Modelling reactive intermediates for predicting photodegradation of organic micropollutants in freshwater environment

在日本从冲绳到北海道采集了50个水库样品和7个湖泊和池塘样品。对水库样品的水化学和吸收光谱特征分析表明，溶解有机质（DOM）的主要变化是分子结构和组成的变化，水库的光生单线态氧量子产率接近以陆源DOM为主的河流和湖泊，但低于以微生物DOM为主的水体。开发了一种新的量子产率估算模型（单模型和多模型）和非线性模型（机器学习模型），可用于高精度地估算水库的量子产率，此外，该模型还可用于估算湖泊、池塘和参考DOM样品的量子产率。

50 reservoir samples and 7 lake and pond samples were taken across Japan from Okinawa to Hokkaido. The characterization of reservoir samples in terms of water chemistry and absorption spectra indicated that major variations in the dissolved organic matter (DOM) were molecular structures and compositions.Quantum yield for photo-production of singlet oxygen of the reservoirs are close to river and lakes dominated by terrestrial DOM but lower than water bodies dominated by microbial DOM.Linear (single and multiple) and non-linear models (machine-learning) were developed and can be used to estimate quantum yield with a high accuracy for reservoirs.In addition, the models can be additionally used for estimating quantum yidl in lakes, ponds, and referenced DOM samples.