Using Knowledge-Guided Machine Learning To Assess Patterns of Areal Change in Waterbodies across the Contiguous United States.

Lake and reservoir surface areas are an important proxy for freshwater availability. Advancements in machine learning (ML) techniques and increased accessibility of remote sensing data products have enabled the analysis of waterbody surface area dynamics on broad spatial scales. However, interpreting the ML results remains a challenge. While ML provides important tools for identifying patterns, the resultant models do not include mechanisms. Thus, the "black-box" nature of ML techniques often lacks ecological meaning. Using ML, we characterized temporal patterns in lake and reservoir surface area change from 1984 to 2016 for 103,930 waterbodies in the contiguous United States. We then employed knowledge-guided machine learning (KGML) to classify all waterbodies into seven ecologically interpretable groups representing distinct patterns of surface area change over time. Many waterbodies were classified as having "no change" (43%), whereas the remaining 57% of waterbodies fell into other groups representing both linear and nonlinear patterns. This analysis demonstrates the potential of KGML not only for identifying ecologically relevant patterns of change across time but also for unraveling complex processes that underpin those changes.

Characterization of organic matter by HRMS in surface waters: Effects of chlorination on molecular fingerprints and correlation with DBP formation potential.

In order to understand and minimize the formation of halogenated disinfection by-products (DBPs), it is important to investigate how dissolved organic matter (DOM) contributes to their generation. In the present study, we analysed the DOM profile of water samples from the Barcelona catchment area by high resolution mass spectrometry (HRMS) and we studied the changes after chlorination. Chlorination produced significant changes in the DOM, decreased the average m/z and Kendrick mass defect (KMD) of their spectra and decreased the number and abundance of lignin-like features. The Van Krevelen (VK) fingerprint exhibited several noticeable changes, including the appearance of highly oxidized peaks in the tannin-like region (average O/C, 0.78 ± 0.08), the appearance of features with low H/C and the disappearance of more than half of the lipids-like features. Up to 657 halogenated peaks were generated during sample chlorination, most of which in the condensed hydrocarbons-like and the lignin-like region of the VK diagram. Around 200 features were found to be strongly correlated (ρ ≥ 0.795) to the formation potential of trihalomethanes (THMs) and 5 were correlated with the formation potential of haloacetonitrile (HANs). They all were plotted in the lignin fraction of the VK diagram, but both groups of features exhibited different nitrogen content: those features related to HANs FP had at least one nitrogen atoms in their structures, whilst those related to THMs did not.