Indian cities flood because their stormwater drains were designed for far less rain than now falls in a single burst, and because the land that used to soak up the excess has been paved and built on. India's 1993 sewerage manual assumed 12 to 20 mm of rain per hour for every city and Mumbai's century-old drains carried 25 mm/h, while monsoon thunderstorms routinely deliver 50 to 70 mm/h. The NDMA estimates that urbanisation raises flood peaks 1.8 to 8 times and flood volumes up to 6 times, so the same storm that a green city absorbed now overruns a concrete one in minutes.
- India's 1993 sewerage manual sized drains for 12 to 20 mm/h; thunderstorms deliver 50 to 70.
- Peak flow follows Q = CiA/360: double the paving and the intensity, and the drain needs 4× capacity.
- Bengaluru's paved area grew 1,028% from 1973 to 2017 while wetlands fell 79% (IISc).
- An IMD study of 12 cities found no systematic trend in heavy-rain days, 1969–2010: the ground changed more than the sky.
- Floods return to the same spots because water still flows to old lake beds and valleys.
The forum version of the question is blunter: why does a road go under water after ten minutes of rain? The answer is arithmetic. A drain is sized for a design rainfall intensity and a share of rain that runs off the surface. When the intensity doubles and the runoff share doubles too, the flow arriving at the drain quadruples.
This post walks through that arithmetic with India's own design numbers, the land-use data from Bengaluru, what the rainfall records say about whether storms are getting heavier, and what the 2025-26 floods in Bengaluru, Mumbai and Delhi showed.
What rainfall are Indian city drains designed for?
Most Indian urban drains were designed for 12 to 25 mm of rain per hour, a fraction of the intensity of a monsoon cloudburst. The NDMA's 2010 urban flooding guidelines record that the Ministry of Urban Development's 1993 Manual on Sewerage gave a single design intensity of 12 to 20 mm/h for all cities, ignoring how differently it rains in Mumbai and Jaipur.
The guidelines warned that any system built to those values will flood whenever rain exceeds 20 mm/h.
City by city
| Network | Design intensity | Where recorded |
|---|---|---|
| Ahmedabad drainage | about 12.5 mm/h | NDMA (2010) |
| Chennai drains | 25 mm/h | IRC SP-50 (1999), via NDMA |
| Mumbai original network, over a century old | 25 mm/h | NDMA (2010) |
| Mumbai new drains | 50 mm/h | IRC SP-50 (1999), via NDMA |
Against those, the same document notes that a thunderstorm dropping 15 mm in 15 minutes is already 60 mm/h, and that severe thunderstorms anywhere produce 50 to 70 mm/h.
See the data as a table
| mm per hour (mm/h) | |
|---|---|
| Ahmedabad drain design | 12.5 |
| 1993 manual, upper value | 20 |
| Chennai drain design | 25 |
| Mumbai old drains | 25 |
| Mumbai redesign (BRIMSTOWAD) | 50 |
| Severe thunderstorm, upper | 70 |
| Mumbai 26 Jul 2005, 12-hour average | 72.9 |
| Bengaluru 24 Sep 2010, about 1 hour | 114.1 |
Source: NDMA National Guidelines: Management of Urban Flooding (2010); Singh, Khole and Rase, MAUSAM (2015) · Mumbai 2005: 875 mm in the 12 hours from 14:30, averaged. Bengaluru 2010: 114.1 mm recorded in a thundershower lasting about an hour.
Mumbai's 26 July 2005 flood is the case study India's own guidelines use. From 14:30 that afternoon the rain gauge at Santacruz recorded 875 mm in 12 hours, part of 944 mm in 24 hours: an average of about 73 mm an hour falling on a drainage network built for 25. At least 479 people died. Yet Colaba, 22 km to the south, recorded only 72 mm the same day. The Chitale Committee that reviewed the disaster found the Mithi river floodplain obstructed, and the NDMA later noted that only 3 of the city's 105 sea outfalls had flood gates to keep the tide out.
NDMA: National Guidelines, Management of Urban Flooding (2010)The Colaba reading shows how local extreme rain can be, and why a single rain gauge per city is not enough to design drains. After the Chitale Committee review, Mumbai began redesigning its stormwater network (the Brihanmumbai Storm Water Disposal project, BRIMSTOWAD) for 50 mm/h.

How does concrete turn rain into a flood?
Concrete raises the runoff coefficient, the share of rainfall that flows off a surface instead of soaking in. Engineers size drains with the rational method, where peak flow is proportional to the runoff coefficient \(C\), the rainfall intensity \(i\) and the catchment area \(A\). Double \(C\) and double \(i\), and the drain must carry four times the water.
Here \(Q\) is peak flow in cubic metres per second, \(i\) is in mm/h and \(A\) is in hectares. The NDMA guidelines tell cities to design all future drains with a runoff coefficient of up to 0.95, which means assuming almost every drop runs off. A worked example for one square kilometre (100 ha) of city:
- Old design. A mixed, partly green neighbourhood with \(C = 0.5\), drains sized for 25 mm/h: \( Q = 0.5 \times 25 \times 100 / 360 = 3.5 \) m³/s.
- Same storm, paved city. \(C = 0.9\) after roofs and roads replace gardens and plots: \( Q = 0.9 \times 25 \times 100 / 360 = 6.3 \) m³/s. The drain is already 1.8 times over capacity.
- A real thunderstorm. \(C = 0.9\), \(i = 60\) mm/h (15 mm in 15 minutes): \( Q = 0.9 \times 60 \times 100 / 360 = 15 \) m³/s, 4.3 times the design flow.
The excess has to go somewhere, and it goes onto the road, which is usually the lowest continuous surface. That is why streets flood within ten minutes of a downpour: the drain fills in the first few minutes and everything after that is surface water.
Is your neighbourhood drain overloaded?
What happened to the lakes and channels that used to hold the water?
They were built over, narrowed or cut off. Bengaluru is the clearest case because it was designed around water: a chain of man-made tanks (kere) on ridges and valleys, connected by stormwater channels called rajakaluves, so that overflow from one lake fed the next.
What IISc measured
Research by T.V. Ramachandra's group at IISc, reported in 2017, found large changes between 1973 and 2017:
- paved surface up 1,028%;
- vegetation down 88%;
- wetlands down 79%;
- about 78% of the city's surface paved by 2017.
The number of lakes fell from about 285 in the early 1970s to 194 in 2006, and a rajakaluve at Jakkasandra had lost about half its width between 1908 and 2017.
The national pattern
The NDMA guidelines describe the same pattern nationally: new housing in low-lying areas, encroachment over drainage channels, sometimes homes built on top of nullahs, and sewage discharged into stormwater drains, which silts them and shrinks their width. Its prescription is blunt:
- pre-monsoon desilting finished by 31 March every year;
- encroachments on drains and floodplains removed (with alternative housing for the poor);
- rainwater harvesting in every building;
- drainage planned by catchment rather than by municipal ward.
| City and event | What fell | What the record shows about drainage |
|---|---|---|
| Mumbai, 26 July 2005 | 944 mm in 24 h at Santacruz | Drains built for 25 mm/h; Mithi river floodplain obstructed (Chitale Committee) |
| Ahmedabad, 2001 | 508 mm in 13 h | Drainage designed for about 12.5 mm/h |
| Bengaluru, 24 September 2010 | 114.1 mm in about an hour | A record thundershower that the city's drains could not take |
| Bengaluru, 18 to 19 May 2025 | 105.5 mm in 24 h (IMD) | About 500 homes flooded; Sai Layout and Silk Board under water; officials cited encroachment and silt |
| Mumbai, 16 August 2025 | 244 mm in 24 h at Santacruz | Highest single August day in five years; Colaba got 83 mm |
| Delhi, 8 August 2026 | 127 mm at Safdarjung by 8:30 am, 204.1 mm for August by evening | About 90% of the month's normal in eight days; Kapashera, MB Road and metro entrances flooded |
Sources: NDMA (2010), Singh et al. (2015), The Federal, All India Radio and Business Today. How Bengaluru's floods spill into its traffic is in why Bengaluru traffic is so slow.

Is it raining harder in Indian cities than before?
Sometimes, but the evidence is more mixed than the headlines. An IMD study of 12 fast-growing cities from 1969 to 2010 found no systematic national trend in heavy-rain days (65 mm or more) or very-heavy-rain days (125 mm or more). Some cities showed rising trends in particular months, including Bengaluru in June, while others such as Bhopal showed falling ones.
If the rain itself has not uniformly intensified, then much of the growth in urban flooding comes from the ground.
That means more pavement, fewer lakes, narrower channels and drains clogged with silt and solid waste. Rain intensity still matters at the margin, and the NDMA guidelines ask cities to build intensity-duration-frequency curves from 15-minute rain data and adjust them for climate change and the urban heat island, which can intensify local showers. How the monsoon itself works is in how the Indian monsoon works.
IMD scientists analysed daily rainfall from 1969 to 2010 for 12 second-tier cities (Ahmedabad, Bengaluru, Bhopal, Bhubaneswar, Guwahati, Hyderabad, Jabalpur, Jaipur, Lucknow, Nagpur, Patna and Pune), counting heavy (65 mm or more) and very heavy (125 mm or more) rain days and testing for trends by season and month. Bhopal and Jaipur showed falling seasonal heavy-rain trends (significant only for Bhopal); Ahmedabad, Hyderabad, Pune and Bengaluru showed rising seasonal rainfall that was not significant for heavy events. Monthly trends were significant in a few cases, such as June in Bengaluru and Patna. Overall, the authors found no systematic trend. Bengaluru's highest one-day total in the period was 177.6 mm, and it recorded 114.1 mm in about an hour on 24 September 2010. The limit: one station per city and data ending in 2010, before the most recent extreme years.
Read the paper (free)India's first national framework for urban floods, written after the 2005 Mumbai disaster. It documents design intensities in use (12 to 20 mm/h in the 1993 sewerage manual, 25 mm/h in old Mumbai, 12.5 mm/h in Ahmedabad), records that urbanisation raises flood peaks 1.8 to 8 times and volumes up to 6 times, and sets 21 actions: Doppler radar coverage, catchment-based drain design, contour mapping at 0.2 to 0.5 m, a runoff coefficient of up to 0.95 for new drains, desilting by 31 March, rainwater harvesting in every building and removal of encroachments on drains. The limit: many deadlines in it (a national urban storm drainage manual by 2012, for one) slipped, and implementation is up to states and city bodies.
Read the guidelines (free)Why do the same spots flood every single year?
Because flooding follows topography, and topography does not change when a lake becomes a layout. The low points of a city are almost always its old water bodies, valley floors and the channels between them. When those are built over, the water still flows to them in a storm, and finds houses, basements and underpasses instead of a tank.
The 2025 Bengaluru floods hit Sai Layout and Horamavu hardest. The city's deputy chief minister said in May 2025 that 166 of 210 identified flood-prone spots had been fixed and ₹2,000 crore allocated for stormwater drains, which also shows how many such spots the city already knows about.
Three repeat offenders show up in almost every Indian city:
- Underpasses and basements. They are, by design, the lowest points around. Without pumps sized for the storm, they fill first.
- Culverts and sea outfalls. A wide drain feeding a narrow culvert under a railway or highway backs up behind it. On the coast, outfalls without flood gates let the sea push back in at high tide: the NDMA noted that only 3 of Mumbai's 105 outfalls had flood gates.
- Mixed sewage and stormwater. Where sewers are missing, homes discharge into stormwater drains, which silt up and lose width, as the NDMA guidelines note.
If you ride through Indian monsoons, the bike side of this is in monsoon maintenance for motorcycles, and reading the forecast well is in how to read an IMD monsoon forecast.

What actually reduces urban flooding?
Restoring the places water used to go, and sizing drains for real storms. The modelling evidence is encouraging: an IISc study of Bengaluru found that a bundle of permeable paving, drainage upgrades and land-use controls would raise average traffic speeds on flood days in 2050 by about 45% and eliminate flood-cancelled trips in its scenario. The NDMA's list is the practical version:
- Map by catchment. GIS contour maps at 0.2 to 0.5 m intervals, and drains planned along natural watersheds, not ward boundaries.
- Design for real rain. Intensity-duration-frequency curves from 15-minute data, adjusted for climate change, and runoff coefficients up to 0.95.
- Desilt before the monsoon. By 31 March, every year, with residents' associations monitoring.
- Hold water where it falls. Rainwater harvesting in every building, permeable surfaces, and lakes kept as lakes, with their inlet and outlet channels open.
- Warn early. Doppler radar coverage for every city and ward-level flood alerts; the NDMA guidelines describe an early-warning system being built for Chennai.
None of these is expensive compared with a single bad flood. The difficulty is that each one requires a city to say no to building in the wrong place, which is the hardest thing for a growing city to do.
What's new
- Delhi gets 90% of August's normal rain in eight days; roads turn into rivers Business Today, Aug 2026
- Santacruz logs 244 mm in a day, its wettest August day in five years All India Radio, Aug 2025
- Bengaluru gets 105.5 mm in 24 hours; 500 homes flooded and three dead The Federal, May 2025
Questions people ask
Why do roads flood within minutes of heavy rain?
Drains sized for 12 to 25 mm/h fill in the first minutes of a 50 to 70 mm/h storm, and paved surfaces send almost all the rain straight to them. Everything after that flows along the road surface to the lowest point.
Why does Mumbai flood every year?
Its old drains were built for 25 mm/h, most of its sea outfalls lacked flood gates against high tide (3 of 105, per the NDMA), and the Mithi river floodplain was obstructed, as the post-2005 Chitale Committee found. The network is being redesigned for 50 mm/h.
Are Bengaluru's floods caused by lake encroachment?
Largely. IISc's work found wetlands fell 79% and paved surface grew over 1,000% between 1973 and 2017, and the city's low points are its former lake beds and channels.
Is climate change making Indian city floods worse?
It adds to the risk, and the NDMA asks cities to adjust design rainfall for it. But an IMD study of 12 cities found no systematic trend in heavy-rain days from 1969 to 2010, so land use and drainage explain much of the rise in flooding.
Sources
- NDMA: National Disaster Management Guidelines, Management of Urban Flooding (2010; design intensities, 1.8 to 8× peaks, Mumbai 2005, C = 0.95, 21 actions)
- Singh, Khole and Rase: Extreme rainfall events and urban floods in the growing cities of India, MAUSAM 66(4) (2015)
- The News Minute: 78% of Bengaluru's surface is paved, IISc study (Ramachandra, Shivamurthy, Aithal) (2017)
- Citizen Matters: IISc study on Bengaluru floods and traffic (Vajjarapu, Verma, Allirani, IJDRR 2020) (2020)
- The Federal: Bengaluru floods, toll rises to 3 (2025)
- All India Radio: Incessant rains paralyse Mumbai (2025)
- Business Today: Delhi-NCR rains, roads turn into rivers (2026)
