Opinion - Leader Page Articles   

Flood forecasting: possibilities and problems

N. Gopal Raj

Just knowing the rain intensity in different places in a city and the amount of rain that has already fallen can help the authorities identify the areas most likely to face flooding and take timely action.

IN MUMBAI, July 26 began like so many other days during the monsoon. It was raining lightly in some places and merely overcast elsewhere. The India Meteorological Department (IMD) had predicted `rather heavy' to `heavy' rain. Even `very heavy rain,' the IMD's highest classification, means only that more than 13 cm could fall in a day. During every monsoon, there are a few days when Mumbai receives so much rain, typically around 15 cm to 20 cm in a day, that the suburban train services are halted or severely disrupted and low-lying areas near the Mithi river get water-logged. The people who live in the city shrug off these events and the financial heart of India barely misses a beat.

But July 26 was not another of those days for Mumbai. It began raining heavily around noon and was a torrent by 2 o'clock in the afternoon. In 24 hours up to 8.30 am on July 27, Vihar Lake, which feeds the Mithi river, had received over a metre of rain and Santa Cruz got 94 cm; much of it had fallen in just 12 hours. This downpour, the highest ever recorded in the country, brought the city to its knees. The flash floods that resulted claimed a huge toll in lives and property.

Preliminary examination of satellite images suggest that when the torrent of rain occurred, two cloud systems, one from the east and the other from the Arabian Sea, merged over Mumbai, says G.S. Bhat of the Indian Institute of Science's Centre for Atmospheric and Oceanic Sciences (CAOS). Both were deep cloud systems before they merged and continued to be deep even afterwards, a sign that they were fed with vast quantities of moisture continuously for several hours. It is not yet clear why this happened.

Although nowhere near the magnitude of the Mumbai rain of July 26-27, bursts of extremely high rainfall are not unknown on the west coast of India. More than 100 intense rainfall events, where at least one of 30 coastal stations along the west coast recorded more than 20 cm of rain in a single day, were reported between 1951 and 1987, says P.A. Francis, a doctoral student at CAOS. In nearly 80 per cent of these cases, the rain did not exceed 30 cm in a day. But there were 13 events when the rainfall was between 30 cm and 40 cm a day, and five events when it was between 40 cm and 50 cm a day. The rain exceeded 50 cm in one day on just three occasions, two of which were recorded in Mumbai. No event that exceeded 60 cm in a single day was recorded during that period.

Nor are spells of intense rain and consequent flooding uncommon in other parts of the country, including the east coast.

Events that produce excessive rain are difficult to forecast, according to J. Srinivasan, also of CAOS. The clouds that produce such rain do not cover a big area and their development is quite sudden, with local factors, such as terrain, playing an important part. So places such as Colaba and Malabar Hill received less than one-tenth of the rain Vihar Lake and Santa Cruz got on July 26-27.

Even if extreme rainfall events cannot be predicted, countries such as the United States have shown that systems can be put in place to provide at least a few hours warning of flash floods. Such a warning system must continually monitor cloud development and the rainfall that occurs in different localities. That gives an idea of the amount of water that would come as run-off. Combined with data about how much water would be removed by drainage systems, it is possible to secure real-time estimates of the extent of flooding and which areas would be most vulnerable.

U.S. scientists have developed a way of estimating rainfall by using data from a variety of different sensors, including satellite imagery, Doppler weather radar, rain gauges as well as surface and upper air observations. The QPESUMS system (an acronym for Quantitative Precipitation Estimation and Segregation Using Multiple Sensors) is able to provide estimates of rainfall and snowfall rates every five minutes with a spatial resolution of one kilometre.

Real-time monitoring

India should perhaps consider developing a simpler system to monitor rainfall in real-time based on a mix of satellite imagery, Doppler weather radars, and rain gauges, suggests Prof. Srinivasan. All three sensors are already available in the country.

The CCD camera on India's Insat-3A satellite can view an entire hemisphere with a resolution of 1 km. Thus any unusual growth of individual clouds as well as the development of the larger monsoon systems in which they are imbedded can be closely watched. Satellite imagery can also be used to indirectly estimate rainfall.

Doppler weather radars can measure rainfall intensity, show wind speeds and direction, and reveal how clouds are growing and whether their composition is changing for up to about 200 km from the radar installation. The Doppler radars can update such information every five minutes compared to 15 minutes to half an hour for satellites. Radar data was "definitely superior" to satellite estimates for heavy rainfall events, provided good radar coverage was available, says Jonathan Gourley who led the QPESUMS development team. But if radar is not available, then satellite data should be used for warning purposes, he adds. India currently has just four operational Doppler weather radars and all of them have been placed on the east coast to keep watch for cyclones. While three of these Doppler radars were imported, one has been developed by the Indian Space Research Organisation (ISRO) and the technology for its production is being transferred to the public sector Bharat Electronics. It is estimated that 40 such radars would suffice to cover the entire country.

Another way to monitor rainfall is to use rain gauges. ISRO has developed automatic weather stations that can not only measure rain at regular intervals (down to every minute, if required) but also wind speed and direction, temperature, and atmospheric pressure. The data is transmitted over a satellite link to a central office for analysis.

Just knowing the rain intensity that is occurring in different places in a city and the amount of rain that has already fallen can help civic authorities identify the areas most likely to face flooding and take timely action. When the rain becomes very intense, as happened in Mumbai, the time between observation of rainfall and the subsequent inundation becomes shorter. In theory, using a storm-scale model or simply extrapolating rain-rates forward in time could produce predictions up to three hours ahead, according to Dr. Gourley. Such rainfall predictions, however, may not be very accurate.

The U.S. experience has been that there is a big increase in warning time when the rain data feeds into a hydrological model that takes into account how quickly water would drain away and then computes the extent of inundation that is likely. Creating such a hydrological model requires detailed information about land use, terrain information such as slopes, and about drainage systems that are in place, points out Pradeep Majumdar of the Indian Institute of Science. Once this information becomes available, a preliminary hydrological model can be created for an Indian city in about six months to a year, he told The Hindu .

In the Indian situation, what may work best is not a complex warning system but a simple one that the civil engineer in a municipal corporation can understand, believes Kapil Gupta of IIT Mumbai. Year before last and again last year, he had submitted a proposal to the Union Ministry of Human Resources Development asking for Rs.65 lakh to set up an experimental system in Mumbai city for early flood warning. In each locality, Dr. Gupta wants to install rain gauges to monitor rainfall and flow loggers to measure how much water was being removed through the drains. Both sensors (which are to be imported) transmit data in real-time to a central location. In the first stage, the system would simply warn civil authorities which localities were receiving more rain than the drainage system could cope with.

"This is a system that can be upgraded over a period of time," he says. In the next stage, with data about topography incorporated, the system would be able to map the extent of flooding. Dr. Gupta wants 10 rain gauges and 10 flow loggers for his experimental system. A full-fledged system for the whole of Mumbai would require 100 of each of these two sensors, he told The Hindu .

However, even if a warning system had been in place, it may well have had difficulty handling an extreme event like the one that hit Mumbai on July 26-27. That was a once in a hundred years event and most forecasting techniques are based on knowledge of events that occur with relatively high frequency, points out Harold Brooks, head of Mesoscale Applications Group at the National Severe Storms Laboratory in the U.S. "My experience with forecasters and with forecasting systems tells me that, even though it isn't easy, forecasting systems can recognise that this is a big event, but nobody can believe the biggest events until it's too late," he told this correspondent by email. On July 26, with radar and QPESUMS, it would have possible to estimate the rain that had fallen and note that the radar still showed a lot of rain approaching the area. This could have given a message of "it's going to stay bad for the next few hours," he says.

Printer friendly page  
Send this article to Friends by E-Mail

Stories in this Section

Mamata Banerjee's unending tantrums
Coming to terms with the oil shock
Flood forecasting: possibilities and problems
Team Manmohan at work, at last
Learning to live with the tiger
Joint declarations and `secret clauses'
A lifetime passion for politics
Our human rights define us
No solution this
Geelani acquittal
Improper behaviour
Employment guarantee

Copyright © 2005, The Hindu. Republication or redissemination of the contents of this screen are expressly prohibited without the written consent of The Hindu