models, when compared to the ground data.