One of the sources for that LCA looks at this part in particular:
It’s a model of global water usage by crop. In their analysis, they estimate about 30% of cotton to be irrigated and the rest from precipitation. Green water is that made naturally available to plants by rain, blue is water piped in for irrigation, grey is recycled.
A lot of that data is sourced from the other source, an even more detailed study on cotton water usage.
About 53% of the global cotton field is irrigated, producing
73% of the global cotton production (Soth et al., 1999). Irrigated
cotton is mainly grown in the Mediterranean and other warm
climatic regions, where freshwater is already in short supply.
Irrigated cotton is mainly located in dry regions: Egypt,
Uzbekistan, and Pakistan. The province Xinjiang of China is
entirely irrigated, whereas in Pakistan and the North of India a
major portion of the crop water requirements of cotton are
met by supplementary irrigation. As a result, in Pakistan
already 31% of all irrigation water is drawn from ground water
and in China the extensive freshwater use has caused falling
water tables (Soth et al., 1999). Nearly 70% of the world’s cotton
crop production is from China, USA, India, Pakistan and
Uzbekistan (USDA, 2004).
But even water for rainfall isn’t “free”… depending on the area, using that to grow cotton means it isn’t available for other crops (the cotton takes it up and then evaporates it back into the atmosphere). It also creates an effect where water is made unavailable in one country for the benefit of consumers in another country. For example:
The total water footprint of an average US citizen
due to the consumption of cotton products is 135 m3
/year–more than three times the global average–out of which about
half is from the use of external water resources. If all world
citizens would consume cotton products at the US rate, other
factors remaining equal, the global water use would increase
by 5% [from 9800 to 10300 Gm 3 /year], which is quite
substantial given that humanity already uses more than half
of the runoff water that is reasonably accessible (Postel et al.,1996)
Water problems in the major
cotton producing areas of the world cannot be solved without
addressing the global issue that consumers are not being held
responsible for some of the economic costs and ecological
impacts, which remain in the producing areas. The water
footprint shows water use from the consumer’s perspective,
while traditional statistics show water use from the producer’s
perspective. This makes it possible to compare the water
demand for North American or European citizens with the
water demand for people in Africa, India or China. In the
context of equitability and sustainability, this is a more useful
comparison than a comparison between the actual water use
in the USA or Europe with the actual water use in an African or
Asian country, simply because the actual water use tells
something about production but not about consumption.
So while dyeing uses very little water comparatively, the production of cotton does take water (both direct rainfall and irrigation) away from other countries that produce a lot of it.
Cotton consumption is responsible for 2.6% of the global
water use. As a global average, 44% of the water use for cotton
growth and processing is not for serving the domestic market
but for export. This means that–roughly spoken–nearly half of
the water problems in the world related to cotton growth and
processing can be attributed to foreign demand for cotton
products
It’s not free, but like so much else in our lucky lives, much of the costs are outsourced to other countries and poorer people.