Bridging stream hydrology across spatial scales

How much of the Earth’s surface is covered by running water? This is an important yet difficult question for scientists who estimate the sources of sinks of carbon to the atmosphere. 

In our recent paper published in Geophysical Research Letters, we show that rivers exhibit deep regularity in their width-length scaling relationship, useful for more accurately estimating the global extent of rivers and streams (Boyd & Allen, 2025). 

Estimating the river and stream surface area is difficult because while large rivers are easily seen from space, smaller streams are often hidden under forest canopies or are too narrow for satellite imagery. Previous global estimates of river surface area have varied widely, ranging from 485,000 km2 (Downing et al., 2012) to 811,000 km2 (Liu et al., 2022). Not only that but the relative amount of large rivers to small streams is poorly constrained, which is important because small streams are profuse emitters of greenhouse gases due to their close hydrologic and chemical connection with groundwater. 

We tackled this problem by studying rivers and streams in the Mississippi River Basin through a nested basin approach, which involved combining low-resolution satellite observations, high-resolution airborne remote sensing data, to hand-measured surveys of very narrow headwater streams.

Virginia Tech graduate student Carter Boyd and undergraduate student Anabelle Fry surveying headwater streams in Indian Peaks Wilderness, CO.

We found a very consistent mathematical relationship across spatial scales. Specifically, at the local scale, widths follow a predictable log-normal distribution while at the basin scale, widths follow a Pareto distribution. This discovery reconciles a long-standing contradiction where small streams and large rivers appeared to follow different statistical rules. By showing how log-normal distributions stack to form an emergent Pareto curve, we provide a new approach to estimate the abundance of the smallest, unmapped streams. 

Stacked histograms of river widths colored by stream order in the Mississippi Basin (note the log-scaled x-axis). The inset donut chart shows the relative contributions of each stream order to basin total river and stream surface area.

Using this new framework, we estimated the total river surface area of the Mississippi Basin to be 17,828 ± 2,563 km2. Our approach captures a higher abundance of mid-sized rivers than previous methods. For example, the smallest “first order” streams, despite being narrow, are so numerous that they account for about 6.3% of the total river surface area in the basin. 

Because these small streams have higher concentrations of greenhouse gasses, accurately accounting for their surface area is important for accurately representing the global carbon budget and by extension, predicting future changes in Earth’s climate. 

Carter Boyd (left) is a former graduate research assistant in the GRG and George Allen (right) is PI of the GRG. Both are coauthors of the study. 

References

Boyd, C. A., & Allen, G. H. (2025). The widths of rivers and streams across spatial scales: A framework for improving river‐atmosphere biogeochemical exchange estimates. Geophysical Research Letters, 52, e2025GL115713. https://doi.org/10.1029/2025GL115713

Liu, S., Kuhn, C., Amatulli, G., Aho, K., Butman, D. E., Allen, G. H., et al. (2022). The importance of hydrology in routing terrestrial carbon to the atmosphere via global streams and rivers. Proceedings of the National Academy of Sciences, 119(11), e2106322119. https://doi.org/10.1073/pnas.2106322119 

Downing, J. A., Cole, J. J., Duarte, C. M., Middelburg, J. J., Melack, J. M., Prairie, Y. T., et al. (2012). Global abundance and size distribution of streams and rivers. Inland Waters, 2(4), 229–236. https://doi.org/10.5268/IW‐2.4.502