Are Arctic rivers speeding up or slowing down? Riverbank positions and migration rates of rivers in Alaska and Arctic Canada (1972-2020)

Abstract:

The pace of Arctic river migration exerts a first-order control on the mobilization timescales for the 1,700 Pg of carbon currently trapped within frozen and thawing permafrost. However, there is no consensus about whether Arctic rivers are responding to regional warming by speeding up or slowing down. Here, we reconstruct migration rates over the period 1972–2020 for Arctic and sub-Arctic rivers spanning 1,500-km of length and a variety of environments. We find that rivers in discontinuous permafrost experienced a systematic acceleration over the last 50 years, whereas rivers in continuous permafrost experienced a systematic slowdown. We identify two competing mechanisms responsible for this bifurcating behavior: a decline in the erosive intensity of river-ice breakup has supported slower migration, whereas thaw of permafrost riverbanks has caused faster migration. Other proposed mechanisms--including Arctic greening and changes in water discharge, water temperature, and riverine sediment loads--are unlikely to be driving the observed trends.

Manuscript citation:

Geyman, E.C. and Lamb, M.P. Are Arctic rivers speeding up or slowing down? Insights from fifty years of satellite observations. In prep/review. 2025.

This dataset consists of 5 parts:

(1) "Arctic_river_timeseries_database.xlsx" - Summary data table listing the n = 20 analyzed rivers, their average migration rates and estimated changes in migration rate over the interval 1972-2020, and various physical and environmental parameters for each river (channel width, latitude and longitude, above-ground biomass density, greening vs. browning trend, floodplain permafrost content, mean annual temperature, etc.)

(2) "Landsat_scene_IDs.xlsx" - Data table listing the scene IDs for the Landsat images used to extract riverbank positions over the period 1972-2020.

(3) "Riverbank_position_shapefiles.zip" - Zip folder containing ESRI shapefiles of the left and right riverbank positions digitized from each Landsat image. The shapefiles are named following the convention: YYYY_MM_DD_lx.shp and YYYY_MM_DD_rx.shp, where YYYY is the year, MM is the month, and DD is the day, and lx and rx represent the left and right riverbanks, respectively. [Note: left and right are based on the convention of looking downstream].  

(4) "Channel_belt_shapefiles.zip" - Zip folder containing ESRI shapefiles of polygons outlining the channel belts for each investigated river reach. 

(5) "MatlabCode.zip" - Zip folder containing Matlab code used to perform the analysis in Geyman & Lamb, 2025 (see citation above).

See the README.txt document for a detailed description of the datasets and metadata.

Final notes:

This dataset builds on the existing dataset of Ielpi et al. (https://doi.org/10.5281/zenodo.7556050), which contains digitized left bank and right bank positions (1972-2020) for n = 10 Arctic river reaches in Alaska and Arctic Canada: the Big River, Kuparuk River, Mackenzie River, Kobuk River, Tanana River, Yukon River, Porcupine River, Kuskokwim River, Stewart River, and Slave River. Here, we build on the existing dataset by adding riverbank positions (1972-2020) for an additional n = 12 river reaches. The combined dataset (n = 22 river reaches) is included here. If using this dataset, please also credit the original authors of the Ielpi et al. dataset.