Identifying critical source areas of non-point source pollution to enhance water quality: Integrated SWAT modeling and multi-variable statistical analysis to reveal key variables and thresholds

•The process-based modeling of SWAT was extended by a machine learning algorithm.•The machine learning algorithm based multi-variable statistical analysis was performed.•Land cover percentage of forest and source areas are the identified variables.•The identified variables affect water quality with...

Full description

Saved in:

Bibliographic Details
Published in:	Water research (Oxford) Vol. 253; p. 121286
Main Authors:	Fang, Shubo, Deitch, Matthew J., Gebremicael, Tesfay G., Angelini, Christine, Ortals, Collin J
Format:	Journal Article
Language:	English
Published:	England Elsevier Ltd 01-04-2024
Subjects:	China Environmental Monitoring Humans Land use and land cover Nitrogen - analysis Non-point source pollution Non-Point Source Pollution - analysis Phosphorus - analysis Physical process-based modeling Rivers Self-organizing mapping Soil Water Pollutants, Chemical - analysis Water Quality Watershed management China Watershed management Non-point source pollution Land use and land cover Physical process-based modeling Self-organizing mapping
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	•The process-based modeling of SWAT was extended by a machine learning algorithm.•The machine learning algorithm based multi-variable statistical analysis was performed.•Land cover percentage of forest and source areas are the identified variables.•The identified variables affect water quality with a threshold effect.•The thresholds for forest and source areas percentage are 37.47 % and 20.26 % respectively. By integrating soil and water assessment tool (SWAT) modeling and land use and land cover (LULC) based multi-variable statistical analysis, this study aimed to identify driving factors, potential thresholds, and critical source areas (CSAs) to enhance water quality in southern Alabama and northwest Florida's Choctawhatchee Watershed. The results revealed the significance of forest cover and of the lumped developed areas and cultivated crops (“Source Areas”) in influencing water quality. The stepwise linear regression analysis based on self-organizing maps (SOMs) showed that a negative correlation between forest percent cover and total nitrogen (TN), organic nitrogen (ORGN), and organic phosphorus (ORGP), highlighting the importance of forests in reducing nutrient loads. Conversely, Source Area percentage was positively correlated with total phosphorus (TP) loads, indicating the influence of human activities on TP levels. The receiver operating characteristic (ROC) curve analysis determined thresholds for forest percentage and Source Area percentage as 37.47 % and 20.26 %, respectively. These thresholds serve as important reference points for identifying CSAs. The CSAs identified based on these thresholds covered a relatively small portion (28 %) but contributed 47 % of TN and 50 % of TP of the whole watershed. The study underscores the importance of considering both physical process-based modeling and multi-variable statistical analysis for a comprehensive understanding of watershed management, i.e., the identification of CSAs and the associated variables and their tipping points to maintain water quality. [Display omitted]
Bibliography:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23
ISSN:	0043-1354 1879-2448
DOI:	10.1016/j.watres.2024.121286