Remote sensing-based modeling and mapping of seasonal water quality dynamics in Vadkert Lake, Hungary

Remote sensing has become increasingly valuable for monitoring inland water quality across space and time. However, detecting key water quality parameters (WQPs) using satellite imagery in small water bodies remains challenging. This study aims to (1) develop regression models for estimating arsenic (As), ammonium (NH 4 + ), chemical oxygen demand (COD), water hardness expressed as calcium oxide equivalent (CaOeq), and total suspended solids (TSS) using Sentinel-2 imagery and in situ measurements from 2019 to 2021 in Vadkert Lake, Hungary; and (2) assess the spatial and seasonal dynamics of these WQPs by applying the models to Sentinel-2 images from four key dates in 2024. The modified normalized difference water index (MNDWI) was applied to isolate water pixels, retaining bands B2 to B8a for their high spatial resolution and relevance. Mean reflectance values around 20 sampling sites were extracted and correlated with measured concentrations of the five WQPs. Stepwise multilinear regression models were developed for As, NH 4 + , and COD, which exhibited the strongest correlations with band reflectance ( R 2 = 0.91–0.99). These models were applied to four seasonal Sentinel-2 images from 2024 to map the spatial and temporal distribution of the WQPs. Results revealed that As levels peaked in summer (76.8 ± 20.7 μg/L) and were spatially uniform, while NH 4 + and COD also peaked in summer (0.2 ± 0.3 mg/L and 7.3 ± 2.01 mg/L, respectively), with elevated values at the southern and eastern lake margins. These findings show that satellite-based seasonal water quality assessment is feasible in small lakes and supports cost-effective environmental management.