2015 Archives - The India Climate Observatory

Lessons of monsoon 2015 for district India

September 16, 2015 by Climate portal editor 2 Comments

By our conventional method of reckoning the adequacy of the 2015 monsoon, this is a year that is amongst the most deficient in rain over a period of 20 years. The monsoon season began late, compared with its usual onset, and apart from a few sustained heavy spells in a few locations, has been less than adequate and also less than normal in every one of our 36 meteorological sub-divisions.

When after eight weeks of the conventional monsoon season it became evident that a combination of factors was causing weak and erratic rainfall, that is when the central and state governments needed to place on alert the regions that were already facing rainfall deficits. At this point, when we have the evidence provided by data for 15 weeks of the monsoon season (until 9 September), every week from early August onwards that passed without such a declaration is a week of preparation and coping lost.

In this commentary, I have described the advantages of using a new methodology that grades rainfall adequacy at the level of the district, to a degree that is very much finer than the five categories of the India Meteorological Department (IMD), which are: normal, excess, deficient, scanty and no rain. The outputs from this methodology (illustrated and described here) are designed to: (1) alert national food and agriculture administrators to impending food insecurity conditions; (2) alert national water resources administrators to impending water scarcities; (3) alert line departments of state ministries and district collectorates to the build up of climatological distress at the district level so that contingency measures can be taken.

Over the conventional season (1 June to 30 September) the inadequacy of rainfall in 2015 is revealed at a glance by the weekly rain report. It makes for a very alarming picture and shows that state administrations and especially district authorities should, by the sixth week which ended on 8 July, have made arrangements to prepare for below-normal rains. In the weekly rain report, each vertical bar corresponds to a week of districts categorised into eleven grades. This provides a weekly national barometer of the number of districts that are in the lower and upper (severely below and above normal rainfall) categories during a given week.

Such a weekly rainfall adequacy report is able to quickly put a stop to the recent tendency of administrations, the media and all those who must manage natural resources (particularly our farmers), to think in terms of an overall seasonal ‘deficit’ or an overall seasonal ‘surplus’. This ‘seasonal’ view must be abandoned because demands for water are not cumulative – they are made several times a day, and become more or less intense according to a cropping calendar, which in turn is influenced by the characteristics of a river basin and of a corresponding agro-ecological zone, and the rural and urban populations in a district.

The difference between the IMD five-grade assessment and the eleven-grade categorisation of rainfall becomes immediately apparent when a comparison is made for any given week. The data source is the same – the weekly tabulation compiled by the IMD’s Hydromet Division (which from this monsoon season provides the data sheets and detailed maps on its ‘customised rainfall information system’, or CRIS, website).

Week 11 of the monsoon season, which is the week until 12 August, provides such an example. Under the five-grade IMD scale, there were 114 districts with normal rain (from -19% to +19%). Under the 11-grade new categorisation, the middle grade is -10% to +10%, and included 66 districts. Under the five-grade IMD scale districts with below normal rainfall fall under deficient (-20% to -59%) or under scanty (-60% to -99%) and for this week the number of districts respectively were 155 and 223. Under the 11-grade new categorisation, there are four grades for below-normal rainfall that is -20%.

Thus while the ‘deficient’ grade includes 155 districts, under the 11-grade system there are 150 districts distributed between two grades – 84 and 66, but we see that a larger number of districts fall in the more severe of the two grades. The signal to be derived from this, at the state and districts administration level, is that if a district remains for two to three weeks at a grade, then contingency measures must be reviewed, readied or rolled out. This is a decision that becomes considerably easier with the 11-grade system when compared with the existing five-grade system.

In the same way, the week by week tabulation of districts under the 11-grade system reveals trends and patterns that are not visible under the existing IMD five-grade assessment. The table shows the distribution of districts by grade across weeks. In each week, the two grades that account for the largest number of districts are highlighted red. We see that for the the first five rain weeks – week ending 3 June to week ending 1 July – the +81% and above grade was one of the top two populated grades. This occurred once more for Week 7, ending 15 July. For the next eight rain weeks – ending 22 July to ending 9 September – the top two populated grades have been in the rainfall grades of -41% to -60%, -61% to -80% and -81% and less. At the country level, this starkly underlines the seriousness of the rainfall deficit.

The uses to which we have put available climatic observations no longer suit an India which is learning to identify the impacts of climate change. Until 2002, the monsoon season was June to September, there was an assessment in May of how well (or not) the monsoon could turn out.

The India Meteorology Department has added computational and analytical resources furiously over the last decade. The new research and observational depth is complemented by the efforts of a Ministry of Earth Sciences which has channelled the copious output from our weather satellites, under the Indian Space Research Organisation (ISRO), and which is interpreted by the National Remote Sensing Centre (NRSC), to serve meteorological needs.

The IMD, with 559 surface observatories, 100 Insat satellite-based data collection platforms, an ‘integrated agro-advisory service of India’ which has provided district-level forecasts since 2008, a High Performance Computing System commissioned in 2010 (whose servers run at Pune, Kolkata, Chennai, Mumbai, Guwahati, Nagpur, Ahmedabad, Bengaluru, Chandigarh, Bhubaneswar, Hyderabad and New Delhi) intelligently consumes an astonishing amount of numerical data every hour.

Over the last four years, more climate and weather ‘products’ (as the IMD system calls them) based on this data and their interpretations have been released via the internet into the public domain. These are reliable, timely (some observation series have even three-hour intervals), and valuable for citizen and administrator alike.

Even so, the IMD’s framing of how its most popular measures are categorised is no longer capable of describing what rain – or the absence of rain – affects our districts. These popular measures are distributed every day, weekly and monthly in the form of ‘departures from normal’ tables, charts and maps. The rain adequacy categories are meant to guide alerts and advisories.

These number four: ‘normal’ is rainfall up to +19% above a given period’s average and also down to -19% from that same average, ‘excess’ is +20% rain and more, ‘deficient’ is -20% to -59%, ‘scanty’ is -60% to -99%, and ‘no rain’ is -100%. These categories can mislead more than they inform, for the difference between an excess of +21% and an excess of +41% can be the difference between water enough to puddle rice fields and a river breaking its banks to ruin those fields.

In today’s concerns that have to do with the impacts of climate change, with the increasing variability of the monsoon season, and especially with the production of food crops, the IMD’s stock measurement ‘product’ is no longer viable. It ought to have been replaced at least a decade ago, for the IMD’s Hydromet Division maintains weekly data by meteorological sub-division and by district. This series of running records compares any given monsoon week’s rainfall, in a district, with the long period average (a 50-year period). Such fineness of detail must be matched by a measuring range-finder with appropriate interpretive indicators. That is why the ‘no rain’, ‘scanty’, ‘deficient’, ‘normal’ or ‘excess’ group of legacy measures must now be replaced. In its place an indicator of eleven grades translates the trends, patterns and messages in IMD’s district-level rainfall data into meaningful and actionable signals.

– Rahul Goswami

Notes

The new 11-grade indicator for assessing weekly rainfall departures in districts uses the same data IMD releases into the public domain, but provides dramatically more useful guidance. This yields the detailed reading required to alert state administrations to drought, drought-like and potential flood conditions. The modified methodology adapts the UN Food and Agriculture Organisation’s ‘Global Information and Early Warning System’ employment of 11 grades.

The weekly tallies of rainfall distribution for meteorological sub-divisions and for states are no longer able to signal administrative action and must be replaced with district-scale and (by 2016 monsoon) urban-scale assessments. The ability of the new 11-grade methodology to provide early warnings of climatic trauma in districts is now clear, and state administrations can respond to growing climatological distress in a targeted manner. Districts and blocks need to be supplied rainfall trends – and not only distribution data – that help farmers and administrators alike better plan for rainfall variability.

A monsoon in more than two halves

August 4, 2015 by Climate portal editor Leave a Comment

Less rain for the remaining two months of the typical monsoon season of four months, but an overall season average that remains as it was forecast in June. This is the confusing monsoon update issued by the Earth System Science Organization (ESSO), the Ministry of Earth Sciences (MoES) and the India Meteorological Department (IMD).

There are aspects of IMD’s treatment of the monsoon season that need correction in our view. One is the long range forecast and its updates. Specific to this update, we are at the halfway stage of what is typically considered the four month monsoon (this too needs revision, as April and May rains were not the usual ‘unseasonal’ passing showers). However, any downward revision of the rainfall average for August and September ought to be an overall downward revision of the season, particularly as June-July have seen very uneven rain.

Consider the highlights of the updated monsoon forecast:

* Quantitatively, the rainfall over the country as a whole during the second half of the season (August to September) is likely to be 84% of LPA with a model error of ±8%.
* The rainfall during August is likely to be 90 ± 9% of LPA as was forecasted in June.
* The season (June to September) rainfall over the country as a whole is likely to be 88% of LPA with a model error of ±4% as was forecasted in June.

We make our criticism constructively, for a significant amount of the material India Climate Portal puts out through our website and our twitter account is taken from the public products released by IMD, ISRO and the Ministry of Earth Sciences, and we fully appreciate the quality of work and commitment of these agencies.

The national mean rainfall (“country as a whole”, as the IMD forecasts call it) must be abandoned as it does not represent the meteorological diversity of a very large country. Each of the 36 met sub-divisions is affected in different ways by the El Nino Southern Oscillation, the Indian Ocean dipole, the Madden-Julian Oscillation and other hemispheric phenomena.

There is no need for this simplification, which in fact achieves the opposite of timely accuracy.
The media in particular (television and radio, print, online) look for an overall message and, without guidance from authorities, picks up ‘top line’ messages that are of little or no use at the district and taluka level, and also for towns and cities. The question for IMD is rather: how will variability in monsoon together with the strengthening El Nino affect local outlooks for August to October. That is why we advocate monthly outlooks for the 36 met sub-divisions, to begin in May and to run until October (that is, half the year and not a third of the year), primarily to prepare local administrations for all possible scenarios.

There is no reason why this cannot be the approach. The Ministry of Earth Sciences coordinates the observation network (satellites included, and our agencies ISRO and NRSC are heavily involved), the IMD uses these data together with a very extensive network of weather stations all over India. The output is excellent quality and in the public domain. Because the meteorological services in India have historically been designed to aid and guide agriculture and cultivation, the agri-met bulletins, alerts and products are copious. Hence IMD/MoES listens to the needs of the agricultural departments and, more recently, disaster management agencies. Unfortunately, the interface with public is still minimal, which this central government can also easily remedy.

Being prudent about forecasts

June 3, 2015 by Climate portal editor Leave a Comment

The Earth System Science Organisation (ESSO), Ministry of Earth Sciences (MoES) and the India Meteorological Department (IMD) released their second long range forecast for monsoon 2015 on 2 June. The ‘headline’ message is that rainfall for the June to September monsoon season is very likely to be 88% of what is normal for the season.

The forecast has been seized upon by various quarters as having serious implications for the production of crop staples (and therefore for food security), for farmers’ livelihoods, for consumer prices and for the availability of water. These are all valid and important aspects that depend entirely or substantially upon the summer monsoon.

But, the IMD, the ESSO and the MoES do not make statements and forecasts on these aspects. They are concerned with what the climatological data and signs point to, and that is what they have told us. How the forecast relates to important aspects of food, farm incomes, water resources and food stocks relies on interpretations. Our advice – to the media, to government agencies and to the private sector – is to go slow on drawing conclusions and when conclusions are required, to make them incrementally.

The wettest pre-monsoon season (March to May) for five years.

Using the handy graphic here, (887KB) we also point out that the pre-monsoon season (March to May) for 2015 has been the wettest in five years. In several meteorological sub-divisions, excess rain has been recorded during this pre-monsoon season. In several districts, the annual rainfall total has already been reached, even before the typical monsoon season of June to September.

This ought to be warning enough to us to be sparing with deciding how forecasts will affect us. The ESSO, IMD and MoES have repeated, in their second long range forecast, that 2015 is an El Niño year which only means that as this sea temperature phenomenon waxes and wanes though the remainder of 2015, so too will the monsoon system react.

It is best to judge our Indian summer monsoon a week at a time, keeping in mind crop calendars and how much water our reservoirs hold. It is always prudent to take precautions such as rationing water (even when it is raining), especially in towns and cities. Likewise, district administrations will do well to assess their local supplies of food staples and match these figures with what food staples their districts are likely to produce during a monsoon whose reliability has now been written off.

The second long range forecast for monsoon 2015 is available here, and the Hindi text can be found here.

Rain, climate, agriculture and Haryana

May 4, 2015 by Climate portal editor Leave a Comment

Every one of Haryana’s 21 districts received excess rain for the period 1 March to 30 April 2015. As these rains have destroyed crops, including food staples, the need to compensate the affected farming families is now paramount. Relief and support are only useful when they are arrive quickly, and unlike administrative conditions two generations ago, state governments and district collectors today can consult data around the clock about conditions in districts and blocks.

Our infographic shows why the March and April rains in Haryana have had destructive effects. The average rainfall in a district of Haryana during this two-month period was 111.6 millimetres – most were in the range of 148 mm (like Ambala) and 81 mm (like Mahendragarh). However, the average for the two months, March and April, is 21.2 mm (Panipat and Faridabad are usually closest to this average).

Although the level of detail available at the district, and indeed even at the weather station level, is comprehensive, the Indian Meteorological Department’s rainfall quantity categories are not granular enough to describe what Haryana has experienced. (This is so for every state that has recorded what is called “unseasonal rain” in these months.)

Amongst the questions that remain unanswered concerning the ‘unseasonal rains’ phenomena – common to Haryana, eastern and western Uttar Pradesh, Bihar, Madhya Pradash, Rajasthan and Gujarat – is the matter of how much rain becomes excess over how many days? This meteorological sub-division (Haryana, Chandigarh and Delhi) normally receives 165 mm in July and 173 mm in August. Therefore, an average amount of 111.6 mm over two months is not a quantity that has surprised the agricultural community.

The matter is one of timing – which lapse the state administration needs to explain, as the weather forecasters first alert state administrations, which then relay alerts and advisories to district administrative staff. Between the claims for compensation for destroyed crops and the political point-scoring, the state government of Haryana has side-stepped the question: what did it do with the weather advisories it was given in March and April 2015?

For our series on the changing rainfall and climate patterns (the Haryana map is the first), we use six categories that begin with 10% above normal and extend to 1000% above normal. This method does better at identifying districts in which agriculture has been hit harder by unseasonal rains and stormy weather.

It’s to be a 93% monsoon says the IMD

April 22, 2015 by Climate portal editor 1 Comment

The India Meteorological Department has just released it’s long-awaited forecast for the 2015 Indian monsoon. In terms of the quantity of rainfall over the duration of the monsoon season (June to September) the IMD has said it will be 93% of the ‘Long Period Average’. This average is based on the years 1951-2000.

What this means is the ‘national’ average rainfall over the monsoon season for India is considered to be 89 centimetres, or 890 millimetres. So, based on the conditions calculated till today, the ‘national’ average rainfall for the June to September monsoon season is likely to be 830 millimetres.

There are caveats and conditions. The first is that the 93% forecast is to be applied to the long period average for each of the 36 meteorological sub-divisions, and a ‘national average’ does not in fact have much meaning without considerable localisation. The second is that the forecasting methodology itself comes with a plus-minus caution. There is “a model error of ± 5%” is the IMD’s caution.

This first forecast and the model that the forecast percentage has emerged from are thanks to the efforts of the Earth System Science Organization (ESSO), under the Ministry of Earth Sciences (MoES), and the India Meteorological Department (IMD), which is the principal government agency in all matters relating to meteorology. This is what the IMD calls a first-stage forecast.

As with all complex models, this one comes with several considerations. The ESSO, through the Indian Institute of Tropical Meteorology (IITM, which is in Pune), also runs what it calls an ‘Experimental Coupled Dynamical Model Forecasting System’. According to this, the monsoon rainfall during the 2015 monsoon season (June to September) averaged over India “is likely to be 91% ±5% of long period model average”. (The IMD forecast is available here, and in Hindi here.)

This is a lower figure than the 93% headline issued by the IMD. This too should be read with care as there are five “category probability forecasts” that are calculated – deficient, below normal, normal, above normal and excess. Each is accompanied by a forecast probability and a climatological probability (see the table). The maximum forecast probability of 35% is for a below normal monsoon, while the maximum climatological probability is for a normal monsoon.

As before, time will tell and the IMD will issue its second long range forecast in June 2015. Our advice to the Ministry of Earth Sciences and to the IMD is to issue its second long range forecast a month from now, in May, and also to confirm these forecasts two months hence in June, when monsoon 2015 will hopefully be active all over the peninsula.

Follow the highs and lows of monsoon 2015

April 22, 2015 by Climate portal editor Leave a Comment

Our coverage of the ‘mausam’, the Indian summer monsoon of 2015, has begun. The unseasonal rains of March and April, which have proved so destructive to farmers, have shown why the conventional monsoon season must be widened. You will find all monsoon-related analysis, data and reports here.

We provide short, focused updates on weather trends. We strengthen the citizen’s understanding of the effects and impacts of climate change with relevant and jargon-free commentary.

We complement the Government of India’s excellent climate and weather monitoring services by advising what you can expect in your district or city, from unseasonal rains or the lack of it. Read our status reporting and analysis here on the India Climate Portal and follow our active twitter feed.