Early warning must also reach farmers without institutional support [Commentary]

For India, an El Niño year has usually meant a weaker monsoon. That relationship is a tendency rather than a rule, and there have been El Niño years that ended with normal rain, but the odds move, and they move against us.

This year the signal was unusually emphatic. The India Meteorological Department put the probability of El Niño developing during the monsoon season at 92%. The United States National Oceanic and Atmospheric Administration gave it a 96% chance of persisting through the coming winter, and several analyses placed it in the ‘super El Niño territory’, comparable in strength to 2015. IMD’s seasonal outlook was for rainfall at 90% of the long period average, with a 60% likelihood that the season would close with a deficit.

The monsoon followed the warning, and Rajasthan felt it early. By the first week of September, the state was 21% short of normal rainfall, with 21 of its districts deficient and only two in surplus. That put the state in step with a country that recorded a 13.8% shortfall between June and August, and where cumulative rainfall stayed about 15% below the long period average through September. Elsewhere the deficit has already turned into formal declarations, as in Maharashtra, which has declared drought in 265 of its 358 talukas.

However, in the fields, the signal came earlier than any declaration. Nationally, kharif sowing lagged by about 16% on July 10, at 53.12 million hectares against 63.27 million a year earlier. Sowing recovered through July and August, but the rain did not. That gap is the lesson of this season. Every El Niño year will bring deficits and declarations, and it will also bring a forecast months in advance.

The question worth asking is not whether the country knew what was coming, but whether the farmers who had to act on that knowledge received it in a form they could use, before the loss rather than after it. In Rajasthan, where drought is a recurring hazard rather than an exception, that question is sharpest. That led us to our investigation.

How India announces a drought warning

The forecast that reached farmers is the output of a system India has spent nearly five decades building. At the forecasting end, the science has improved steadily and performs well. The larger difficulty lies in the last stretch, between the bulletin and the field, which is the dissemination.

Drought early warning in India rests largely on the Agromet Advisory Services, which were set up in 1977, when IMD began issuing state-level bulletins through All India Radio and the newspapers, the only channels with sufficient rural reach at the time. Doordarshan followed, with programmes such as Krishi Darshan. All of it was one-way, generalised and tied to a broadcast schedule. The service was brought down to the district level in 2008, restructured under the Gramin Krishi Mausam Sewa, and extended to the block level in 2018 in partnership with the Indian Council of Agricultural Research (ICAR). More than 130 Agromet Field Units and a network of District Agromet Units housed at Krishi Vigyan Kendras (KVKs) now generate the advisories.

What a farmer is meant to receive is not a single alert but a sequence that sharpens as the season approaches. At the longest range, six to eight months out, IMD’s Multi-Model Ensemble and its statistical forecasting system indicate the broad character of the coming monsoon. This is where an El Niño signal first enters the system. Medium-range bulletins, issued twice a week, convert five-day forecasts into specific farm operations such as  ‘hold back sowing’, ‘delay the irrigation’, ‘bring the harvest forward’. Short-range warnings for sudden events reach phones in near real time.

The delivery has changed more than the science. Where a bulletin once travelled by radio and newsprint, it now moves through SMS, WhatsApp groups run by Krishi Vigyan Kendras, applications such as Meghdoot, the Kisan Portal, and the social media accounts of state agriculture departments. On paper this is a considerable advance: from generic and delayed to targeted and immediate.

What the warning did this year

Between May and August this year, we spoke with farmers across the geographic zones of Rajasthan, one of India’s historically most drought-exposed states. Half had sustained contact with their local Krishi Vigyan Kendra (KVK) and used digital advisory channels actively. The other half had no such institutional link with a KVK. The two groups differed in how they received the advisories and in how they responded to El Niño.

Among the first group, the El Niño warning did precisely what an early warning system is supposed to do. It arrived early enough to change the shape of the kharif.

“It was an El Niño year, so I didn’t sow the rice all at once. I watched which way the rain was going and sowed it in portions, one section at a time,” said Rampal, a farmer from Sri Ganganagar.

Others changed the crop itself, moving away from anything thirsty.

“This year I didn’t sow crops that need a lot of water. Instead I sowed bajra and guar, because they need less water,” said another farmer Sheeshpal from Pratapgarh. “This year I didn’t plant soybeans. I sowed bajra (pearl millet) in its place,” said Rajpal, a farmer from Baran.

One farmer described something more precise than crop substitution, a calculation that set the advisory against his own irrigation capacity and left him a deliberate margin. In his 25-bigha land, he sowed ten. If the rain failed, the tube well could carry that much; and he had saved the cost of sowing the other fifteen.

Among the second group, the same season in the same districts, under the same advisory zone, produced a different account entirely.

“We are relying on god. We have sown the crop and are waiting for the rain, and the crop is being ruined,” said Girdhari, a farmer from Jaisalmer. This brings us to a striking contrast i.e. one infrastructure and one season, yet opposite outcomes.

Where it breaks

It is not a simple story about access. The two groups fail in different ways, and the two failures do not answer to the same remedy.

Farmers already inside the system do receive their advisories. Their complaint is about quality, which has two parts. The first part is the distance between a forecast’s accuracy, meaning whether it matches what happens, and its precision, meaning the scale at which it is issued. A forecast can be entirely correct at the block level and still be too coarse for the microclimate of one plot, a limitation they attribute to a thin network of automated weather stations.

The second is newer, and is a direct product of the shift to digital delivery. When advisories came over All India Radio, the channel carried the authority. A forecast heard on the state broadcaster was official because of where it appeared. That signal has collapsed. Government advisories, private forecasting services and outright speculation now arrive through the same interface, formatted alike, and the work of telling them apart has been transferred to the farmer.

“Messages come from every direction, government, private, or anyone else, and we just glance at them and delete them. We can’t tell which warning is genuine and which is false,” says Dhanraj, a farmer from Hanumangarh.

For farmers outside the extension network the obstacle arrives much earlier. Patchy network coverage and lack of awareness that such a service exists mean the advisory comes late or not at all. Where it does land, it is often written in technical or non-regional language and is dismissed rather than acted on, which leaves the farmer no better placed than if nothing had been sent. This group is less exposed to open social media, but the same collapse of institutional markers operates on WhatsApp, without the digital fluency to compensate for it.

“Once, a message came on WhatsApp saying the monsoon would be very good this year, and people were saying the same thing. We trusted it and bought more seed. Later it turned dry, and we had to bear a big loss,” complains Mohanlal, a farmer from Jodhpur.

They also rely on traditional knowledge. Reading the onset of rain, cloud behaviour and seasonal indicators persists across both. What differs is its weight. For farmers outside the advisory network it remains the primary basis of decision-making rather than a supplement to the forecast.

What would have to change

The gap here is not a technological one, and it will not close by adding more channels. Early warning dissemination systems built around existing points of institutional contact reach, by default, those who already have institutional contact. The farmers who most need anticipatory information are the ones furthest from the machinery that provides it.

Certain steps would make drought early warning more effective for those who need it most. The first step is to bring the agencies together. Right now, the work is split across IMD, the Ministry of Agriculture and Farmers Welfare, the National Disaster Management Authority and ISRO’s remote sensing centres. Each does its part, but no one is answerable for whether the advisory actually reaches the farmer. The second is to stop sending the same advisory to everyone. A single format will always fail those who are least equipped to use it.

Farmers outside the KVK network need their own route: audio messages in their own language, voice response systems they can call, and outreach through KVKs. This should run alongside the digital advisories as a separate channel, not be added on later. The third is to rebuild trust by letting farmers talk to a person, along the lines of Karnataka’s Varuna Mitra service. It connects a farmer directly with an agrometeorologist, so the farmer is no longer just receiving a broadcast. A real person on the line can give back the sense of authority that advisories lost once they moved onto phones alongside every other message. The fourth is more precision. More automated weather stations would make a forecast that is right for a block usable on a single plot.

India has committed, under Sendai Framework Target G, to substantially increase access to early warning by 2030, and the Prime Minister’s Ten-Point Agenda on disaster risk reduction pairs technology-led dissemination with building local capacity. The first has comfortably outpaced the second.

Aggregate performance conceals this. A drought early warning system serves two purposes at once: securing food production in the aggregate, and protecting the livelihoods of the farmers whose production it rests on. A system can succeed at the first while failing at the second, because national food security requires only that enough farmers be reached, not all of them. Which is exactly why it matters who is reached and who is not.

El Niño 2026 was forecast accurately, months in advance, by a public system that has been building this capability since 1977. For the farmer who staggered the paddy sowing and watched the rain before committing the next plot, the forecast system worked. For the farmer waiting on a crop that is failing, it may as well not have existed. The forecast was the same. The warning was not.

Banner image: Damaged fields in Rajasthan in 2015. Representative image. (AP Photo/Deepak Sharma)

Vandana Choudhary is a project scientist at the National Institute of Disaster Management. Milap Punia is the Vice Chancellor of Maharshi Dayanand University and a professor at the Centre for the Study of Regional Development, Jawaharlal Nehru University.

The findings in the commentary draw from the authors’ ongoing research study on the multi-hazard assessment of heat waves and droughts in Rajasthan. The research findings are yet to be formally published.

Read more: El Niño, the monsoon, and the confusion of Kerala’s fishers [Commentary]