Pivotal asservation of Water Quality in Kuwait territorial waters availed in conjunction with machine learning techniques and satellite imagery

Dana M. AlSalloumi, Nayef Z. AlMutairi, and Mohammad A. AlYaqout

Abstract

Seawater pollution has been identified as being the most serious menace to both humankind and aquatic ecosystems. Any regulation or policy ought to be preceded by seawater evaluations. The conventional techniques for determining the quality of saltwater have been demonstrated to be laborious and expensive. On the contrary, using in-situ measurements, earth observation data from satellites can provide thorough, swiftly, and affordable information about water bodies. As a result, this research proposes an alternative strategy for identifying water quality index (WQI) utilizing machine learning techniques that simulate seawater characteristics using in-situ measurements and remotely sensed data. Fuzzy logic was utilized to estimate WQI in terms of six parameters which are chlorophyll, pH, dissolved oxygen (DO), conductivity, salinity, and turbidity. Data records of these parameters were collected over a year (2016-2017) from 15 buoy stations deployed in Kuwait’s territorial waters, five of which were in Kuwait Bay.

Additionally, two different machine learning techniques were used to correlate spectral bands from Landsat-8 with in-situ measurements, which are Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). An appropriate estimate of WQI in Kuwaiti territorial seas was determined by fuzzy logic. Furthermore, it was discovered that GA outperformed PSO in its capacity to observe WQI over the Bay of Kuwait. Based on spatial mapping, it was revealed that there is a pressure on Kuwait bay due to the concentration of anthropogenic activities. For further studies, it is recommended to apply other methods to estimate WQI and compare the results to fuzzy logic results.