Dear IWP&DC,

Almost every year we see extreme rainfall, and another meteorological record broken. This is not unexpected: as time passes the likelihood increases that the conditions for an extreme event will be met. These extreme events offer new estimates of probable maximum rainfall that should in turn result in a new estimate for probable maximum flood (PMF). In addition, changes in land use and increasing urbanisation suggest that the risk of flood will increase in future.

We have recently produced new estimates of probable maximum precipitation (PMP) for parts of England and the whole of India. These PMP values are higher than existing estimates, and if realised will have catastrophic effects on public health and safety. In the light of this, current practice in dealing with maximum floods should be reconsidered.

•Public safety. India’s worst dam disaster happened in 1979, when the Machhu dam failure led to 1400 deaths. The design flood of the dam was increased several times, but we calculate that the PMF may be more than 1.5 times the latest estimate. In England there have been instances of overtopping but no fatalities since introduction of the Reservoirs Act in 1930. However, lesser floods can and have led to drowning during inundation of the flood plain.

•Public perception. Many people living in central England were caught out this year by a flood said to have a return period of 100 years. But without data extending for 200 years or so this assertion cannot be made with any degree of confidence. Land use changes and increased winter rainfall may reduce the return period to less than 50 years.

•Planning. In England, houses are routinely built in areas of known flood risk. In India, which has a long history of dam failure, there has been a variable level of flood protection because revision of the design standard has been on an ad hoc basis. It is essential that meteorological and hydrological data are collected and provided to those responsible for planning. In England, however, the number of rain gauges is decreasing. In India the number of rain gauges has risen this century, but each has to represent about 600km2, so valuable detail of the spatial variability of the rainfall will not be visible.

•Public participation. Often authorities are unwilling to speak to the public. This is regretted, as there is often a large body of local historical knowledge that could be used to make a better assessment of the flood hazard. Flood data are often collected on an ad hoc basis: greater participation by the public may put the process on a more regular footing.

•Forecasting. Prediction is an important part of flood management that has been neglected. Flood warning systems require accurate hydrometeorological data. Similarly, the timing of flash flood peaks in small catchments will need much better data than are collected at present. It is only after an extreme flood that the weakness of flood warning and emergency planning procedures become apparent. The accuracy of radar-based rainfall data is rather limited during extreme flood events. There is a multiplicity of agencies involved in flood warning arrangements and the likely failure of more than one part of the infrastructure during a flood. Emergency planning schemes should have a greater emphasis on local involvement, rather than using external agencies that may not be able to arrive in time to prevent loss of life.

•Probability. Many people misunderstand the probabilites of flood frequency analysis. To speak of a 100-year event implies to many that the event will occur one in every 100 years. If a one in 2000 year event was swiftly followed by a one in 50 year event, many members of the public would view it as a failure of the authorities to prevent flood damage, rather than as a matter of chance — especially since, for many, the effect in terms of flood damage would be the same.

The authors have initiated an experiment by producing revised estimates of PMP for parts of England and the whole of India, which may bring significant changes in peak floods. How these estimates affect planning of public health and safety will represent the outcome of this experiment.