Effects of climate change on the water cycle

Hervé DOUVILLE, Chercheur au Centre National de la Recherche Météorologique, Co-auteur du dernier rapport du GIEC

I will discuss the anthropogenic disturbances to the water cycle, a topic that is hard to explain in minutes as changes in the water cycle are highly variable in both space, from region to region, and time, from season to season, and are not just caused by human-induced climate change, but also by land use, changes in land use and the abstraction of water resources.

There is a growing anthropization of the water cycle, and climate change is just one of the facets of the anthropogenic disturbances of this cycle.

I will begin by looking at the observed changes and their possible attribution to human activities, before briefly discussing the projections related to changes in the water cycle throughout the 21st century.

Regarding the observed changes in the water cycle, they can be documented based on both in situ observations, using rain gauges or flow measurements, but also, and this for several decades now, based on spatial observations, and notably since the early 21st century, based on a very interesting mission, the GRACE mission, that enables us to document changes in the stocks of water, both surface water and groundwater, which in some regions represents the main stock of water resources.

When we look at the map of the changes recorded since 2002, we can see that the response varies greatly from region to region, with regions where the resource is increasing in blue, and regions where it is substantially decreasing in red, and this is the case in particular of the Aral Sea.

Here again I want to stress that this is not solely or mainly due to climate change, but rather to excessive abstraction due to intensive rice-growing in the region.

Beyond the detection of observed changes, climate models allow us to explore the attribution of causes.

We can force these via both the observed changes in greenhouse gases and anthropogenic aerosols, and we can thus observe in our simulations that throughout the 20th century, the effect of greenhouse gases on the water cycle was largely masked by the effect of anthropogenic aerosols.

These aerosols, for the most part, have a cooling effect on land surfaces, as they reflect a significant portion of incoming solar radiation.

As they are emitted into the northern hemisphere, due to the presence of many industrialized countries in this hemisphere, this has led to a difference in warming between the two hemispheres, which in turn has triggered a shift in the Inter-Tropical Convergence Zone, and in particular monsoon climates, towards the southern hemisphere.

In addition, this reflective effect limits surface warming, limits the increase of evaporation at the surface linked to greenhouse gases, thus limiting aridification, namely in the middle latitudes of the northern hemisphere.

From the end of the 20th century, atmospheric depollution efforts made it possible to limit anthropogenic aerosol emissions, while, sadly, greenhouse gas emissions have continued to increase.

From then on, we can easily see that the attribution of changes in the water cycle to anthropogenic effects was tricky until the end of the 20th century, and only started gaining in importance in the early 21st century.

Thanks to the combined use of both observations and models, these attribution studies allow us to state, in particular, that the increase in rainfall observed in the northern hemisphere's high latitudes is of anthropogenic origin, that the increase in the frequency and intensity of heavy rainfall, on a day-to-day basis, is also accentuated by human activities, and, in addition, as I have shown here, that the frequency and intensity of agricultural droughts in certain parts of the world, here in the IPCC's sense of the word represented by hexagons, has also increased, particularly in western Europe and the Mediterranean region, and this is largely due to human-induced climate change.

Regarding future changes in the water cycle, in order to describe and document these changes, we rely solely on digital modeling, and particularly on simulations guided by different changes envisaged for the concentrations of greenhouse gases and anthropogenic aerosols, this according to different socio-economic scenarios.

If we take, for example, a median scenario, known in the IPCC's jargon as the "SSP2-4.5" scenario, we note, for example, regarding seasonal changes in rainfall, heterogeneous spatial changes from region to region, with an increase in rainfall during most seasons or high latitudes, but, for example, if we focus on Europe, this increase in rainfall is mostly observed in winter or high latitudes, and not so much in summer.

Conversely, in summer, we note a sharp decrease in rainfall across southern Europe.

We therefore see, at European level, an increase in the seasonality of rainfall, with more rain during the rainy season, and less rain during the dry season.

If we look at matters on a global scale, we can see that many subtropical regions that are already semi-arid will experience a decrease in rainfall during most seasons, leading to an annual drop.

This is the case not only for the Mediterranean region, but also for California, for South Africa, for Chile, and for a part of Australia.

All of these semi-arid climates will sadly become even more arid because of the warmer climate, which will obviously cause problems, especially for water resources used in agriculture.

The models also give us access to changes in rainfall patterns on a day-to-day basis.

We can also look at, on the one hand, the number of days without rain, on a regional scale, and on the other, the average intensity of rainfall, that is to say, annual rainfall that is not divided by the number of days in the year, but by the number of rainy days.

We consider a rainy day to be a day when rainfall exceeds 1mm.

Still in the case of a median scenario, we can see that average rainfall intensity will increase in a near-uniform manner around the world.

This will concern moderate rainfall events, but also extreme rainfall events, with an increased risk of flooding associated with such events.

Regarding the number of rain-free days, the response is more spatially varied and more variable from region to region.

We can see areas marked in blue, where the number of rainy days is increasing, and areas marked in red, where it is the number of dry days that are increasing, notably in the Amazon region, but also in the semi-arid regions I mentioned earlier, which are already regions that are relatively vulnerable to water scarcity, and therefore, since the number of rainfall events will drop, we risk having a greater variability in annual rainfall, which will depend on a lower number of events.

This increase in the water cycle's variability from one year to the next is particularly flagrant when looking at tropical regions.

Here, I am showing the changes, on average, of rainfall, in blue, and of runoff, in ocher, according to the level of global warming reached in the projections for the 21st century, regardless of the envisaged scenario.

We can see that average runoff and average rainfall increase in accordance with the level of warming.

What the dotted lines show, however, is that the changes in the interannual variability of these variables, for rainfall but even more so for runoff, increases a lot more than the changes in the average data, meaning that we will once again have much more volatile water resources in regions that rely heavily on water, notably for their own agriculture.

The consequence of this is that our adaptation policies need to be very prudent, they cannot be based solely on changes in average data, but also on changes in the variability over time, and changes from one model to the next.

We have here, for a level of warming of no more than 5°C, error bars that show that beyond the median response given by all of the models, some models show even more dramatic changes in the variability or in the increase in runoff and rainfall in the tropics, with, there again, a risk of flooding, but also, due to increased variability, a risk of drought, which could be increased compared to the median response, which is often commented on in the IPCC's reports.

It is therefore very important for us to have adaptation strategies that are based on as many models as possible, and not just one illustrative model, on all the plausible scenarios in terms of greenhouse gas emissions, and that take into account both average changes and the temporal variability of the resources and water flows.