Anomaly Detection Algorithm for Gas Flaring Investigation: The RXD Performance on Landsat 8/9 and Sentinel-2 Data in Southern Iran

Abstract

Satellite-based techniques are increasingly used to monitor gas flaring (GF) and assess its environmental impacts. This study applies the Reed-Xiaoli Detector (RXD) algorithm to identify GF locations in the Pars Special Economic Energy Zone (PSEEZ) in Iran, one of the world’s largest gas reserves. RXD was first implemented on shortwave infrared (SWIR) images from the OLI sensor on Landsat 8/9, and its performance was then evaluated using MSI data from Sentinel-2A/B. All available cloud-free OLI and MSI SWIR reflectance data from 2016 to 2022 were processed. To improve detection robustness, a new step based on the annual persistence of the RXD metric was introduced, ensuring an optimal balance between sensitivity and accuracy. Ground-truth GF locations from field surveys and very high-resolution satellite imagery were used to assess RXD performance. Results show that RXD, on both OLI and MSI, enhances the identification, mapping and monitoring of gas flaring sites compared with existing satellite GF products using similar spectral bands. The demonstrated reliability of RXD outputs highlights the algorithm as a valuable component of a multi-source GF monitoring framework, supporting decision-making for the sustainable management of industrial activities.

Keywords: gas flaring, PSEEZ, anomaly detection, RXD, daytime shortwave infrared data, OLI˗L8/9, MSI˗S2