Private forecasters no match for us, says India’s Met chief

In August last year, KJ Ramesh took charge as the Director-General of India Meteorological Department (IMD), the country’s national meteorological service. Ramesh is heading IMD at a time when it is in the middle of expanding their observatory network and has plans to improve its technological prowess for more accurate forecasts. Ramesh spoke to DNA on issues ranging from IMD’s expansion plans, forecasting systems and coping with private weather forecasters. Excerpts.

IMD is in the middle of operational and infrastructure expansion. Can you explain what it does it involve? 

It is happening in two directions. Consolidating the existing observatory systems and augmenting them as per emerging needs and expanding them to go to the next level. To render services we need to better forecast services, with assistance of forecast models of different ranges and finer resolutions. There is a transition within IMD too to test new variants. For all that to happen, we need high performance computing systems to churn out more sophisticated models and the ministry (earth sciences ministry) will make them available. Orders have already been placed by ministry for high performance computing.

At the moment, we have 1.25 petaflops peak computing powers, very soon we will take them to 8-10 petaflops, this year. This will help in leapfrogging in forecast skills. Once computing gets commissioned, new models will be tested.

What are the targets for adding physical infrastructure? 

The Indian Air Force will be adding 12 radars while IMD will be adding 11 radars to augment the radar network. We commissioning in such a way that each district has one automatic weather station (AWS) and two automatic rain gauges. That is the sort of network we have established. After that, for the next three years, we will have two to five AWS’ and then about 15 rain gauges. In cities, we will densify the network for capturing heavy rainfall (data). These systems will help us cover the plains and later we will add more infrastructure in the hills.

This year, IMD gave primacy to dynamical forecast model over statistical model for this year's monsoon. How has this helped improving the forecast?

Now, after migrating to dynamical techniques, it is giving much more flexibility in terms of assessing month to month variability. We have used both the models, statistical and dynamical models. Incidentally, both models gave the 96% (+/- 5% error margin) figure as far as the all-India rainfall is concerned that too, based on conditions up to March. Early April conditions of El Nino, we showed the prospect of El Nino developing in August and September as projected by every model. That prospect and associated probability is coming below 50%. It was more than 50% till March. The reduced probability and a moderate El Nino has been confirmed by the Australian Bureau of Meteorology too. This was not possible to envisage using a statistical model, it is impossible. Because of continuous coupling of ocean and atmosphere in the prediction process, this is possible. Continuous interface and interplay was not available in the statistical model.

IMD is often criticized for jargon heavy bulletins and forecasts. How is IMD addressing that and are there plans to increase social media presence? 

Changing the outreach pattern and public friendly is a simple thing to do. We are working on to generate more content to provide it on multiple platforms. Pushing the information will not be a problem. We are trying to work on the terminologies, it would be a gradual process. That transformation will keep happening, the heat wave bulletins are already more generic in nature.

Is there a possibility of the El Nino becoming stronger as monsoon approaches or once it sets in?

Ocean uncertainties will be high as long as the variability of temperature is moderate. With sea-surface temperatures at 0.5 degrees, probability of its occurrence is decreasing, from March to early April, it has decreased. As long as moderate variabilities of sea-surface temperatures occur, that will result in more uncertainties. That is why we said during our briefing on first monsoon forecast, we said that more clarity would come when we go for second update of the monsoon forecast, because temperature changes are moderate in nature. At this point, we can say that El Nino is going to be moderate

How is IMD coping with competition from private weather forecasters? Skymet is collaborating with Maharashtra government to run automatic weather stations in rural areas. 

As far as the service is concerned there is no competition, from anybody. Because nobody can have that knowledge base or system support base to render service of this magnitude. I don't think anybody has and they will have ahead. Private sector will never invest in this kind of thing, which is of only public good. That is the mandate of national weather service. What they are doing and how state governments are engaging them is the following. At the moment, private forecasters are only limiting their services to give modules to TV Channels and tailored information to newspapers, or in vernacular languages. 

We cannot work to generate a module for TV channels, three hourly updates. Our information is available, we provide it to DD too. But they prepare an infographics. There is always an IT company in between to package the information and present it in a manner that is required. They (private forecasters) pull the information from our website and package it and serve their clients. 

Now, with crop insurance, risk transfer has to be accounted. To claim insurance, you need proof of damage due to weather conditions. Only to have that mechanism, some states have set up their own automatic weather stations. Karnataka, Andhra Pradesh have set up their own networks at sub-taluk level. 

These weather stations record climate data and here possibilities are explored by private weather companies. They want to serve those clients. Maharashtra is not setting up any network, they are asking the companies to set up network and states seek data when there is crop damage as evidence. In one way, it is good as states don’t have to invest in networks and maintenance. 

So you are calling them weather companies and not weather forecasters? 

They are not forecasters, where is the forecast? They are trying to show their capabilities by capturing publicly available data sets of predictions.  Predictions of US, UK, European Union, Australia and Japan, Canada are available in public domain. You capture everything and make your own assessment. 

Are they also using IMD forecast models? 

Our things we have not put it in public domain. They may be using satellite data, automatic weather station, rain gauge data. 

In terms of the changing climate, are we seeing more extreme weather and rain events?

Rainfall events of 10 cm (100mm)  or more per day are on the increase, definitely. Moderate intensity events of 5cm per day are on the decline, while the seasonal quantum remains the same. At the same time, number of rainy days are also gradually decreasing. That means that lesser rainy days and higher intensity events are bringing the same quantum of water. That is why we have a challenge of capturing and storing. It is not gradual. Wherever this gradual rainfall is not happening, rain-fed agriculture will have a problem, such as pulses, oilseeds. They need a gradual spell of rains. These are some of the impacts at local levels, in pockets. Either it is high or there is no rain. We should have means of providing supplementary irrigation to cope with this.