India’s farmers forecasted the monsoon long before satellites existed. One of them told me exactly how.

I spent a day with a 74-year-old farmer near Barmer who predicted the 2022 monsoon better than the regional IMD forecast. No instruments. No app. Just a tree, some ants, and four generations of accumulated observation. It changed how I think about climate knowledge entirely.

The day Bhagwan Singh embarrassed the meteorologists

I almost didn’t go. The drive from Jodhpur to the village where Bhagwan Singh lives takes about three hours on roads that the word ‘road’ flatters somewhat. My translator, a young agricultural extension worker named Ramesh, had told me we were going to meet a farmer who ‘knows things about rain.’ I was expecting folk wisdom and polite conversation.

What I got was a man who sat under a Khejri tree, pointed at an Arjun sapling twenty feet away, and told me the southwest monsoon that year would arrive ten days late, come in lighter than average during the first week, and that I shouldn’t trust anything the sky told me in the second week of July.

I wrote it down. I checked it later. He was right on all three counts.

The IMD district forecast for that period said ‘normal to below normal rainfall.’ It said nothing about timing or distribution within the season. Bhagwan Singh gave me a timeline, a calibration for intensity, and a specific week to watch. He was 74 years old and had never owned a smartphone.

That’s not folk wisdom. That’s a more granular forecast than the official one. The only honest question is why we don’t take it seriously.

How did he do it? The Arjun tree, he explained through Ramesh, does something specific in the weeks before the monsoon arrives. Its leaves begin to curl upward at the edges – a response to shifts in atmospheric humidity that the tree has evolved to detect over millions of years. When that curling happens, and in what pattern, tells him something about when the rain is coming. When the leaves stay flat, it’s a different story.

He also watches ants. A specific kind, near the base of the tree. Whether they’re moving their eggs upward or downward in the soil tells him about pressure changes underground – the kind of changes that precede significant rainfall by days. He described the two patterns with his hands, demonstrating the difference, and the precision of his description was striking. This was not vague. This was specific, repeatable, observable.

He learned it from his father. His father learned it from his grandfather. The system has been tested on the same land, by the same family, across at least four generations. That’s a hundred years of iterative field research on a specific microclimate.

What Bhagwan Singh is actually doing

Here’s the thing I didn’t understand before that visit: Bhagwan Singh isn’t reading signs. He’s reading data.

The Arjun tree’s leaf behaviour reflects real atmospheric conditions – shifts in humidity and temperature differentials that precede monsoon onset. The tree is responding to physics. Bhagwan Singh has learned to read the tree’s response. The information chain is: atmosphere changes → tree responds → farmer reads tree. That’s not mysticism. That’s an observation system with a biological sensor in the middle.

The same logic applies to the ants. Changes in soil pressure and moisture that precede heavy rainfall trigger specific behavioural responses in ant colonies. The farmer observing those responses is, in effect, using the ants as instruments. Remarkably sensitive, highly localised instruments that no weather station near Barmer can currently replicate.

Researchers at agricultural universities in Andhra Pradesh, Rajasthan, and Maharashtra have been documenting traditional bio-indicators for about fifteen years now. A 2019 study in the Indian Journal of Traditional Knowledge catalogued 116 such indicators from across the country and assessed the scientific basis of each. The majority had clear mechanistic explanations. The Arjun tree’s behaviour reflects real atmospheric physics. The Indian roller’s pre-dawn call correlates with barometric pressure changes. The red ant movement reflects soil moisture shifts.

None of this means traditional forecasting is always accurate, or more accurate than modern meteorology. It isn’t. But for specific microclimates, for the kind of localised prediction that matters to a farmer deciding whether to sow this week or next, it can be more useful than a regional forecast that averages across an entire district.

The Nakshatra calendar: older than the IMD by about 2,400 years

Bhagwan Singh also mentioned the Nakshatras. He didn’t call them that – he used the local terms – but he was describing the lunar mansion system that divides the Indian agricultural year into 27 periods of 13 days each, each with associated weather patterns and agricultural guidance.

The Nakshatra system is documented in the Krishi Parashara, an agricultural text estimated at around 400 BC. That’s roughly the time when Aristotle was writing about weather in Greece. The Indian text is more specific, more practical, and more focused on agricultural decision-making. It gives you not just a description of climate patterns but a set of instructions for what to do about them.

I find this genuinely extraordinary. Two and a half thousand years ago, Indian farmers had developed a systematic, documented framework for making agricultural decisions based on seasonal weather patterns. And we spent most of the 20th century treating it as superstition.

Five conversations I’ve had that changed my mind about this

Since that day with Bhagwan Singh, I’ve been deliberately seeking out similar conversations. Here are five that stuck with me.

Barmer, Rajasthan – the beekeeper who reads the desert

About eight kilometres from Bhagwan Singh’s village, I met a woman named Parvati who keeps bees and also manages a small herd of goats. She told me that when her bees build their honeycomb downward – elongating the cells toward the ground – she knows dry conditions are coming and she needs to move the herd closer to the seasonal river. When they build upward, the rains are approaching.

I couldn’t verify this. I don’t know if it’s always accurate. But she’d been doing it for thirty years, her mother had done it before her, and she described it not as a belief but as a system she’d tested and could explain: bees respond to humidity and temperature changes that affect the wax, which affects how they build. Again – biological sensor, observational reader, practical output.

Vidarbha, Maharashtra – the Indian roller and the monsoon

A cotton farmer in Yavatmal told me that the Indian roller – the Neelkanth, the bird that appears on the dashami of Dussehra – calls three times before dawn in a particular way when the southwest monsoon is seven to ten days away. Not its regular call. A different register.

I’ve since read that this corresponds to barometric pressure changes the bird detects with its inner ear before any human instrument in the vicinity registers them. The farmer doesn’t know that. He knows the call. The mechanism doesn’t matter to him. The forecast does.

Ziro Valley, Arunachal Pradesh – the dragonfly and the planting window

The Apatani people of Ziro Valley have one of the most productive traditional farming systems I’ve ever read about – a rice-fish-millet integrated polyculture that UNDP has recognised as a model of sustainable agriculture. The timing of the rice transplanting is determined, in part, by the emergence of a specific species of dragonfly.

When I asked an Apatani elder why the dragonfly, he looked at me with the kind of patience you reserve for a slow student. The dragonfly emerges when certain soil conditions are met, he said. Those conditions are the right conditions for transplanting. The dragonfly is telling you the soil is ready. He wasn’t wrong – the dragonfly’s emergence correlates with soil temperature and moisture reaching thresholds that favour transplanting success.

Odisha coast – the swallow and the cyclone

A fishing community leader in Puri told me that before the 1999 super cyclone, the white-bellied swallows had been behaving strangely for four days. They weren’t feeding normally. They were moving inland in an unusual pattern. His grandfather, who’d lived through the 1971 cyclone, recognised it. He told the community to move. Several families who listened survived. Several who didn’t, didn’t.

The Centre for Science and Environment has documented other cases where traditional early warning – bird behaviour, wave patterns, sky colour at specific times – preceded official cyclone advisories by 48 to 72 hours. Not because the traditional forecasters are clairvoyant, but because the biological systems they’re reading respond to atmospheric changes faster than the monitoring networks in those areas can detect them.

Barmer again – the conversation that depressed me

On my way back through Barmer, I stopped and talked to Bhagwan Singh’s son, Mahendra, who was visiting from Jodhpur. He’s 42, works in construction, has a smartphone with a weather app.

I asked him if he’d learned his father’s system. He hadn’t. He’d been away for most of his twenties and thirties. He’d meant to learn it. Time had passed.

I asked him if his children would learn it. He shrugged. They were in school in Jodhpur. They didn’t come to the village much.

That conversation is why we’re building what we’re building at ClimateVarsity. The knowledge in Bhagwan Singh’s head is irreplaceable. And it is not going to survive in the default trajectory of Indian modernisation.

Why this knowledge is disappearing – and why the timing is terrible

There are three things I’d point to when asked why traditional climate knowledge is vanishing from rural India.

The first is the Green Revolution’s disruption of agricultural practice. When Indian farming shifted to high-yield varieties, synthetic inputs, and irrigation-based systems in the 1960s and 70s, the intergenerational context for transmitting traditional ecological knowledge was broken. Farmers who adopted the new system were farming differently from their parents. The old knowledge was less immediately applicable. A generation later, less of it got passed on.

The second is urbanisation. Mahendra in Jodhpur is not an exception – he’s the rule. The children who would have been the next generation of traditional knowledge holders are in cities now. This is broadly good for them economically. It is bad for the knowledge.

The third – and this one irritates me – is institutional dismissal. For decades, agricultural extension services, universities, and government programmes treated traditional practices as things to replace rather than things to learn from. The assumption that modern science superseded indigenous observation wasn’t just intellectually lazy – it was factually wrong in many cases, and the cost is that we failed to document what we were throwing away.

We had centuries of longitudinal field research across thousands of microclimates, conducted by people whose livelihoods depended on getting it right. And we called it superstition.

The timing of this loss is particularly bad. India’s climate is becoming less predictable, not more. The patterns that modern meteorological models were trained on are shifting. Extreme events are increasing. The conditions in which regional forecasting models were built – conditions of relative variability within known bounds – are changing.

The communities that evolved their knowledge systems in conditions of high variability – the desert communities of Rajasthan, the coastal communities of Odisha, the high-altitude communities of the Himalayas – are potentially more useful guides to navigating climate disruption than we’ve understood. They’ve done it before. They know what strategies survive.

We don’t have time to rediscover that by starting from scratch.

What serious integration actually looks like – and what it doesn’t

I want to be careful here, because ‘integrating traditional knowledge with modern science’ is a sentence that sounds good and gets used to mean very different things.

It doesn’t mean treating everything farmers believe as equally valid. Some traditional practices are based on superstition, some are harmful, and some simply don’t work. The point isn’t uncritical acceptance. The point is serious engagement: document it, test it, understand the mechanism, and build on what’s real.

It doesn’t mean making traditional knowledge a museum exhibit. The framing that traditional knowledge is ‘cultural heritage to be preserved’ is well-intentioned but wrong. It implies the knowledge is historical, passive, and primarily of value to the communities that hold it. That’s not what Bhagwan Singh’s forecasting system is. It’s operationally useful, right now, in conditions of climate change.

What it actually means in practice

Researchers at the National Institute of Hydrology have started piloting what they call ‘indigenous knowledge-informed drought monitoring’ – which is exactly what it sounds like. Community members systematically document bio-indicator observations. Those observations are combined with satellite data and IMD records to produce composite assessments of drought risk at the local level. Early results, from pilot sites in Rajasthan and Maharashtra, suggest the composite assessments are more locally actionable than either traditional or modern data alone.

This is the model. Not traditional knowledge replacing modern science. Not modern science validating traditional knowledge from a position of superiority. Both systems observing at different scales, capturing different signals, combining into something more useful than either alone.

The IPCC’s Sixth Assessment Report explicitly acknowledges this. It states – and this is in the report, not in a press release about the report – that indigenous and local knowledge systems hold key insights for climate adaptation, particularly where scientific data coverage is sparse. That’s a formal scientific body saying: there is knowledge in communities that we should be using.

What we’re doing about it at ClimateVarsity

When I started ClimateVarsity, I knew I wanted traditional ecological knowledge in the curriculum. What I underestimated was how much resistance that would meet – not from communities, but from educated, well-meaning people who worried it would undermine scientific credibility.

My answer to that concern is Bhagwan Singh. He’s not undermining scientific credibility. He’s extending it. He’s doing something that the nearest weather station can’t do: providing a hyperlocal, phenologically grounded, multi-generational observational record of climate patterns in a specific microclimate. That’s not a threat to climate science. That’s a data source that climate science hasn’t figured out how to use yet.

In our programmes, we teach traditional bio-indicators alongside modern climate science – not as alternatives, but as complementary observation systems. We ask learners: what do the plants and animals near you tell you about the weather? That question reconnects people with forms of environmental attention that urbanisation and smartphone weather apps have eroded. It also, in communities where traditional knowledge is still alive, gives dignity and credibility to what elders know.

We also document. Every programme we run is an opportunity to capture traditional climate knowledge from the communities we work with. It’s not systematic enough yet – that’s something we’re building toward – but we’ve started.

And we’re explicit with our CSR and development funding partners that this isn’t a soft add-on. Traditional ecological knowledge is a climate adaptation resource. Losing it has a measurable cost in resilience. Preserving and integrating it has a measurable benefit. We’re making that case with the same rigour we apply to everything else.

Back to Bhagwan Singh

I’ve been back to Barmer once since that first visit. Bhagwan Singh was in good health. He told me the 2023 monsoon would be erratic – ‘confused,’ was how Ramesh translated it – with a false start in June and the real rain not arriving until mid-July.

He was, again, roughly right.

His son Mahendra happened to be there that day. He sat with us for a while, listening. At one point, he asked his father to show him the Arjun tree again – which leaves to watch, what to look for. Bhagwan Singh showed him.

I don’t know if Mahendra will remember it. I don’t know if he’ll ever have occasion to use it. But I noticed that he wrote it down in his phone. Which felt, in a small way, like something.

The question isn’t whether traditional climate knowledge is valuable. It obviously is. The question is whether we’ll get out of our own way long enough to preserve it.