This dataset is public for a manuscript under review. The dataset consists of one csv file containing estimated river ice (length) fraction based on images from USGS Landsat satellite missions. Documentation for the Global River Ice Dataset This csv file contains 840,187 records (rows) of river ice length fraction derived from Landsat satellite images. For each river ice fraction, we have the following associated properties: 1. date: The date on which the image was captured (format: YYYY-MM-DD; data type: string) 2. river_ice_fraction: The length fraction of ice cover (format: decimal number; data type: float; range: [0, 1]) 3. cloud_fraction: The length fraction of cloud cover (format: decimal number; data type: float; range: [0, 1]) 4. topo_shadow: Average topographic shadow along river centerline (format: decimal number; data type: float; range: [0, 1] with 0 meaning fully in shadow) 5. LANDSAT_SCENE_ID: The unique Landsat TOA image identifier (data type: string) 6. PATH: The path of the Landsat image (under WRS-2), roughly related to the longitudinal position of the image (data type: integer) 7. ROW: The row of the Landsat image (under WRS-2), roughly related to the latitudinal position of the image (data type: integer) 8. N_river_pixel: Number of total centerline points intersecting with the Landsat image (data type: integer) 9. N_clear_river_pixel: Number of centerline points free from cloud or cloud shadow cover that intersected with the Landsat image (data type: integer) The dataset used for analysis is a subset (No. of records: 407,542) of this dataset with implementing the following three constraints: \(N\_clear\_river\_pixel ≥ 333\) \(topo\_shadow ≥ 0.95\) \(cloud\_fraction ≤0.25\)