Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions
Fluvial sediment pulses pose a significant threat to the overall ecological health of river systems. Nonetheless, the scarcity of monitored and published data underscores the importance of devising innovative methods for understanding and measuring how river systems react to the introduction of sedi...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
KeAi Communications Co., Ltd.
2025-06-01
|
| Series: | International Journal of Sediment Research |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1001627925000150 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850275926049292288 |
|---|---|
| author | Abhinav Sharma Celso Castro-Bolinaga Natalie Nelson Aaron Mittelstet |
| author_facet | Abhinav Sharma Celso Castro-Bolinaga Natalie Nelson Aaron Mittelstet |
| author_sort | Abhinav Sharma |
| collection | DOAJ |
| description | Fluvial sediment pulses pose a significant threat to the overall ecological health of river systems. Nonetheless, the scarcity of monitored and published data underscores the importance of devising innovative methods for understanding and measuring how river systems react to the introduction of sediments across the fluvial domain. The objective of this study was to create a modeling framework based on reflectance–turbidity that can be applied in regions with both limited and abundant data. Various combinations of predictor variables, training algorithms including linear regression and additional machine learning methods, and input data availability scenarios were examined to comprehend the factors influencing turbidity prediction on a regional scale. The results indicated that, for Washington state, the random forest algorithm, utilizing a combination of reflectance-based predictors and sediment delivery index (SDI) as predictors, produced the most accurate outcomes (data rich: NSE = 0.54, RSR = 0.68, data scarce: NSE = 0.47, RSR = 0.73). However, when tested on three locations in Washington experiencing sediment pulses, the reflectance–based turbidity prediction model consistently underestimated the peak high and peak low turbidity levels for the Elwha River. The model also exhibited consistent inaccuracies in predicting the initial phase of sediment pulses following the Oso Landslide. Nevertheless, promising results were observed for the Toutle River, downstream to the St. Mt. Helens Volcanic eruption site. Overall, the inclusion of SDI in the model enhanced its efficiency and transferability. By enabling the reconstruction of fluvial sediment pulses in data-scarce regions following dam removals, this integrated approach contributes to advancing our understanding of how rivers respond quantitatively and predictively to these disturbances in sediment supply. |
| format | Article |
| id | doaj-art-e141b709d893473ba7f448176951e6ec |
| institution | OA Journals |
| issn | 1001-6279 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | KeAi Communications Co., Ltd. |
| record_format | Article |
| series | International Journal of Sediment Research |
| spelling | doaj-art-e141b709d893473ba7f448176951e6ec2025-08-20T01:50:30ZengKeAi Communications Co., Ltd.International Journal of Sediment Research1001-62792025-06-0140352353610.1016/j.ijsrc.2025.02.002Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regionsAbhinav Sharma0Celso Castro-Bolinaga1Natalie Nelson2Aaron Mittelstet3Department of Biological and Agricultural Engineering, North Carolina State University, Raleigh 27695, USADepartment of Biological and Agricultural Engineering, North Carolina State University, Raleigh 27695, USA; Corresponding author.Department of Biological and Agricultural Engineering, North Carolina State University, Raleigh 27695, USA; Center for Geospatial Analytics, North Carolina State University, Raleigh 27695, USADepartment of Biological Systems Engineering, University of Nebraska–Lincoln, Lincoln 68583, USAFluvial sediment pulses pose a significant threat to the overall ecological health of river systems. Nonetheless, the scarcity of monitored and published data underscores the importance of devising innovative methods for understanding and measuring how river systems react to the introduction of sediments across the fluvial domain. The objective of this study was to create a modeling framework based on reflectance–turbidity that can be applied in regions with both limited and abundant data. Various combinations of predictor variables, training algorithms including linear regression and additional machine learning methods, and input data availability scenarios were examined to comprehend the factors influencing turbidity prediction on a regional scale. The results indicated that, for Washington state, the random forest algorithm, utilizing a combination of reflectance-based predictors and sediment delivery index (SDI) as predictors, produced the most accurate outcomes (data rich: NSE = 0.54, RSR = 0.68, data scarce: NSE = 0.47, RSR = 0.73). However, when tested on three locations in Washington experiencing sediment pulses, the reflectance–based turbidity prediction model consistently underestimated the peak high and peak low turbidity levels for the Elwha River. The model also exhibited consistent inaccuracies in predicting the initial phase of sediment pulses following the Oso Landslide. Nevertheless, promising results were observed for the Toutle River, downstream to the St. Mt. Helens Volcanic eruption site. Overall, the inclusion of SDI in the model enhanced its efficiency and transferability. By enabling the reconstruction of fluvial sediment pulses in data-scarce regions following dam removals, this integrated approach contributes to advancing our understanding of how rivers respond quantitatively and predictively to these disturbances in sediment supply.http://www.sciencedirect.com/science/article/pii/S1001627925000150Remote sensingSediment transportPhysically informed predictorsSediment pulsesTurbidityHydrology |
| spellingShingle | Abhinav Sharma Celso Castro-Bolinaga Natalie Nelson Aaron Mittelstet Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions International Journal of Sediment Research Remote sensing Sediment transport Physically informed predictors Sediment pulses Turbidity Hydrology |
| title | Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions |
| title_full | Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions |
| title_fullStr | Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions |
| title_full_unstemmed | Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions |
| title_short | Simulating fluvial sediment pulses using remote sensing and machine learning: Development of a modeling framework applicable to data rich and scarce regions |
| title_sort | simulating fluvial sediment pulses using remote sensing and machine learning development of a modeling framework applicable to data rich and scarce regions |
| topic | Remote sensing Sediment transport Physically informed predictors Sediment pulses Turbidity Hydrology |
| url | http://www.sciencedirect.com/science/article/pii/S1001627925000150 |
| work_keys_str_mv | AT abhinavsharma simulatingfluvialsedimentpulsesusingremotesensingandmachinelearningdevelopmentofamodelingframeworkapplicabletodatarichandscarceregions AT celsocastrobolinaga simulatingfluvialsedimentpulsesusingremotesensingandmachinelearningdevelopmentofamodelingframeworkapplicabletodatarichandscarceregions AT natalienelson simulatingfluvialsedimentpulsesusingremotesensingandmachinelearningdevelopmentofamodelingframeworkapplicabletodatarichandscarceregions AT aaronmittelstet simulatingfluvialsedimentpulsesusingremotesensingandmachinelearningdevelopmentofamodelingframeworkapplicabletodatarichandscarceregions |