Spatiotemporal distribution of dissolved organic matter in coastal groundwater systems of tidal flats with machine learning–based biogeochemical regulation analysis

Abstract Dissolved organic matter (DOM) is an important variable in the coastal groundwater ecosystem that plays an essential role in regulating the biogeochemistry of tidal flats. The present research uses ML and RS techniques along with field measurements for investigating the distribution of DOM in coastal groundwater and the biogeochemical controls affecting it. RF model and partial dependence plot (PDP) were employed for the purpose of finding out the influential environmental factors. It was found that salinity played a dominant role in the regulation of DOM, with higher DOM values being observed when salinity exceeded 40 psu. In addition, vegetation and water temperature played an important role, with NDVI ranging from 0.25 to 0.45 and temperatures over 22°C favoring microbial decomposition, respectively. The RF model exhibited good predictive capability with an R 2 of 0.91 on the test set data, with an RMSE value of 1.14 mg/L and MAE value of 0.92 mg/L. The spatial analyses showed that DOM values were higher in the southern marsh region and were lower in the northern marine‐affected regions. The hybrid framework can serve as an effective method to identify biogeochemical controls and analyze DOM dynamics in the coastal groundwater ecosystem.