doi:10.1016/j.gloplacha.2008.03.001 
Copyright © 2008 Elsevier B.V. All rights reserved.
Climatic and anthropogenic factors affecting river discharge to the global ocean, 1951–2000
J.D. Millimana, 
, 
, K.L. Farnsworthb, 1, P.D. Jonesc, K.H. Xua and L.C. Smithd
aSchool of Marine Science, College of William and Mary, Gloucester Point, VA 23062, USA
bU.S. Geological Survey, Pacific Science Center, Coastal and Marine Geology Program, 400 Natural Bridges Drive, Santa Cruz, CA 95060, USA
cClimatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, UK
dDepartment of Geography, 1255 Bunche Hall, University of California-Los Angeles, CA 90095, USA
Received 24 January 2008;
accepted 10 March 2008.
Available online 25 March 2008.
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
Abstract
During the last half of the 20th century, cumulative annual discharge from 137 representative rivers (watershed areas ranging from 0.3 to 6300 × 103 km2) to the global ocean remained constant, although annual discharge from about one-third of these rivers changed by more than 30%. Discharge trends for many rivers reflected mostly changes in precipitation, primarily in response to short- and longer-term atmospheric–oceanic signals; with the notable exception of the Parana, Mississippi, Niger and Cunene rivers, few of these “normal" rivers experienced significant changes in either discharge or precipitation. Cumulative discharge from many mid-latitude rivers, in contrast, decreased by 60%, reflecting in large part impacts due to damming, irrigation and interbasin water transfers. A number of high-latitude and high-altitude rivers experienced increased discharge despite generally declining precipitation. Poorly constrained meteorological and hydrological data do not seem to explain fully these “excess” rivers; changed seasonality in discharge, decreased storage and/or decreased evapotranspiration also may play important roles.
Keywords: global river discharge; precipitation; irrigation; Arctic; evapotranspiration; dams
 |
Fig. 1. (A) 1915–2004 discharge trend for the San Diego River (southern California; below) and the Pacific Decadal Oscillation (PDO) index (above). The abrupt shift from warm to cold PDO in 1942 coincided with a period of decreased river discharge, the warm shift in 1978 with increased discharge. Although highly variable, mean annual San Diego discharge during the two warm PDO intervals averaged ~ 1.2 m3 s− 1 compared to 0.18 m3 s− 1 during the cold PDO interval 1942–77. The 12 greatest discharge years between 1915 and 2004 occurred during warm PDO, mostly during El Niño years. (B) Five-year running means of Susquehanna River (northeastern North America) discharge and the winter (December–March) North Atlantic Oscillation index, 1892–2004. A close correlation between the two trends occurred between a warm shift in the Atlantic Multidecadal Oscillation (AMO) in 1928 and its next warm shift in1980 (see Trenberth et al., 2007, their Fig. 3.33). Prior to 1928 Susquehanna discharge appears to have been inversely correlated with the NAO; there seems to be no clear correlation between the two after 1980.
Fig. 2. (A) 1951–2000 discharge trends for 137 global rivers, 109 of which are based on ≥ 45 yr of data (Table S1). With few exceptions (Kolyma, Lena, Fraser), < 10% changes are not statistically significant (Table S2). Shaded regions represent water-scarce areas, mean annual runoff < 100 mm/yr. (B) 1951–2000 global precipitation trends based on 0.5-degree grid cells; data from Climatic Research Unit (CRU), University of East Anglia.
Fig. 3. Latitudinal ΔP (solid curves) and ΔQ (dots) trends in the Americas (left), Euro-Africa (center), and Austral-Asia (right). Shaded areas define latitudinal belts in which the two trends diverge considerably.
Fig. 4. 1951–2000 precipitation (P — dashed lines) and runoff (R — solid lines) trends for three rivers that we classify as normal rivers. The Mississippi (A) had statistically significant increases in both P (14%, 99% significance) and R (30%, 99.5%). The Niger (B) experienced significant declines in both P (− 15%, 99.5%) and R (−22%, 99.5%), whereas the Loire (B) showed no statistically significant change in either parameter.
Fig. 5. 1951–2000 precipitation (P — dashed lines) and runoff (R — solid lines) trends for three rivers that we classify as deficit rivers. Precipitation in the Rio Bravo/Rio Grande watershed (A) increased by 22% (95% significance), but runoff decreased by 38% (90% significance). In contrast, the Yellow River (B) had no significant ΔP but − 82% ΔR (99.5% significance). The Colorado (Texas) (C) experienced a + 30% ΔP but no significant change in ΔR.
Fig. 6. 1951–2000 precipitation (P — dashed lines) and runoff (R — solid lines) trends for three rivers that we classify as excess rivers. The Lena River (A) showed a statistically significant decrease in P (− 6%, 90% significance) but a statistically significant increase in ΔR (6%, 90%). The Indigirka River (B) experienced − 13% ΔP (97.5% significance), but ΔR (7%) was not statistically significant. In contrast, the Brahmaputra River (C) showed non-significant ΔP (4%), but a statistically significant (97.5%) ΔR (25%).
 |
Fig. 7. (A) 1951–2000 percentage changes in precipitation (P) and runoff (R) for 58 large rivers. Normal rivers (green dots) represent rivers in which ΔR primarily reflects ΔP. ΔR in deficit rivers (red diamonds) is less than inferred by ΔP, whereas ΔR in excess rivers (blue triangles) is greater than inferred by ΔP. Open symbols represent rivers whose classifications are questionable (Table S2). (B) Flow regulation (total reservoir capacity as percentage of mean annual discharge) and irrigation indices (unitless quotients of irrigated basin areas divided by virgin mean annual discharge as defined by Nilsson et al. (2005) for normal, deficit and excess large. 0 values from Table S2 are plotted as 0.1. (C) Global distribution of normal, deficit and excess watersheds. Large-river (dots) classification from Table S2; small-river classification (open circles) based on ΔR (Table S1) and ΔP estimated from Fig. 1B. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 8. 1951–2000 trends in precipitation (above) and runoff (below) during winter (Dec–March), summer (July–Sept) and transition (April, May, Oct, Nov) months in the Yenisei (left) and Lena (right) watersheds. Precipitation data from CRU. Discharge data for Yenisei River (at Igarka, upstream basin area 94% of total watershed) and Lena River (at Kusur, upstream basin area 97% of total watershed) from GRDC.
Abstract
Purchase PDF (2338 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (0)
Purchase PDF (208 K)
