Preparation of rice cultivation map based on phenological characteristics using time series of sentinel 1 images

Subject Areas : Agriculture, rangeland, watershed and forestry

sayyad asghari saraskanroud ¹ , hosein sharifi tolaroud ² , Behrouz Sobhani ³

1 - ئئ
2 - Msc student, Mohaghegh Ardabili University, Ardabil, Iran
3 - University of Mohaghegh Ardabili

Received: 2022-04-30 Accepted : 2022-08-21 Published : 2024-09-29

Keywords: Radar backscatering coefficient, polarization, Paddy Rice field, remote sensing,

Abstract :

The purpose of this study is to identify Paddy Rice field and prepare a land use map based on the phenological characteristics of rice plants using the backscatering of radar data in the Google Earth engine Platform. In order to increase the accuracy of backscatering intensity changes, a 2-year time series was selected. Then, initial identification and land classification were performed. First, the relationship between VV polarization backscatering process and phenological cycle of rice plant was investigated. The results of VV polarization backscatering diagram analysis show that in the first stage of rice plant growth due to moisture, flooding and lack of sufficient vegetation, the backscatering rate was lower. In the second stage of rice plant growth, the amount of backscatering is higher due to increased vegetation and water surface coverage. While in the third stage of rice plant growth, due to the ripening of rice plant seeds and also drying of the paddy fields to harvest, the backscatering rate decreases. Then, by managing time periods and using color combination, the types of uses were identified. After initial identification, in order to achieve better results, a user map of the central part of the effort was prepared using the supervised classification method using the most similar algorithm. After preparing the user map, the accuracy of the map was evaluated using ground samples. The overall accuracy and kappa coefficient of this algorithm are 91.57% and 0.75, respectively. The results showed that the use of phenological data reprocessing time series in accordance with phenology in classifications increases the accuracy of classification. The results also show that the use of Sentinel 1 images along with the Google Earth engine Platform will have a high efficiency in monitoring paddy lands in the northern regions due to the presence of clouds.