Detection of Anthropogenic River Avulsion Using Multi-Sensor Remote Sensing Fusion and Historical Correlation: A Case Study of the Iput River Basin (Belarus)

This study presents a methodological framework for detecting and validating 
anthropogenically induced river avulsion using multi-sensor remote sensing fusion. 
The study area encompasses the Iput-Sozh interfluve in southeastern Belarus, 
where historical industrialization is hypothesized to have caused a significant 
hydrological reorganization. By integrating Sentinel-1 Synthetic Aperture Radar 
(SAR), Sentinel-2 multispectral imagery (specifically the Red-Edge derived NDRE 
index), Landsat 8 thermal data, and FABDEM topographic data through Principal 
Component Analysis (PCA), we identified a 12-km linear paleochannel feature 
previously unmapped in regional hydrological assessments. The identified feature 
exhibits a consistent hydro-geomorphic signature characterized by topographic 
depression (50–100 cm below surrounding terrain), elevated soil moisture indices, 
and reduced surface temperature. Historical document analysis, including 
engineering reports from 1896 detailing a 0.75-m hydraulic head increase at the 
Dobrush Paper Factory dam, provides temporal constraints suggesting channel 
abandonment occurred during the period 1870–1900. Independent validation was 
achieved through analysis of the May 2022 flood event, during which satellite 
imagery confirmed preferential water flow along the mapped paleochannel.
Vegetation succession analysis indicates an abandonment age of approximately
110–180 years, consistent with the historical industrial timeline. This methodology
demonstrates the potential for detecting obscured anthropogenic landscape
modifications through multi-sensor data fusion and historical correlation.