Satellites reveal hotspots of global river extent change

Abstract

Rivers are among the most diverse, dynamic, and productive ecosystems on Earth. River flow regimes are constantly changing, but characterizing and understanding such changes have been challenging from a long-term and global perspective. By analyzing water extent variations observed from four-decade Landsat imagery, we here provide a global attribution of the recent changes in river regime to morphological dynamics (e.g., channel shifting and anabranching), expansion induced by new dams, and hydrological signals of widening and narrowing. Morphological dynamics prevailed in ~20% of the global river area. Booming reservoir constructions, mostly skewed in Asia and South America, contributed to ~32% of the river widening. The remaining hydrological signals were characterized by contrasting hotspots, including prominent river widening in alpine and pan-Arctic regions and narrowing in the arid/semi-arid continental interiors, driven by varying trends in climate forcing, cryospheric response to warming, and human water management. Our findings suggest that the recent river extent dynamics diverge based on hydroclimate and socio-economic conditions, and besides reflecting ongoing morphodynamical processes, river extent changes show close connections with external forcings, including climate change and anthropogenic interference.

Fig. 1. The global distribution of different types of river extent changes in the early 21st century: morphological dynamics (Type-M), hydrological signals (Type-H), and new reservoir-type river reaches (Type-R).

a The global map of different types of changes. b, c The areal statistics of different reach types in six continents (NA North America, SA South America, EU Europe, AF Africa, AS Asia, and OC Oceania) and in the 25 mega basins. Zoom-in maps (d–i) exemplify different types of water extent changes on the Occurrence Change Intensity (OCI) map.

Fig. 2. River widening caused by new dams (Type-R).

a The ratio of expanded water extent (areas of significantly increased inundation frequency) to the basin-wide total river area reveals dam-related river widening across specific regions of the world. The bottom insets (from left to right) are four typical giant dams of the world, the Serra de Mesa Dam (Brazil), the Alqueva Dam (Portugal), the Upper Atbara and Setit Dams Complex (Sudan), and the Three Gorges Dam (China). b Statistics of the total expanded river area due to new dams in counties/regions. c Comparison of the increased river area due to dam construction with other types of expansions (Type-H) in each continent.

Fig. 3. Water extent changes from 1984–1999 to 2000–2018 on Type-H rivers (excluding Type-M (morphological dynamics) and Type-R (new reservoir-type rivers).

a The global map of water extent changes composited from the percentage of increase (PI, blue), the percentage of decrease (PD, red), and the percentage of generally stable (PGS, green) at the level-6 basin scale, with transparency of each color ranging from 0% to 100%. b The percentages of different classes (PGS, PD, and PI) aggregated in each continent.

Fig. 4. Water extent changes in 25 mega river basins.

a the percentage of increase (PI). b the percentage of decrease (PD). c the percentage of general stable river areas (PGS).

Fig. 5. Hotspots of river extent changes.

The upper map shows the top four largest hotspots with positive net increase (a–d in blue outlines) and those with negative net decrease (e–h red outlines), overlaid on the map showing the basin-wide relative magnitude of net increase (the areal difference between increase and decrease divided by the total river area). The green circles on the map indicate the locations of the zoom-in maps within the labeled hotspot region.

Fig. 6. The time series of precipitation (P) and evapotranspiration (E) anomaly for the eight river change hotspots.

See locations of the hotspots (labeled as a–h) in Fig. 5. The solid red lines show the fitted linear trend (p-value for the trend significance) of the time series. The dashed lines represent averages in the first epoch (1984–1999) and the latest epoch (2000–2018). The P and E data are averaged from different sources of datasets (see Methods), with the gray shades representing uncertainty according to the inter-model standard deviation.

Fig. 7. Flow cutoff and recovery in the Yellow River.

a The percentage of significant increase class in the Yellow River basin. b The OCI map on the lower reaches. The distribution of gauging stations is shown in red dots in the maps of a, b. c The dried-up distance and duration (no available data in 1984–1986 and 1990) in the Yellow River. d The long-term discharge anomalies measured at six gauging stations along the river, with mean anomalies shown in solid black line. The solid brown line denotes the mean discharge anomalies during 1990–1999 and 2000–2018.

Fig. 8. The distribution of stable river extent and its correlation with nighttime light intensity.

The map shows the percentage of river area with generally stable extent (PGS, %) in each level-6 basin. The top 15 countries/regions with the highest PGS and the lowest PGS among the top 50 counties with the largest river area are shown on the bottom inset. The left inset shows the PGS statistics in relation to nighttime lights for each basin where the population density is larger than 1 person/km².