Edit

Mapping floods to assess risks and vulnerabilities in the Chennai Metropolitan Area

Satyarupa Shekhar and Craig D'Souza

April 5, 2016

Context

Tamil Nadu experiences severe water shortages, and water stagnation and flooding every year. The recent policy focus in the state has been on groundwater recharge through rain water harvesting, but in practice, the public water utility has been acquiring 'water fields' - agricultural lands in the peripheries of Chennai where the level and quality of ground water is amenable for supply to the city's burgeoning water demands. Simultaneously, poor planning practices and enforcement of building rules has resulted in the majority of the city's lakes and ponds built over, obstructing the natural hydrology. Unfortunately, successive governments have allowed for weaker plans and poor enforcement of the rules; they have even pushed for amendments that regularise violations and exemptions that will benefit the more affluent. For example, the Tamil Nadu government is currently in the process of framing exemptions for multi-storey buildings, reducing the parking requirements within property premises, and reducing the permissible distance between buildings and aquifers, among others. This degradation of the urban ecology has amplified the magnitude of risks that Chennai faces, particularly weakening the resilience of the city's residents.

The city has a poorly planned and severely inadequate storm water drain network, which is in a high state of disrepair despite the large amounts of funds allocated to the construction. A significant feature of the government's spending on all public infrastructures is that the focus is on construction but negligible attention is paid to maintenance. The built channels for water do not adequately meet the infrastructure needs of the city. The Buckingham Canal, the city's main waterway, was built to convey water from the inland to the sea but has been encroached. The commuter train line is almost wholly built over the Canal. The government has also frequently allocated land and built low income housing in large marshlands and natural catchment areas in the city, amplifying the vulnerabilities of the urban poor. All of this has resulted in Chennai experiencing water stagnation every monsoon and severe floods every 7-10 years.

In 2013-2015 we worked with the Corporation of Chennai (CoC), the city government, on building their capacity to collect and use (spatial) data for planning and monitoring. Despite several contracts and commissions to private businesses and universities, the city government lacks data and maps that it can use to plan and monitor public services, including storm water drains. In October 2014, we worked with city engineers to map water stagnation points in the city and documented possible causes and solutions to create a micro plan. The relevant department relied on an incomplete map that was neither geo-referenced nor to scale. During our engagement we saw that city engineers relied on their latent knowledge to address problems that were of a routine nature, but because of the informal nature of the knowledge and actions, the interventions remain ad hoc and in reaction to water overflows or stagnation. For example, they were aware that the drain network was illegally connected to the sewerage network at several places and this was being carried to the city's waterways and water bodies untreated, but were helpless in institutionalising this knowledge and leveraging it for planning and monitoring. At that time, CoC had requested our support in creating a geo-referenced map of the sub-surface storm water drain network but we did not have the capacity to undertake this task.

The flood map can help us make a compelling case for local actions to combat climate change. Often, city officials have wrongly attributed the devastation from unexpected flooding to the results of climate change when in fact it is a result of poor planning and infrastructure. It has diluted the strategies that cities need to adopt to address climate change and improve resilience.

Objectives

The objective of our short research project was to make recommendations for the designation of flood risk zones within the Chennai Metropolitan Area. In order to do this, we needed to determine the actual extent of the inundation caused due to the floods in November and December 2015. Next we examined the possible influence of some factors in determining whether a given part of the metropolitan area will flood or not to identify the most relevant causal factors behind flood inundation in different parts of the city.

While the potential determining factors are many, only a few of them such as height above Mean Sea Level, proximity to water bodies (existing and reclaimed), proximity to water channels and size of upstream catchment area were considered. Other factors such as soil type, prevailing soil moisture levels, slope of land, effect of operations of upstream manmade water structures were not considered for the current timeline and resources.  These limitations in scope could be addressed by developing a watershed model for the Chennai watersheds which factor in these variables.

The geographical scope of this exercise is limited to the Chennai Metropolitan Area. The temporal scope is limited to the floods of the first week of December, since based on available rainfall data and news reporting the flood inundation was at its peak during this period.

Methodology

We primarily relied on three sources of information. We identified the flood inundated areas from the satellite image we obtained from the National Remote Sensing Centre, ISRO for the 4th of December through a Web Map Service (WMS). This image captures the extent of flooding in the region with reasonable accuracy. For the map of water bodies we used the 1955 US Army map. To estimate the elevation of the Chennai Metropolitan Area we used a Digital Elevation Model (DEM) from NASA.[1] Using these maps and information, we created the following four maps

  1. A coastal flood risk estimation map
  2. A historic water bodies map
  3. A water channels map
  4. A flood inundation map (for the 4th of December 2015)

Map 1 - Coastal Flood Risk Estimation Mapping

This map attempts to predict or estimate zones in proximity to the coast within the Chennai metropolitan area that are at risk of flooding due to the heavy rains. The line of reasoning follows that coastal areas below sea level or within a few metres of sea level are more prone to flooding both because of ingress of sea water as well as the longer time taken for drainage of storm water.

We edited the DEM data to indicate the areas of high and low vulnerability with different colours in QGIS. The blue colours in the map indicate areas below Mean Sea Level (0 m) red indicates 0-4 m, pink 4-8 m and shades of brown indicate elevations greater than 8 m.

Legend

dem_chennairiverbasins



Map 2 - Historic Water Bodies Mapping

Digitising both the currently existing as well as the former water bodies in the Chennai Metropolitan Area is important to understand the effect that water bodies and their reclamation may have on flood inundation. The US Army Topographic Maps (1955) showing water bodies for the Chennai area are the best available reference showing old water bodies in the region.[2]We first edited these maps in GIMP/Photoshop, a photo editing software, to filter out pixels from the map that showed water bodies. This was possible by converting into a binary format, with blue showing water bodies and white showing everything else. We then converted the jpg to a geo-referenced tiff file, which in turn, we converted to a polygon vector showing water bodies.

Since this map was pre-satellite era we cannot be completely certain of the precise location of these old water bodies. However, the fact that the water bodies shown on the map clearly align with several currently existing water bodies lends credibility to it.


Map 3 - Chennai Drainage System Mapping

In addition to the water bodies, we also identified water channels, i.e. the path water would take when it rains. The term ' does not necessarily indicate that water flows via these channels perennially.

As a first step we applied a ‘sink fill’ algorithm to the DEM to smoothen out any gaps. Next we made flow direction and flow accumulation rasters using tools from the SAGA toolbox in QGIS.[3] The flow accumulation raster assigns different values to each pixel on the raster. If the pixel has a large number of pixels upstream bringing water towards it, then it will have a correspondingly high number on the flow accumulation raster, i.e. larger water channels. After we prepared the flow accumulation raster we filtered out pixels with higher values using raster calculator. We then used the QGIS Polygonise tool to convert the raster files to vector, to obtain a water courses map with different line patterns indicating the size of the watersheds of each water channel.

Legend

Chennai River Basins 30m SRTM DEM


watercourses



Map 4 - Flood Inundation Mapping

To complete the analysis we used a map of actual flood inundation for the city. Bhuvan, the web-platform of the National Remote Sensing Centre, published three maps of flood inundation of the Chennai Metropolitan Area for the dates of December 3, 4, 5 and 7 as Web Map Services that they gave us access to. We found the map for the December 4 most suitable for this analysis because it had the highest resolution. We viewed the map using QGIS, took geo-referenced snapshots and saved these as png files. We then converted the png files to polygon vector format using the QGIS Polygonize tool to obtain a vactor file that shows us the boundaries of the flood inundated areas.

Legend

watercourses


NRSC Satellite Observations - Flood Inundation



Insights

Overlaying the satelite observations of inundated areas in the December 2015 floods with the Madras City 1909 map (from Project 3) shows a significant correlation between the built environment in 1909 and the floodplains of 2015. 

The 1909 urban areas (shown in pink on the map) are generally built in locations that were not inundated in 2015.

Between 1909 and the present day, the continued expansion of Chennai has resulted in property development on these floodprone areas. 

Legend

Madras City 1909


1955 Water Bodies


NRSC Satellite Observations


Next Steps

A number of tasks can be carried out to improve our understanding of flood risk across Chennai

1. Incorporating the full Web Mapping Service data from the National Remote Sensing Centre

2. Mapping the results of "Survey of Losses sustained during the Chennai Floods"

3. Mapping the georeferenced tweets from #ChennaiFloods and #ChennaiRains to understand where and when the flooding occured.

4. Estimating or acquiring appropriate rainfall data for the December 2015 event.

5. Developing a numerical hydraulic model of the flood event and calibrating it to this data. 

Get involved

ChennaiFloodManagement.org is an open, stakeholder powered platform. If you can help us with any of the above tasks, please get in contact with Satyarupa at satyarupa.shekhar@cag.org.in

 

References

Note the spatial data used for each map is referenced in the "Map Metadata" button.

Additional references used are listed below.

[1] SRTM 1arc second (https://lta.cr.usgs.gov/SRTM1Arc), NASA

[2] http://www.lib.utexas.edu/maps/ams/india/

[3] http://gracilis.carleton.ca/CUOSGwiki/index.php/Exploring_the_Hydrological_Tools_in_QGIS