Comments to pages 108 - 112 in Cool it.
writes (on page 109): " . . our question is how climate change will
affect water stress."
Nearly all climate computer
models project increased precipitation in the future, because global
warming will lead to evaporation of more water, which then will rain
down again somewhere on the globe. The crucial question is: Where will
the extra rain fall - will it fall e.g. over the sea or in land areas
with dense populations that can prosper from the rain?
One may consult the latest IPCC report (2207), WG II, and read on this
subject in its section 3.5.1. That section presents two forecasts of
the projected number of people that will live under water stress. One
is produced by Arnell et al. (2004), and is cited by Lomborg. The other
is produced by Alcamo et al. (2007) and is not cited by Lomborg, even
though it was published before Lomborg finished his work on the water
shortage chapter (cf. his note 812). The reference is: J. Alcamo et al.
(2007): Hydrological Sciences Journal 52(2): 247-275.
The IPCC report presents estimates from the two papers. Alcamo et al.
predict much higher numbers of people under water stress than do Arnell
et al. This is so for the number of people under water stress in the
2050s, but the same is true when looking further into the future
(Alcamo: 2070s, Arnell: 2080s).
In Lomborg´s book, the text relies exclusively on Arnell et al.
when it states that ". . global warming actually reduces the number of people living
in water-stressed areas." But in those scenarios investigated by Alcamo
et al., global warming will
mean more people under water
The explanation why the Arnell et al. study leads to fewer
water-stressed people lies in India. All models show greatly increased
precipitation in northern India and easternmost Pakistan. Some models -
but not all - also show increased precipitation in southern India. In
Arnell´s study, the extra precipitation in India and Pakistand
lifts more than 1 billion people out of water stress. That is a lot -
it is nearly all inhabitants of the Indian subcontinent. In contrast,
Alcamo et al. project that although there will be a greatly increased
precipitation in north India, this will not be sufficient to lift the
inhabitants out of water stress. Here, the same definition of water
stress is applied in both cases, viz. less than 1,000 m³ per
How come that the two studies lead to different results? It is not
because of different estimates of population growth - both studies use
the same IPCC scenarios regarding population growth.
The study of Arnell et al. has some shortcomings. For instance, it does
not include changes in glacier melt, and it does not clearly indicate
the seasonality of the precipitation. The study of Alcamo et al. also
has some shortcomings, but seems more advanced in some respects. Thus,
it includes the effect of rising temperatures on evapotranspiration -
i.e., when it is warmer, crops use more water, and therefore need more
irrigation, and it is more specific concerning water use efficiency.
Some important caveats should be mentioned. Arnell et al. have this
imprtant caveat (p. 50): ". . the increases in runoff generally occur
during high flow seasons, and may not alleviate dry season problems if
this estra water is not stored: the extra water may lead to increased
flooding, rather than reduced water resources stress." Alcamo et al.
are more specific on this point. They state clearly that the extra
precipitation in north India will fall in the rainy season - i.e. when
it rains, it will rain more heavily than before. Not only will average
rainfall increase, there will also be an increase in extremely strong
and damaging rain showers, especially in western India. This will lead
to more frequent events like the flooding in Mumbai in 2005, and to
more floodings of river valleys. Also, Nohara et al. (2006), cited by
Lomborg, show that all extra river flow in the Ganges will be in the
already wet season. Accoring to that study, total river flow at the end
of the 21st century will increase by about 18 % - that is not very much
relative to population growth during the century.
Remember that during the latter half of the 21st century, Himalayan
glaciers will be much reduced, and give less water flow in the large
rivers during the dry season (see Lomborg-errors on glaciers here). As stated
above, this effect is not
included in the study by Arnell et al. So the situation in north India
will be that there will be less water than before in the dry season,
and more water than before in the wet season. It will require a very
large effort to build sufficiently large water reservoirs to have
sufficient water all year round. These reservoirs will have to be very
strongly built and extremely large if they shall prevent floodings due
to catastrophic rain events.
Arnell et al. remember to mention the caveat that although there will
be much more water for the many people in north India, this may
possibly be a curse rather than a blessing, with crops being destroyed
by floods. Lomborg, on the other hand, forgets to mention this caveat.
He also forgets to mention that some of the models in Arnell et al.
indicate reduced rainfall in south India, and he especially forgets to
mention that there are two studies - one indicating a net advantage
overall, and the other indicating a net disadvantage overall. He only
mentions the former.
Page 111 top: " We could bring basic water and sanitation to all of
these people within a decade for about $4 billion annually. ." In the note: "Four billion
dollars per year is the equivalent to $10 billion from 2007 to
2015 at a 5 percent discount rate. "
main text is wrong. As it is is formulated, the reader must understand
that the costs of $4 billion annually are spent within a decade.
However, as explained only in the note, if the costs are spent within a
decade, they will be $10 billion annually. An additional flaw is that
the recalculation of the yearly costs by using a discount rate
introduces a subjectivity in the estimate, because the result is
extremely dependent on the choice of discount rate. Here is chosen 5 %,
which is different from the rate of 3 % chosen in Copenhagen Consensus,
which means that the costs indicated here are not comparable to other
costs presented by Lomborg.