Proposal for a new customization process for a data-based water quality index using a random forest approach

Environmental Pollution, Год журнала: 2023, Номер 323, С. 121222 - 121222

Опубликована: Фев. 6, 2023

Язык: Английский

---

Identification of driving factors of algal growth in the South-to-North Water Diversion Project by Transformer-based deep learning

Water Biology and Security, Год журнала: 2023, Номер 2(3), С. 100184 - 100184

Опубликована: Апрель 20, 2023

Accurate and credible identification of the drivers algal growth is essential for sustainable utilization scientific management freshwater. In this study, we developed a deep learning-based Transformer model, named Bloomformer-1, end-to-end without needing extensive priori knowledge or prior experiments. The Middle Route South-to-North Water Diversion Project (MRP) was used as study site to demonstrate that Bloomformer-1 exhibited more robust performance (with highest R2, 0.80 0.94, lowest RMSE, 0.22–0.43 ​μg/L) compared four widely traditional machine learning models, namely extra trees regression (ETR), gradient boosting tree (GBRT), support vector (SVR), multiple linear (MLR). addition, had higher interpretability (including transferability understandability) than which meant it trustworthy results could be directly applied real scenarios. Finally, determined total phosphorus (TP) most important driver MRP, especially in Henan section canal, although nitrogen (TN) effect on Hebei section. Based these results, loading controlling whole MRP proposed an control strategy.

Язык: Английский

---

Vertical distribution rules and factors influencing phytoplankton in front of a drinking water reservoir outlet

The Science of The Total Environment, Год журнала: 2023, Номер 902, С. 166512 - 166512

Опубликована: Авг. 22, 2023

Язык: Английский

---

Predicting Daily River Chlorophyll Concentrations at a Continental Scale

Water Resources Research, Год журнала: 2023, Номер 59(11)

Опубликована: Ноя. 1, 2023

Abstract Eutrophication is one of the largest threats to aquatic ecosystems and chlorophyll a measurements are relevant indicators trophic state algal abundance. Many studies have modeled in rivers but model development testing has largely occurred at individual sites which hampers creating generalized models capable making broad‐scale predictions. To address this gap, we compiled large data set concentrations matched other water quality, meteorological, reach characteristic for diverse 82 streams across United States. We used extreme gradient boosting, tree‐based machine learning algorithm, predict daily concentrations. Furthermore, tested several practical considerations models, such as predictions not included training or utility situ quality versus universally available remotely estimated inputs. Predictions were very strongly correlated observations when compared against randomly withheld subset days; however, had lower accuracy applied completely novel from training. Turbidity total nitrogen two most important variables predicting . Although improved estimates identified more during interpretation, using only remote inputs still resulted highly with small bias. Testing many allowed identification common highlighted challenges applying data‐driven new larger spatial scales.

Язык: Английский

---

Enhancement of quality and quantity of woody biomass produced in forests using machine learning algorithms

Biomass and Bioenergy, Год журнала: 2023, Номер 175, С. 106884 - 106884

Опубликована: Июнь 24, 2023

Язык: Английский

---