2015-06-02

by Judith Curry

Ahmedabad’s Heat Action Plan offers a proven track record and model to protect their residents from heat waves.  My Georgia Tech/CFAN colleagues Violeta Toma, Peter Webster and Mark Jelinek are enabling this Plan with a pioneering heat wave forecast product to help the Ahmedabad Municipal Council determine when to implement a heat alert.

With a death toll now exceeding 2000, the current heat wave in India is the 5th deadliest  heat wave in the world and 2nd deadliest in India [link]. The greatest loss of life in the current heat wave is among outdoor workers. The deadliest heatwave on record in India was in 1998, killing 2541 people.

The current heatwave began on May 21, with temperatures in many regions exceeding 45C, and reaching 47.6C and beyond [link].  Delhi has endured seven consecutive days over 44C, the worst extreme heat event recorded in a decade. The heat wave was caused in large part by sparse pre-monsoon season rain showers, leaving large parts of India arid and dry. Cooling monsoon rains are expected within a few days, which will bring the heat wave to an end.

Of course, the heat wave is being blamed on human caused climate change [link, link]:

The combination of rapidly expanding population, and increasing extreme heat events reducing food yields suggests a perfect storm is brewing. We can only hope that negotiators at the UN climate talks in Bonn next week step up to the mark to ensure concerted and effective climate action.

https://theconversation.com/is-indias-heatwave-a-freak-event-a-statistician-investigates-42636

Regardless of what caused this heat wave event, I don’t think that international negotiations and reducing CO2 emissions is going to help address the problem of India’s heat waves, at least in the near term.  But tactical adaption, following the model developed for Ahmedabad, can make a huge difference with minimal cost.

Tactical adaptation

The idea of tactical adaptation to weather extremes is to make use of probabilistic medium range weather forecasts as part of a strategy for planning and implementing mitigatory actions beginning about a week in advance of the predicted event.

Ahmedabad is a city of 7 million people in the northwestern state of Gujarat. Ahmedabad’s Heat Action Plan is introduced in this recent article [link]:

“Ahmedabad’s Heat Action Plan – South Asia’s first early warning system against extreme heatwaves – is tailored to help protect the city’s vulnerable communities during these disasters.”

After a heatwave hit Ahmedabad, a western city of 5.5 million people, in May 2010, killing over 1,300 people, local authorities mapped areas with “high-risk” populations including slums, as part of an action plan.

They also built up public awareness of the risks of high temperatures and set up “cooling spaces” in temples, public buildings and malls in the sizzling summer months.

In Ahmedabad, authorities have focused on public awareness as the first step in preventing heat-related deaths. This includes telling people how to protect themselves via campaigns on television, radio and newspapers, as well as through messaging platforms such as WhatsApp.

The government also alerts residents to forecasts of very high temperatures through hospitals, community groups, media outlets and government agencies. Health workers have also been trained to recognize the symptoms of heat stress and ensure emergency rooms and ambulances are stocked with ice packs.

A recent blog post from the NRDC describes the Heat Action Plan and the partnership between  the Ahmedabad Municipal Council and scientists and policy experts at  the Indian Institute of Public Health, Gandhinagar, Public Health Foundation of India, Natural Resources Defense Council, Icahn School of Medicine at Mount Sinai, Rollins School of Public Health at Emory University and Georgia Institute of Technology’s CFAN group.  Excerpts:

One thing that is so attractive about the Ahmedabad’s Heat Action Plan is that most of its effective strategies are very straight-forward to implement. There are many actions a city government can take immediately to save lives. [Below] is a summary of what the AMC and other stakeholders are doing right now in the city to warn and protect residents from the ongoing extreme heat.

Posted large hoardings (billboards) & banners with heat preparedness messaging across the city, including in vulnerable slum areas and at all government hospitals & Urban Health Centers. Distributed more than 400,000 handouts of heat preparedness messaging .

Held meeting with all zones’ Assistant Medical Officers and Junior Entomologists to discuss the HAP 2015, identifying drinking water sources and adding more throughout the city. Installed more than 1,100 drinking water stations around the city in collaboration with NGOs and social organizations.

Conducted a survey of all construction sites to see availability of drinking water and shelter for rest in afternoon for laborers working there. All construction sites will be given an order from the Estate Department to provide water and shelter to laborers.

Keeping open all gardens zones as cooling spaces, including gardens that are generally closed over the noon hour.

Have issued media advisory as in all leading newspapers, electronic and radio media to alert the public when there were temperatures forecast hot enough to quality as “orange days alerts” for upcoming next 3-4 days.

Sending WhatsApp texts messages among all AMC department groups to sensitize citizens to avoid outdoors in afternoons until 4:30 pm during extreme heat days.

Providing regular instructions to various departments like water department for no water supply cut and electricity department for no power cut during extreme heat days. Labor & Traffic police department received instructions to ensure safety of their workforce.

Providing regular follow-up with all medical officers and government hospitals to take stock of their preparedness and management of heat-related cases. Keeping a close track on daily reports of heat-related illness & deaths from all government hospitals and on daily all-cause deaths as well to ensure there is no steep rise in death toll.

Conducted sensitization & ToT (Training of Trainers) workshops for various NGOs working for vulnerable communities.

The Heat Action Plan offers ways to mitigate the health impacts of rising temperatures, including mapping high-risk heat areas, increasing access to drinking water stations & green spaces for shade, reducing urban heat island effects, ensuring new buildings are more heat-resilient, and developing transportation systems that help people avoid heat stress.

The documents describing the Heat Action Plan can be found [here].  See also a paper published in the Int. J. Environ. Res. Public Health [link]. This figure by Jeremy Hess from a presentation made by Peter Webster provides a good summary:



Forecasting Ahmedabad’s heat waves

Effective implementation of the Heat Action Plan requires a good forecast out 7 days or more in advance of the event.  Weather forecasts for Ahmedabad are made by the Indian Meteorological Department (5 day, deterministic), weather.com (10 day, deterministic), accuweather.com (45 day, deterministic).  The accuracy of the publicly available forecast products was insufficient at 7+ day lead times to support the Heat Action Plan; further, these forecast products are all deterministic, rather than probabilistic (which is needed at longer forecast horizons.)

In 2013, led by Violeta Toma and Peter Webster,Climate Forecast Applications Network (CFAN) began participating in the formulation of Ahmedabad’s  Heat Action Plan.  CFAN was tasked with providing a 7-day probabilistic forecast of maximum temperatures (based on the ECMWF ensemble forecast system) during the pre-monsoon hot season. During the Ahmadabad hot season, prior to the monsoon rains, maximum daily temperatures are on average around 41°C (106°F), with record values of over 46°C (~115°C), such as during May 2010.

The temperature thresholds were developed in collaboration with the local authorities and the public health officials in Ahmadabad based on research relating mortality and morbidity with maximum temperatures. There are four distinct alert levels:

“Safe/White” when the temperatures are below 41C,

“Hot/Yellow” when the temperatures are between 41C and 42.9C,

“Very Hot/Orange”, when the temperatures are between 43C-44.9C

“Extreme Heat/Red” at temperatures of 45C and over.

For each alert level, there is a set of actions to be taken to diminish the effects on the population.

The challenge to producing useful medium-range heat wave forecasts is to overcome the biases seen in the forecasts made by IMD and accuweather.com and weather.com (accuweather.com and weather.com forecasts have a high false alarm rate for Amedabad heat waves).

CFAN’s heat wave forecasts for Ahmedabad takes into account both historical and recent bias of the ensemble forecasts, using the quantile-to-quantile adjustment, the adaptive (Kalman-type) decaying averaging bias correction, and kernel dressing to address the under-dispersion of the ensembles.

For the adjustments/corrections to work properly, the accuracy of the observational data set is important.  There seems to be an inconsistency between the IMD reported maximum temperatures and the temperatures reported by the NOAA/NCDC based on automatic WMO station measurements. At the peak of the hot season,  there have been several instances where the NCDC measurements are higher than the IMD measurements. Based on our conversation with the IMD, the sites are some distance apart and they use different instruments. While theses technicalities can explain the differences between the measurements, its still unclear what is the actual temperature in the city. Moreover, we don’t know what is the effect of urbanization on the inner city temperatures, or if the met station location is representative for the entire city.

The heat wave forecasts are presented via a web-based dashboard.  The forecast below was initialized on June 1, 2015,  and includes daily forecasts out for 7 days.  A color coded alert level warns for the next day’s forecast.  The daily maximum temperature forecast is provided as an ensemble mean, and also as the probability of exceeding certain thresholds.

Recent past forecasts are also provided, to show forecast consistency.



Verification  statistics for April-May- June 2014 are shown below.  The root-mean-square-error is seen to be below 1.2 for all lead times out to 8 days.

The ability to forecast the extreme heat waves is illustrated by the Hit Rate (probability of detection) and the False Alarm Ratio.

Note that based on the IMD observational records, Ahmedabad had only one extremely hot day during the 2014 hot season.   Hence the results for the “Extremely Hot” instances are not statistically robust. Additionally, there appears to be a systematic bias in the IMD records compared to NCDC and METAR data, with IMD reporting smaller values for temperatures in the very hot/ extremely hot bracket. This might lead to CFAN large false alarm rate in the extremely hot bracket.

Overall, the contingency tables show that our threshold-based forecast has performed quite well with very large probability of detection and small false alarm rate.

The current heat warning system in Ahmadabad takes into account the effect of maximum daily temperature and does not account for either atmospheric humidity or minimum daily temperature. However, these last two factors appear to be as critical as the maximum daily temperature.  Thus incorporating atmospheric humidity an minimum daily temperature is something we plan to address in the future.

JC reflections

The Ahmedabad Heat Action Plan is a marvelous example of tactical adaptation to climate.  CFAN (under the leadership of Peter Webster), has been working for over a decade to try to apply CFAN’s cutting edge capabilities in extended range prediction of extreme weather events.  It is not an easy thing to pull off.

The successful ingredients in this case are an enlightened local government, a willingness to accept weather information from a source that is not the IMD, a boundary organization (the NRDC) and a consortium of researchers from both India and the U.S.  The IMD barrier is a nontrivial one, Peter Webster relates this anecdote:

When we told the IMD that we would use probabilistic methods, their comment was “Professor Peter, you must know that probabilities do not work on the subcontinent.”

It is almost easier to work with genuinely undeveloped countries, who do not have their own meteorological agencies struggling for recognition and for their jobs.

This idea of tactical adaption to extreme weather events is one that Peter Webster has been working on for two decades (I coined the term ‘tactical adaptation’ in this context).  The most mature example is Bangladesh flood forecasts [link].  At the heart of tactical adaptation is probabilistic weather forecasts on timescales of a week to months.  Making these forecasts actionable requires a detailed plan of action for mitigatory actions to take at intermediate and then shorter time frames.  The basic principles work for any weather hazard, whether caused by climate change or natural variability.  The advantage of tactical adaptation is its low cost.   Tactical adaptation to extreme weather events is deemed of particular importance to the developing world, where extreme weather events are routinely associated with large loss of life, relentless impoverishment, and destabilization.

Focusing more resources on medium to extended range weather forecasts, along with developing action plans tied to these forecasts, would go a long way towards alleviating problems associated with extreme weather events.  And provide a far better investment in human development and economic benefits in the near term than reducing carbon emissions.

A society that learns to deal with present hazards through tactical adaptation will be more adept at dealing with future hazards that may be more frequent and intense.

Filed under: Adaptation

Show more