# Global flood risk under climate change (part 1): observational evidence

Fig. 1: A view from a plane en route from Bangkok to Chiang Mai in October 2011 ([1])

A beautiful, yet daunting, image of the sunset taken during the 2011 Thailand floods reveals the full extent of one of the world’s largest natural disasters. About 30,000 square kilometres were flooded – an area equivalent to that of Belgium – causing over 800 fatalities. Scientists from UCL suggest that another flood, at least as devastating as this one, can occur in Thailand within the next two decades [2]. Just how concerned should we be that floods of this magnitude can happen again? Can we infer any future trends in flood frequency due to climate change from observations?

This snack series is divided into two parts: Part one of this snack will look at the observational evidence for changes in flood frequency. Part two at modelling problems and how model results can allow us to make a more detailed statement about the direction and magnitude of change in global flood risk under climate change.

According to the International Disaster Database, floods are already the most frequent of all natural disasters – 46% of reported disasters are attributed to flood events. The United Nations estimate that about one in seven people living on this planet are likely to be affected by a 100-year event (see Box 1 for background information on the 100-year flood event).

Box 1: The 1-in-a-100-year flood

It is important to point out that the ‘100-year event’ is not a ﬂood that occurs ‘once every 100 years’ as commonly perceived. The standard hydrological term in fact refers to a ﬂood that has a one percent chance of being exceeded in any given year. This means that e.g. for a river flood, the probability of a flood with a river flow equal to or greater than that of a 100-year flood would be 0.01 or 1%. However, the probabilities accumulate over time. The chance of a certain-size flood occurring during any period can be calculated as:

Pt = 1-(1-Pe)n

where Pt is the probability of occurrence over the entire time period, n, and Pe is the probability of occurrence in any year.

Fig. 2: Probability of a 1-in-a-100-year flood ([3])

As illustrated in Fig. 2, if you live in a 100-year ﬂood zone, there is a greater than 26% chance that you will see at least one 100-year ﬂood over a period of 30 years (and, similarly, more than a 63.4% chance over 100 years).One should bear in mind that since the 100-year flood level is statistically computed using existing data, as more data comes in, the level of the 100-year flood will change.

Flood risk can increase because of many factors …

A number of factors can change flood risk. We can generally distinguish between changes in the climate system (climatic factors) as well as human-induced changes to hydrological systems as explained below. The relative importance of these factors tends to vary with location.

Changes in the climate and atmospheric system concern the characteristics of precipitation (duration, intensity, volume, seasonality, timing, phase – i.e. rain or snow) and temperature (affecting snowmelt, evaporation etc.). For example, more intense rainfall is favoured in the future due to the so-called Clausius-Clapeyron relationship: atmospheric moisture is expected to increase by about 6-7% per degree Celsius warming.

It is also important to note that seasonality and climate variability (e.g. related to the El Nino Southern Oscillation or North Atlantic Oscillation) as well as the frequency and tracks of tropical storms can affect flood risk.

Apart from changes in the climatic system, there are multiple non-climatic factors exacerbating flood hazard. Land-use changes, deforestation, urbanization, the elimination of wetlands and washlands, as well as separation of flood plains from the channel by levees and construction of artificial river banks affect flood hazard.  Many of those “human” factors lead to a reduction of the available water storage capacity and an increased exposure of the population to flood damages. Urbanization not only increases the portion of impervious area such as roads, pavements, or parking lots and leads to a growth in the flood amplitude as water simply cannot seep into the ground. In many regions, people have also been developing and settling in flood-prone areas.

What do flow observations in rivers tell us?

In order to look at trends in flooding, long-term datasets are needed. The flow in rivers, also termed stream flow by hydrologists, is often used as a proxy to investigate potential changes in the hydrological regime because of the availability of longer-term records – but there are a number of problems associated. We should keep the above mentioned factors in mind when looking at flood risk trends from observations.

Intuition would suggest that precipitation trends are visible in stream flow records. Looking at the average yearly streamflow, however, is problematic. The most recent analyses [4] do not support earlier work [5] that reported an increasing trend in global river discharge associated with global warming during the 20th century. With precipitation amount extremes increasing, as well as the possibility of less frequent precipitation events, the effect on average annual streamflow is inconclusive and highly dependent on the region.

The most evident, increasing flood trends appear to be in northern high latitudes, where the largest warming temperature trends have been observed and have led to permafrost thawing and the loss of frozen water. The warmer climate also allows storm systems in the extratropics to transport more water vapour into the higher latitudes.

The picture for other regions, however, is patchy – with varying degrees in the direction of change as well as its significance. In the Amazon basin, an increase of discharge extremes has been observed over recent decades [6]. Other regional studies, for example on Russia [7], show no evidence of a trend in extreme flooding. Interestingly, annual streamflow trends in many regions appear to reflect wetting trends in winter months [8]. However, strong trends of decreasing streamflow were found to be widespread across Europe and Asia in spring and summer months as a balancing factor on average yearly stream flow.

So why can’t we find clear trends in flow records?

River flow processes exhibit a strong natural variability, thus records of less than 30 years are almost certainly too short for detection of climate change. In addition, large floods are extreme events – in order to evaluate changes in the frequency of extreme events (for example a 1,000 year flood) and to detect trends in the intensity of flooding outside of natural variability requires decades, if not centuries, of continuous observations. River flow records are still too short, incomplete and do not cover all parts of the world. Due to financial constraints, several countries have reduced their observation networks. In addition, most large rivers, especially those for which a long-term stream flow record exists, have been impacted by human influences such as dam construction or land use. Streamflow observations from so-called ‘near-natural catchments’ – not impacted by human development – are therefore of paramount importance for detection and attribution studies.

We can infer that there is currently no clear and widespread evidence for observed changes in flooding. Part 2 of this snack series on global flooding under climate change will deal with modelling problems as well as current model results.  There are regional indications that flood risk is changing, however, will modelling gives us the bigger picture? Can we sit back and relax or should we start to worry about the future?

Citations / references:

[1] Photograph by Mith Huang, shared under the Creative Common License on Flickr. Link: http://www.flickr.com/photos/mith17/

[2] Gale, E.L., & M.A. Saunders (2013): The 2011 Thailand flood – climate causes and return periods. Weather 68 (9), pp. 233–237.

[3] Nelson, S.A. (2013): The risk to New Orleans – Present and future. What is the 100-year flood / 100-year storm? Link: http://www.tulane.edu/~sanelson/New_Orleans_and_Hurricanes/New_Orleans_Risk.htm

[4] Labat, D., Aeris, Y. G., Probst, J. L., & J.L. Guyot (2004): Evidence for global runoff increase related to climate warming, Adv. Water Res. 27, pp. 631–642.

[5] Milliman, J. D., Farnsworth, K. L., Jones, P. D., Xu, K. H., & L.C. Smith (2008): Climatic and anthropogenic factors affecting river discharge to the global ocean, 1951–2000. Global Planet Change 62, pp. 187-194.

[6] Espinoza-Villar, J. C., Guyot, J. L., Ronchail, J., Cochonneau, G., Filizola, N.,  Fraizy, P., Labat, D., de Oliveira, E., Ordoñez, J. J., P. Vauchel. (2009): Contrasting regional discharge evolutions in the Amazon basin (1974–2004). Journal of Hydrology 375 (3-4), pp. 297-311

[7] Shiklomanov, A.I., & R.B. Lammers (2007): Record Russian river discharge in 2007 and the limits of analysis. Environ. Res. Lett. 4, 045015.

[8] Stahl, K., Hisdal, H., Hannaford, J., Tallaksen, L. M., van Lanen, H. A. J., Sauquet, E., Demuth, S., Fendekova, M. & J. Jódar (2010): Streamflow trends in Europe: evidence from a dataset of near-natural catchments. Hydrology and Earth System Sciences 14 (12), pp. 2367–2382.

000000

#### Latest posts by Matthias Schmid (see all)

SciSnack Disclaimer: We write in SciSnack to improve our skills in the art of scientific communication. We therefore welcome comments concerning the clarity, focus, language, structure and flow of our articles. We only accept constructive feedback. All comments are manually approved and anything slightly nasty will not be accepted.