Google earth engine based assessment of seasonal spatial distribution of water hyacinth in north Bolgoda lake, Sri Lanka

Abstract

Water hyacinth (WH), a highly invasive aquatic plant is found on every continent except Antarctica. Due to the significant damages caused by these plants to the society, economy and to aquatic ecosystems, it is important to monitor the spatial distribution with respect to seasonal variability to help employ prompt and effective mitigation strategies. In this study, we characterized the seasonality of WH over four years (2019-2022) for the North Bolgoda lake, Sri Lanka. To identify high incidence areas the lake was divided into four sections (A, B, C and D). A pixel based Random Forest (RF) classification utilizing five spectral indices along with the raw spectral bands of Sentinel-2 scenes were used to map the hyacinth coverage with an accuracy greater than 98% by using the GEE (Google Earth Engine) platform. We found that the highest abundance of hyacinth for all sections occurs in the wet season (May – November) and that section A has the maximum WH coverage in all seasons followed by sections C, B and D. This study provides a freely available automated framework for continuous monitoring of WH to help improve environmental decision making.