Environmental Engineering Reference
In-Depth Information
although some (but not all) of these issues are resolvable via dynamical
downscaling, the logistical requirements for so doing make dynamical
downscaling infeasible for analysis of the broad suite of GCM output in the
IPCC AR4 (and upcoming AR5) archives.
We have used as our motivation here the Milly et al. (2005) study, which
analyzed multidecadal runoff from 12 GCMs for mid-21st century as com-
pared with early 20th century runoff (24 separate model runs were analyzed;
however, some models had multiple ensembles, which were averaged). An
important difference, however, is that for reasons articulated in Chapter 2
we performed our analysis in a way that links projected runoff changes with
changes in the global mean temperature. Accordingly, we proceeded as fol-
lows. We extracted IPCC AR4 archived runoff output along with surface air
temperature, precipitation, and evapotranspiration for 23 models for which
global monthly output for (in most cases) 1870 through 2100 was archived.
For each model and emissions scenario, we computed the global mean tem-
perature for each year. We then computed a 30-year moving average of the
global mean temperatures, and from this time series, we determined the year
at which the global mean temperature was larger than for 1985 (1971-2000
mean) by 0.5°C, 1°C, 1.5°C, and 2.5°C. We then computed 30-year moving
averages of runoff for each model, emissions scenario (A2, A1B, and B1),
and grid cell, and overlaid the runoff values for 1985 and the year in which
the moving average global temperature rise was the given amount (0.5, …
2.5) to the U.S. hydrological regions. We then computed runoff changes
as percentages per degree C of global warming, and took the median over
models and global temperature changes of 1°C, 1.5°C, and 2°C. We also
computed equivalent standard errors of the medians (see Table 5.3). Figure
O.2 shows the results. In general, runoff decreases over most of the country,
with the exception of the northeast and the northwest. Closer examination
of similar results (not shown) for precipitation and evapotranspiration indi-
cate that (a) changes in runoff result from a combination of slight increases
in precipitation (in the multi-model median) over most of the country with
the exception of the Southwest, and increasing evapotranspiration over all
but the Southwest (where lower precipitation apparently limits moisture
available for evapotranspiration); (b) the general patterns of changes (runoff
sensitivities per degree C) are roughly constant over the range 1-2°C global
temperature rise; and (c) the patterns of changes and sensitivites per degree
C are similar for the different emissions scenarios (hence justifying pooling
of the results over emissions scenarios).
We also conducted a similar analysis for global runoff changes, which
are also shown in Figure O.2. The figure shows (a) increases (in the multi-
Search WWH ::




Custom Search