Lessons not learnt

The image of rain over Mumbai at 3-39 p.m., local time, on July 26, captured by the Tropical Rainfall Measuring Mission (TRMM) satellite. The image shows rain intensity as measured by TRMM's sensors, with the areas that received the heaviest rainfall indicated in dark red and lighter rainfall in blue. A dark red area of heavy precipitation sits directly over Mumbai, revealing rain rates as high as 50 mm/hour immediately around the city. TRMM measured the precipitation rates in the centre using the TRMM precipitation radar, the only radar capable of measuring precipitation rate from space. The outer swath has been measured by the TRMM Microwave Imager. (Below) The rainfall accumulations between July 25 and July 27. The dark red area of heavy -

The intensity of the rain that lashed Mumbai could not have been predicted accurately, but the meteorological office could have responded to the situation more effectively if it had a better data collecting mechanism and a more professional approach to work.

ACCORDING to the India Meteorological Department's (IMD) terminology for the classification of rainfall intensity, "Rather Heavy Rain" stands for 3.5 cm to 7.49 cm of rain over a 24 hour period, "Heavy Rain" for 7.5 cm to 12.49 cm and "Very Heavy Rain" for rain in excess of 12.5 cm. Clearly, the terminology is grossly inadequate to classify the unprecedented rain of July 26 in Mumbai. The meteorological station at Santacruz in North Mumbai recorded a whopping 94.4 cm of rain in 24 hours. Some pockets seem to have received even higher rainfall (Table 1). But, given the relative frequencies of rainfall with different intensity during the monsoon period, the terminology is not without logic. Table 2, for example, shows the relative frequencies for different rainfall intensity along the West Coast, where one witnesses spells of heavy rainfall during the monsoon.

Historically, in more than a century of rainfall records of the IMD, the number of occasions Mumbai received more than 20 cm of rainfall in a day is 50; and, the number of occasions with more than 30 cm is only 13 (Table 3). The highest previously recorded day's rainfall in Mumbai is 57.56 cm on July 4, 1974. In fact, the July 26 rain in Mumbai was truly extraordinary even when compared with heaviest rainfall recorded in places such as Cherapunji (Table 4). But pertinently, in Mumbai's case the rainfall (and hence the scale of the weather system that caused it) was extremely localised. The station at Colaba in the southern end of Mumbai recorded only 7.34 cm of rain on the same day.

The extraordinary precipitation seems to have been spread across 20 to 30 km only. In meteorological parlance, processes that extend over such small scales (20 km to 200 km) are known as "mesoscale phenomena", as against "synoptic scale phenomena" (200 km to 2,000 km), on which the routine observational meteorology is based. Usually, "mesoscale phenomena" last for a very short duration (a few hours) while "synoptic scale phenomena" can last for a couple of days.

Given the available synoptic and modelling techniques, forecasting such a rare and extreme, and that too highly localised, event would seem to be nearly impossible. The techniques of synoptic, or short range, forecasting (one to three days ahead) are based on the knowledge of events that occur with relatively high frequency. As Harold Brooks of the United States' National Oceanic and Atmospheric Administration (NOAA) points out: "Essentially, we're looking for ingredients to come together to produce such an extreme event and, in this case, they are unlikely to have been observed often enough to learn from them. We might be able to predict a very heavy (about 30 cm) event, but not the incredible event that occurred."

Even numerical weather prediction models require input data at appropriate spatial resolution (grid points) to be fed into the model as initial conditions. Roughly speaking, to analyse a complete wave of atmospheric disturbance any numerical model would require at least data at five grid points. Hence if the intrinsic resolution of a numerical weather prediction model is, say, 30 km (which would be a good model), the scale at which it can possibly predict processes or the formation of weather systems would be 120 km. Even for very high-resolution mesoscale forecasting models (say 5 km to 10 km resolution), it would be a tough ask to predict a phenomena on such a small scale as across 20 km.

In fact, given the meteorological systems that were developing in the days prior to the event, most models did predict heavy precipitation. But no model, at any forecasting centre of the world, could predict the extremely intense precipitation that occurred. Both the IMD, in its 24-hour forecast using a limited-area regional model, and the National Centre for Medium Range Weather Forecasting (NCMRWF), in its 48-hour forecast using a mesoscale model (with a 38 km resolution), had predicted 8cm to 16 cm of rain over Mumbai up to 8-30 a.m. of July 27.

Some models around the world, like the one at the United Kingdom Meteorological Office, apparently forecast up to 30 cm of rain. Even this, as the frequencies of intense rainfall events show, is of garden variety, and not anything unusual. However, according to Akhilesh Gupta of the NCMRWF, in a re-run of its model, with some improved input data assimilation techniques and changed input parameters (whose details are not known yet), the U.K. Meteorological Office found that its model predicted rainfall up to 80 cm. Notwithstanding this, it would be fair to say at this point of time that, given the paucity of wind and precipitation rate data, even the cause of the event (which is unlike any other in the past) is puzzling to say the least, though some speculations have been made .

CONSIDERING that the event was not amenable to prediction, the question that naturally arises is whether the people of Mumbai could have been forewarned, say a few hours ahead, of the impending disaster. For instance, the first warning, based on the local forecast for Mumbai and suburbs, that the IMD sent out to all the agencies concerned on July 26, first at 1-00 p.m. and repeated thereafter, said: "Rather heavy to heavy rain accompanied by strong gusty winds likely in city and suburbs. Heavy rainfall likely to occur at a few places with very heavy rainfall at isolated places over Konkanpatti-Goa during next 48 hours. Heavy to very heavy rainfall likely to occur at isolated places over madhya [central] Maharashtra and Marathwada during the same period."

Given the classification of the intensity of rainfall, this warning hardly conveys the gravity of the developing situation. But with the observed rainfall at that time (about 1 cm of cumulative rain) and the limited instrumentation that is in place in the region, such a warning would seem natural and would have been considered appropriate by the IMD as well.

The instrumentation for weather-related measurements in Mumbai include manually operated rain gauges at variously located stations, whose data are recorded every three hours and transmitted, two continuous self-recording rain gauges at Colaba and Santacruz meteorological stations, and an X-band weather radar. The radar is basically a wind-and-storm detection instrument which can, at best, be used to detect cloudiness and measure size and depth of clouds.

If an S-band Doppler weather radar or radars (with a range of about 300 km) had been in place, one would have got wind speeds of the rapid convection systems that were continuously feeding high levels of moisture into the clouds in a localised fashion and the high precipitation rates. Such a radar would have been able to show not only how much rain had fallen and the rate at which it had fallen, but also the extent of rain that was approaching the area hours ahead. However, while any such system would have probably warned of a rainfall intensity of 30 cm to 50 cm, a rainfall intensity of above 90 cm could not have been anticipated at all given the history of intense rainfall events in the region. But the event has certainly brought home the urgency to implement the proposed Doppler Weather Radar (DWR) network along coastal India. It would also perhaps call for a rethink on the density of the network along the western coast. Unlike the network on the east for cyclone detection, a fewer number have been proposed along the west.

There was a rapid increase in the rate of precipitation at the Santacruz meteorological station. In the three-hour duration between 2-30 p.m. and 5-30 p.m. on July 26, it poured nearly 39 cm; more rain per hour than a full day's rain in many intense rainfall days during monsoon. Unfortunately, the instrumentation network of the IMD that is in place does not provide real-time data online. While the manually operated rain gauges give data only in three-hourly bins, the self-recording ones are not automated to transmit in real-time. The data has to be retrieved and scrutinised before it can be used in analysis.

However, given the intense rainfall along Goa and coastal Karnataka in the preceding days, and given the sudden cloudburst-like activity in Mumbai, the meteorological officials should have responded to the situation better and obtained the rain gauge data more frequently, say on a half-hourly or hourly basis. The first half-hour in the beginning of the next three-hourly bin (2-30 p.m. to 5-30 p.m.) itself would have indicated the extreme nature of the event. That would have probably sufficed to issue an immediate red alert following the earlier warning. But this was not done and a couple of crucial hours were perhaps lost. In fact, the sad part is that till date, data from the continuous self-recording rain gauges are not available even to the IMD headquarters in Delhi for a proper diagnosis of the event both by the IMD and other researchers across the country.

Of course, instrumentation and data are not the only issues. The entire system - the administration, the civic authorities and the scientific service arms of the government - does not seem to have learnt from the past and respond to calamities effectively. Whether it was the Kandla cyclone of June 1998 or the Orissa supercyclone of October 1999 or the present record-breaking burst of Mumbai rain, there is all round lack of professionalism. The system seems to react only after the event and is never prepared for the worst-case scenario. Radars and instrumentation will, of course, be put in place in due course given the Rs.500 crores IMD upgradation proposal of Kapil Sibal, the Minister of Science and Technology. But, it is learnt that the IMD has not even responded with an action plan yet, three months after the meeting on the matter.

But how could it? The organisation has been functioning without a chief for several months now. The acting chief has not even been given the charge to handle current affairs. A joint secretary from the Indian Administrative Service (IAS) cadre has been made the overall in charge. He is not only in charge of the IMD's operations but a permanent representative at the World Meteorological Organisation (WMO) where scientific matters are discussed. The Minister and the Department of Science and Technology (DST), under which the IMD functions, do not seem to see the urgency to appoint a Director-General for it to function effectively.