Every year, by the time June approaches, Indians start anticipating and discussing the arrival of the monsoon. This year, of course, the monsoon has far greater significance -- a good monsoon will not only impact rising prices but also have a bearing on the elections to follow.

"It would be a normal monsoon this year. So it should be good news for the people," says Dr Rajeevan, a weatherman who joined the IMD 23 years ago.

In a detailed conversation with Archana Masih on his day off on a quiet Saturday morning, he spoke about the complexities about the Indian weather system, why predictions have gone wrong at times and why if the Americans are asked to issue a forecast for India, they will also do badly.

This year the monsoon has far more importance with rising prices and so much depending on how good the monsoon will be. It will also have a bearing on the elections that are to follow. What is the monsoon forecast for this year?

This year the forecast is that it will come by May 29, of course there is a margin of 3 to 4 days, it can come early or come late. It is almost a normal onset.

We issue many forecasts for the monsoon -- one of course is the monsoon onset, that is when Kerala [Images] will get the first rain -- it is very important. It may not have economic value but definitely has a psychological value that 'Ah, the monsoons have come.' It is a tricky issue.

The Indian Meteorological Department, IMD, gives its forecast in April which is updated in June after more data comes in. In April we give an advanced forecast so that the government has enough time to respond. It is very difficult but we are trying to do it and are doing it for many years now.

In addition to that we issue a separate forecast for monsoon rainfall for the country as a whole. That is very important -- whether there will be a drought, normal monsoon or excess monsoon.

We issued the first forecast for this in April which said that it would be a normal monsoon this year. So it should be good news for the people. At present we don't see any adverse things that will affect the monsoon performance. There may not be plenty of rain but definitely it will be normal rain which will be sufficient for good agriculture. It is not bad news.

What changes are there likely to be?

It can change by hardly 2% to 3%. Definitely it is not going to be a drought this year. There is more chance of a normal monsoon than anything else.

How do you go about predicting the Great Indian Monsoon? Last year you used a technique called the ensemble method.

There are two main methods of issuing forecasts -- statistical or empirical method that IMD has been using for more than 100 years now. The first forecast was issued in 1886, and India was the first country to issue a seasonal forecast. So we are the pioneers. The US, for instance, uses statistical method plus dynamical method, in which they run a numerical model using a big computer.

There are so many things we don't know about the monsoons, there is more information about the Pacific Ocean. They know more about the American climate, the Indian monsoon climate on the other hand is very complex. People have been working on it, even in the US, there are so many universities working on it.

What are the primary complexities about the Indian climate?

There are so many complexities that are involved -- topography, the influence of the ocean. India is locked from north and south by the Himalayas and the ocean. The Himalayas have a non linear topography and the monsoon system that is formed is very tricky, it is not a very simple, its relationship with ocean dynamics is very difficult.

We don't know how the Indian Ocean is behaving, we have more information available about the Atlantic Ocean. They have more data, more research, more modeling, observation -- so there is plenty of data about the Pacific and Atlantic Oceans.

I am not saying that people are not working on the Indian monsoon. In India and abroad people are working on it but as far as our monsoon meteorology is concerned, it is said God has given us a little problem -- it is called predictability.

For example, you can better predict the Australian or African monsoon but when it come to India -- the signal to noise ratio is very small -- that is, basically you are trying to predict weather into feature -- so what happens is there is a lot of noise in the ocean data, so as you continue assimilating data over days, this noise will grow and you get more noise than signal.

So God has given us a very small signal to noise ratio. Nobody can help us with this, there are many reasons for this. This signal to noise ratio is predictability and so the limit to predictability is itself very bad over the Indian monsoon compared to other monsoons.

Nature itself has given us some limitations. Within these limitations we try to do maximum as possible.

There was a difference in the onset of the monsoon as predicted by your department last year while the monsoon arrived later.

That was about the onset of the monsoon but our rainfall prediction was off by 11% which is not really good. Last year the model did not perform very well but that does not mean that the model is bad. Models are not expected to do well every year. No model is perfect.

Models are being updated every 5, 6 years. We get new ideas and data into the analysis. We apply all this and use these to improve. So the ensemble model which we introduced last year is basically an attempt from us to improve the capability of the statistical model. Last year the prediction wasn't pretty good.

The IMD in the last five years has got its prediction wrong thrice (2004/2006/2007) and right twice (2003/2005) -- is that correct?

2004 it was bad. 2006 it was okay. We need to have a correct forecast for drought. We should be able to tell if a drought will occur in the monsoon season. We should be able to tell that in advance so that the government can take some kind of mitigation measures.

We did this last year also, so our emphasis is to predict a drought. Drought is important, this is not to say that other monsoon scenarios are not as important. The impact of drought is more than a good monsoon. When rainfall goes on the negative side, its impact on economy and agriculture is very large. When it rains more the impact is not much.

What is the signal for drought this year?

The possibility of drought is very less this year. Climatologically, take 100 years of monsoon rain data -- for 20 years we had drought -- even if a person has to issue a forecast, there is a 20% probability of drought. This 20% probability is always there.

When IMD gives a forecast for drought that probability is 5%, it means that drought probability has come down -- it is almost ruled out. We are giving a probability less than the climatological probability. 5% is nothing, it is good news.

So this year the drought possibility is very less?

Yes. Excess rainfall is also not very much. We have more probability confined to normal monsoon. There is 60% probability for a normal monsoon.

You have said no set of equations or formula can predict the Indian monsoon consistently and accurately, is it like you have mentioned earlier in the interview, primarily because of our weather system or are there other reasons like lack of state of the art of equipment?

There are plenty of reasons, one is of course, that our weather system is highly complex. The person who designed the first computer Von Newman, he once made a remark that seasonal forecast is the second most difficult prediction problem in the world. The first is to predict human behaviour.

That is the seriousness of the problem. But that does not mean we can rest on that. We have to do the work. Still we need to know many things about the Indian monsoon. We do know but we need to know much more. We need more observation from the Indian Ocean.

Why don't we have enough data from the Indian Ocean?

There are lots of reasons but now things are improving. Under Kapil Sibal (the Union minister for science and technology) we are putting automatic stations, automatic rain gauges, Doppler radars, buoys, data platforms in the ocean -- so it is being done. The future is definitely bright.

What about the IMD's weather channel that is going to be launched? What can you tell us about it?

They already have a project team. I understand they have appointed some private agents to see the feasibility of sustaining the channel. The feasibility report is being submitted. The government is very serious about it.

We have to teach the people, because the required awareness is not there. In the USA and Australia -- the information, not only forecast but real time information about weather; if El Nino is developing -- the climate information data is being conveyed to the people. People should be receptive and know what kind of data is being given, how it can be made use of.

We issue forecasts that the monsoon will be like this but we don't know how people are using that information. Whether they need something else, how they use this information in agriculture, we don't know how to gauge that by simply issuing a forecast. We need many application oriented projects.

In Australia when you issue a forecast, farmers are told if you believe this forecast what is the risk involved. The farmer says by going with this forecast, his agricultural produce will have so much percenatge of loss. If I don't, I will have a greater percentage of loss. We do not do this kind of assessment in India. We do not know how much risk is involved.

When do you begin work on predicting the monsoon every year?

We do it round the year but we begin collecting data in October, some data from December goes into the model and we have continuous monitoring. We don't wake up at the last moment and collect data.

How many places in India do you get this data from?

For seasonal forecasts, we take data from the Atlantic Ocean, Pacific Ocean, Indian Ocean, South Indian Ocean, China region. For day to day forecasts IMD has 550 observatories, 30 to 40 upper air observatories -- we send balloons -- and radars. Rain gauge network and radar network around the coast. We don't have a very good mesoscale network which is very important for predicting thunderstorms and strong winds. Like the Mumbai rains happened, we need a high resolution mesoscale network every 2 to 5 kms. A Doppler radar was asked for by Mumbai, it is coming.

Isn't it true that we don't have state of the art equipment to predict our weather?

Not completely true. We do have our own network. Some instruments are obsolete and we need better equipment and technology -- Doppler radar is a new technology and we have only 6, 7 of them in India. We need weather stations, observatories to be automated. The government is doing it.

In this financial year, they will install 1,000 automatic rain gauges, which will be introduced for the first time. We have some automatic weather stations. We have already installed 150 or so and next year some 1,000 are to be installed, so the government is taking action. We definitely need more data, good observations, better equipment and the government is doing it.

There is so much stress laid on the monsoon, understandably so because 70% of our country is dependent on agriculture -- but since we have been experiencing such bitter winters and severe summers which has taken so many lives in the recent years -- we don't seem to have information about how these months are going to be. What are we doing about it?

We seem to be more worried about the monsoon rainfall and we have 100 of problems with the monsoon season itself but we appreciate a need for issuing a separate forecast. So far nothing has been done, I admit but definitely we are planning to do that.

Why are our winters becoming colder and summers hotter? 2007 was the fourth warmest year since 1901 -- and among the 10 warmest years since 1901, 8 have been after 1998.

Our temperatures are increasing. Warming is continuing, it is a part of global warming. The maximum day time temperature is increasing, not night time temperature. That means you have more heat waves. It has been increasing for the last 20 years. It is also observed that the intensity of rain has increased. It rains a heavy amount in a short time, like what happened in Mumbai. Research and theory has shown that the intensity of precipitation has increased in many parts of the country.

Our country also sees a lot of havoc being caused by floods each year? Why don't we have proper prediction for that?

We can predict 2, 3 days in advance. I can't tell you now where it will be flooded in August. I can tell you 2, 3 days in advance.

It may not be enough?

But it can still save lives. People can be asked not to go out etc. The nullahs can be cleaned. Many floods are also man-made because of inaction.

When is the monsoon hitting Mumbai?

Let it come to Kerala. If everything goes very well, by June 7-8, it should come to Mumbai. It is very difficult to say.

In countries like the US, much importance is given to weather forecasts on news channels. People watch the weather on television before setting out, how do they bring about a largely accurate forecast?

One reason is their science and technology is better. Secondly, their weather pattern, their science of weather is different. Here up to 3 o' clock, there will be no cloud but at 4 you will get a cloud and by 5 it will rain. This won't happen in the US. If it's going to rain there at 4 pm, by morning there will be clouds. Even two days prior it will be known that clouds are coming. But not so in our country.

If the Americans are asked to issue a forecast for India, they will also do badly (laughs). That's what I feel because our weather pattern is difficult. When you assess a forecast skill, don't assess the IMD's skill with the skill of the USA. It is like comparing Tendulkar with a school cricket boy. It is meaningless.

You should compare the weather forecasting capability of India with any country in the tropics, that is a reasonable comparison. I am not saying the IMD is doing the best job, but it is uneasy for me to compare the IMD skill US or UK. You can compare India to Thailand or to some extent Korea, Singapore. Among the tropical countries, we are leaders, that's what I feel.

You are hopeful and feel things are improving.

Definitely. But we need more people. Many students come for research in meteorology. There are courses in IIT, Indian Institute of Science but they are not coming here and go out. They may be working on the monsoon but could be doing something else -- like studying the weather over Scandanavia or the Belgium rain. We need 200 good officers in IMD. Even if we want, we can't get 200 good people trained in meteorology.

Are there enough people opting for meteorology?

Education of meteorology is not being done very well in many universities. It is being taught in Cochin university, Poona university, Andhra university. Most universities teach physics, we need universities to come up with these courses.

Once students get their PhD, there should be a mechanism to absorb them in India and give them good jobs. I know some good students from IISc but next thing I know they are working in some lab outside India. Many people opt for IT now, so IT has really spoiled the whole science scene.