The research, led by Hannah Schwedhelm of the Technical University of Munich together with colleagues at BOKU University in Vienna and hydrosolutions GmbH in Zurich, addresses a stubborn problem. Global datasets describe river networks, flow estimates and soil erosion, but they typically omit smaller rivers, focus on suspended fine sediment, or demand detailed local measurements that simply do not exist in Central Asia. Sediment transport and hydromorphological data for the region are so sparse that assessing morphological processes at the basin scale has been nearly impossible. To fill the gap, the team modified an existing network-scale sediment connectivity model called CASCADE, originally developed for the Mekong basin, so that it could run entirely on openly available geospatial data combined with statistically varied input parameters.

The study area covers the mountainous portions of five countries—Uzbekistan, Tajikistan, Turkmenistan, Kyrgyzstan and Kazakhstan—and focuses on four main catchments: the Syr Darya and Amu Darya, which drain toward the Aral Sea, the Chu-Talas, which terminates in the Kazakh desert, and the endorheic Issyk-Kul basin. Together these basins drain the entire Central Asian mountain range, from the Hindukush in the south through the Pamir and Gissar-Alay in the center to the Tien Shan in the northeast. Runoff in these rivers is strongly controlled by snow and glacier melt, with gauging stations classified as nivo-pluvial, nival, nivo-glacial or glacial-nival. Measured sediment data exist for only 18 stations, where suspended sediment concentrations are available as decadal monthly averages—a thin foundation for any regional assessment.

The technical core of the work lies in how the team rebuilt the model’s inputs. A river network was delineated from the hydrologically conditioned HydroSHEDS digital elevation model at 3-arcsecond resolution and divided into 1-kilometer segments. Because bedload transport occurs mainly during high-flow events, the researchers derived a channel-forming discharge, the Q5 percentile, by correcting mean monthly flows from the global Flo1K dataset with a factor linked to a seasonal variation index calibrated against regional discharge records. River widths came from a modified version of the RivWidthCloud algorithm, retooled in Google Earth Engine to exploit the higher spatial resolution of the Sentinel-2 satellite mission instead of Landsat, allowing narrower channels to be measured. Dam locations were drawn from the global GROD dataset, each barrier assigned a sediment trapping efficiency that reduces the load passed downstream.

With no measured grain-size distributions or roughness values available at the segment scale, the team took an unusual statistical approach. They generated synthetic distributions for five sediment classes—silt, sand, gravel, cobble and boulders—with the finer classes subdivided into fine, medium and coarse fractions, yielding eleven classes in total. For each class, ten evenly spaced median grain diameters were combined with three levels of sorting, producing 330 distinct grain-size distributions. These were paired with seven riverbed roughness values ranging from smooth to very rough, resulting in 2,310 simulation runs per catchment. The outputs were then aggregated into spatially explicit probabilities: the likelihood that each sediment class is transported, entrained or deposited in each 1-kilometer river segment, along with a connectivity index describing how many upstream reaches successfully deliver sediment to a given point.

Validation was a challenge in a region with almost no sediment measurements, so the team relied on two complementary checks. First, they compared modeled transport probabilities for silt and sand against the 18 stations with suspended sediment records; the model returned probabilities between 76 and 94 percent, with mean values of 90 percent for silt and 94 percent for sand, consistent with the observed regular transport of fine material. Second, they ran an internal consistency assessment against high-resolution 2D numerical hydraulic models at four sites, which had been hydraulically calibrated against measured water levels. The comparison showed overall good agreement and reproduced the physically expected pattern of rising transport probability with decreasing grain size, though the modified CASCADE model showed a consistent tendency toward overestimation, with a mean error of roughly 16 percent and a mean bias of 10 percent.

The resulting dataset, freely available for download and integrated into the HydroPlan-CA hydropower decision support system, reveals striking spatial patterns. In the Naryn catchment, a mountainous sub-basin of the Syr Darya used to illustrate the results, silt and sand show high to very high transport probabilities throughout, gravel is mobilized in nearly every segment including small tributaries, and boulders move only in the main river where discharge peaks. Deposition probabilities generally increase downstream, while entrainment dominates the steep tributaries, indicating that mobilized gravel is flushed toward the main stem. Notably, connectivity is highest in the tributaries and lowest in the main river, where growing upstream catchments, local deposition and dams interrupt the sediment cascade—a structural insight that reach-scale measurements alone could never provide.

The authors frame the dataset around three practical application domains. For site comparison, reaches with low transport probabilities may offer engineering stability, but deposition-prone sites risk reservoir siltation, and highly connected reaches are critical for sediment continuity—an inherent trade-off between feasibility and sustainability. For design and operations, high transport probabilities signal the need for reinforced structures, sediment flushing or bypass systems, and desanders to protect turbines. For impact assessment, the data identify where impoundment or flow alteration may trigger riverbed coarsening, siltation or loss of habitat heterogeneity. At the two example sites, Suyok and Shakimardan, the contrast is instructive: coarse boulder and cobble transport dominates at Suyok, while Shakimardan faces fine-sediment deposition but strong upstream connectivity, demanding different mitigation strategies at each.

The team is candid about limitations. The 1-kilometer resolution smooths over local hydraulic variability, the connectivity index is the most sensitive output to input uncertainties, and river width derived from satellite imagery is likely the largest source of error—though the Parker-Klingeman transport formula used is relatively insensitive to width. The brute-force simulation of 330 grain-size distributions is computationally heavy, and where expert knowledge exists the approach could be simplified. The dataset is explicitly a decision-support tool for early-stage, basin-wide planning rather than a substitute for site-specific studies, and it does not prescribe where hydropower should or should not be built. Still, by making sediment processes visible across entire basins for the first time, the study gives Central Asian planners a transparent, consistent basis for weighing energy ambitions against the morphological and ecological health of rivers that millions of people, and countless ecosystems, depend on.

Subject of Research: Sediment transport and connectivity modeling for sustainable hydropower planning in Central Asian river basins

Article Title: Modeling Sediment Transport Potential and Connectivity to Support Sustainable Hydropower Development in Central Asia

Article References: Schwedhelm, H., De Keyser, J., Siegfried, T., Krone, S., Dallmeier, A., Casas-Mulet, R., & Rüther, N. (2026). Modeling Sediment Transport Potential and Connectivity to Support Sustainable Hydropower Development in Central Asia. Environmental Management, 76(9), Article 313. https://doi.org/10.1007/s00267-026-02598-8

Image Credits: AI Generated

DOI: 10.1007/s00267-026-02598-8

Keywords: sediment transport, sediment connectivity, hydropower, Central Asia, CASCADE model, river morphology, bedload, reservoir siltation, Sentinel-2, geomorphology, sustainable development, river ecosystems