| This paper describes an assessment of the implications of future climate
change for river runoff across the entire world, using six climate models which have been
driven by the SRES emissions scenarios. Streamflow is simulated at a spatial resolution
of 0.5°x0.5° using a macro-scale hydrological model, and summed to produce
total runoff for
almost 1200 catchments. The effects of climate change have been compared with the effects
of natural multi-decadal climatic variability, as determined from a long unforced climate
simulation using HadCM3. By the 2020s, change in runoff due to climate change in
approximately a third of the catchments is less than that due to natural variability but,
by the 2080s, this falls to between 10 and 30%. The climate models produce broadly similar
changes in runoff, with increases in high latitudes, east Africa and south and east Asia,
and decreases in southern and eastern Europe, western Russia, north Africa and the Middle
East, central and southern Africa, much of North America, most of South America, and south
and east Asia. The pattern of change in runoff is largely determined by simulated change
in precipitation, offset by a general increase in evaporation. There is little difference
in the pattern of change between different emissions scenarios (for a given model), and
only by the 2080s is there evidence that the magnitudes of change in runoff vary, with
emissions scenario A1FI producing the greatest change and B1 the smallest. The
inter-annual variability in runoff increases in most catchments due to climate
change — even though the inter-annual variability in precipitation is not changed — and
the frequency of flow below the current 10-year return period minimum annual runoff
increases by a factor of three in Europe and southern Africa and of two across North
America. Across most of the world climate change does not alter the timing of flows
through the year but, in the marginal zone between cool and mild climates, higher
temperatures mean that peak streamflow moves from spring to winter as less winter
precipitation falls as snow. The spatial pattern of changes in the 10-year return period
maximum monthly runoff follows changes in annual runoff.
Keywords: SRES emissions scenarios, climate change impacts on runoff, multi-decadal
variability, macro-scale hydrological model, drought frequency, flood frequency |