Applications of remote sensing and geographic information system on salinity assessment: a case study on Bang Pakong River, Thailand

Abstract

The Bang Pakong River (BPK) has an annual salinity issue. The river is an important source of freshwater supply in the Eastern Economic Corridor (EEC) of Thailand, which has major agricultural and industrial activities. This research aims to understand and assess the past and future salinity distribution in the BPK River by using remote sensing and time series forecasting through water quality and Landsat image time series data. Regarding to remote sensing, this research employs four machine learning algorithms—multilinear regression, decision trees, random forest, and artificial neural networks—by applying reflectance bands of satellite images and measured salinity values. Random forest is the best model with higher R2 and lower RMSE in both original and bootstrapped data, indicating the model of choice for salinity study. It is also noted that the salinity increases annually which is evident from the time sequence salinity maps, in particular, the downstream part has more prominent salinity changes, especially to BPK9.5 (Mueang Chachoengsao District). Compared to two Time Series Analysis—Seasonal Auto-regressive Integrated Moving Average (SARIMA) and Nonlinear Autoregressive Neural Network (NARNET) Model, in which SARIMA model can predict the salinity with lower RMSE. Based on the forecasted salinity values of SARIMA, maximum and minimum salinity of downstream part in February 2024 is 63.57 - 14.01 dS/m, while in the upstream part, the range of 16.94 - 1.44 dS/m is observed. As regard to climate factors, precipitation and the salinity have the inverse relationship with -0.17 of Pearson correlation coefficient. In addition, temperatures of both water and atmosphere have fair correlations: 0.22 (air) and 0.1 (water) with salinity. However, the water level has no relationship with the salinity, in which Pearson correlation value is 0.01 with EC and -0.02 with TDS respectively. Moreover, the suggestions for salinity prevention are discussed in two main parts: (a) Capability within the local community and administrative body and (b) Capability at the national level.