American rivers are transporting more sediment in less time

Understanding how sediment is produced, mobilized, and delivered through river networks is essential for addressing challenges in water quality, infrastructure management, and landscape evolution. Yet, long-term assessments of sediment dynamics have been hindered by sparse sampling that misses the short-lived events responsible for most annual transport. This study develops a deep learning framework that combines high-frequency turbidity sensors and long-term hydrometeorological datasets to reconstruct daily suspended-sediment flux across 175 minimally regulated U.S. catchments from 1985–2023. By leveraging LSTM models and data-driven attribution techniques, the work produces a continental-scale record capable of resolving multi-decadal shifts in both sediment yield and the timing of transport. Results show that many rivers deliver a greater portion of their annual sediment load in shorter, more extreme pulses; the median time required to transport 90% of the load shrank from 69 days to 50 days, with one-third of basins exhibiting increased temporal inequality. To explain these trends, interpretable machine-learning methods were applied to quantify the relative influence of hydroclimatic forcing and land-use disturbance. Analysis of those drivers reveals that deforestation, urban expansion, and intensifying precipitation events collectively drive the observed acceleration and concentration of sediment transport. By reconstructing a detailed sediment history for U.S. rivers, this thesis provides a new basis for understanding how climate and land-use change are reshaping sediment regimes. The findings have direct implications for sediment budgeting, aquatic habitat protection, reservoir and flood-control infrastructure, and the design of best-management practices.

Name thesis:degree_name

Master of Science

Level thesis:degree_level

masters

Discipline thesis:degree_discipline

Civil Engineering

Department dc:contributor.department

Civil and Environmental Engineering

Grantor dc:publisher

Virginia Tech

Year dc:date.issued

2026

Author dc:creator

• Sigdel, Nishchal Nath

Chair dc:contributor.committeechair

• Husic, Admin

Committee members dc:contributor.committeemember

• Saksena, Siddharth
• Kirker, Ashleigh N.

• Suspended sediment concentration
• high-frequency sensors
• temporal inequality
• deep learning

Statement dc:rights

• In Copyright

Licence dc:rights.uri

• http://rightsstatements.org/vocab/InC/1.0/

Language dc:language.iso

en

Dc Identifier Other

vt_gsexam:45443

OAI identifier oai:identifier

oai:vtechworks.lib.vt.edu:10919/140610