Identification of Aquatic Habitats of <i>Anopheles</i> Mosquito Using Time-series Analysis of Sentinel-1 data through Google Earth Engine

Abstract

Malaria, a severe disease transmitted by Anopheles mosquitoes, presents a substantial public health concern. Since Anopheles mosquitoes thrive in water-rich environments, accurately mapping surface water is essential for assessing malaria risk and managing mosquito populations. This study seeks to identify areas prone to water accumulation, which create habitats conducive to mosquito breeding. Initially, high-risk months were extracted using precipitation and temperature data. Subsequently, the Sentinel-1 Water Index (SWI) was utilized to analyse seven years of monthly Sentinel-1 time-series images via Google Earth Engine (GEE). To enhance our findings, we integrated monthly surface-water maps using a weighted majority voting strategy. Validation efforts included collecting 520 samples, half of which were water bodies identified through field observations and Google Earth Pro, with masks generated using the Segment Anything Model (SAM) algorithm. Object-based evaluation was employed, treating each water body as a distinct entity. The results revealed an overall accuracy of 96.1% and a kappa coefficient of 92.2% in water body detection, underscoring the method's effectiveness. This method, which outperformed other approaches in the domain and machine learning classifiers, is straightforward to implement, rapid, and does not require training data. Furthermore, while field monitoring may be challenging, the findings of this study could aid health authorities in identifying high-risk areas for disease control and prevention efforts.