Burning less, growing more: Why NASA is betting on Mitti Labs.

Tech for good: fire, farming, and the future of rice stubble management

Each winter, a thick, toxic haze settles over northern India. The skies above Punjab and Delhi turn gray. Schools close. Flights are grounded. Millions struggle to breathe.

At the center of this crisis is an agricultural dilemma. After the rice harvest, farmers are left with a field full of straw and just two weeks to clear it before planting wheat. The fastest solution? Burn it. Every year, this practice repeats across millions of acres. And every year, the costs mount. It is estimated that nearly half of the total annual PM2.5 pollution load comes to Delhi during the peak season comes from stubble burning.

But what if we could see the fires and quantify the emissions? Understand where they happen, why they happen, and what could replace them? What if we could reward farmers not for speed, but for sustainability?

That is the future Mitti Labs is building. And now, NASA is backing us to do it.

Why NASA trusts Mitti Labs

NASA already operates global fire monitoring programs, but they’ve been limited by coarse-resolution models and patchy field data. NASA largely relies on MODIS (MCD14ML) and VIIRS (VNP14IMGML) for active fire locations and burned areas. These are very consistent data streams and help support global, coarse-resolution products including the Global Fire Emissions Database at gridded monthly 0.25º resolution. However, these sensors struggle with small fields, confusion with tilling practices and flooding activities, and inability to precisely determine residue characteristics.

That is where Mitti’s technology shines. Mitti Labs uses a stack of high-resolution, multi-frequency satellite imagery – including synthetic aperture radar (SAR), thermal, and optical data and combines it with deep learning to track crop burning at the level of individual fields. Our technology tracks fields using high resolution X-band, stripmap quad pol L-band, dual pol C-band, and time series optical to characterize burn types, such as partial, pile, or broadcast burning, as well as estimate biomass and fuel loads. The team deploys physical retrievals that fuse C-band SAR, where parameters reflect relatively strong surface scattering and high extinction characteristics, with L-band that reflects the enhanced penetration capabilities and different scattering mechanisms, to estimate residue biomass during harvest. The prototype uses cross-frequency attention that enables dynamic feature weighting based on vegetation conditions, allowing networks to adaptively emphasize the most informative frequency combinations for different biomass levels. 

Mitti Labs then uses these retrievals to drive a convolution neural net architecture that tracks cropland burning type, timing, and intensity at field scale across entire landscapes with over 90% accuracy (precision, recall, F1 >0.90). They complement this with on-the-ground observations of phenotypic parameters, residue combustion, tillage, and soil microbiota sequencing pre- and post- burn to quantify impacts and emissions. This innovative technology is enabling a level of accuracy NASA wants — and until now, hasn’t had. The team is now expanding the prototype to work with USDA and EPA on national inventories. 

“It’s a validation of our science, our mission, and our ability to drive impact at scale. This partnership is helping NASA refine its own models, while empowering Mitti Labs to build a Digital Twin that governments, NGOs, and farmers can use to reduce cropland burning — and emissions — at scale.” Mitti Labs co-founder Nate Torbick.

But this isn’t just about data collection. Mitti’s tools are being deployed to enhance the integrity of carbon credits and grow climate-smart practices. Now this technology is helping governments, NGOs, and the private sector understand what interventions are working, and where support is needed most.

Burning less, growing more

It’s easy to talk about pollution from a distance. But up close, these are deeply human decisions.

Over three million farmers across Punjab and Haryana grow rice-wheat in rotation. Many are smallholders working under tight constraints: short planting windows, declining groundwater, unreliable subsidies. Burning straw isn’t ideal, but for many, it’s the only viable option. And this means nearly 23 Mn tons of residue burning each season.

Mitti’s work meets these farmers where they are. By partnering with trusted local NGOs, we build long-term relationships grounded in trust and respect. Our team doesn’t parachute in with tech. We listen first — to village leaders, to local norms, to cultural context. We want to make agricultural sustainability the smarter, easier choice.

Technology rooted in nature and community

From a satellite’s eye to a farmer’s field, Mitti Labs works across scales. But our center of gravity is always the same: the land and farmer.

We build our tools not just to track change, but to enable it. Our goal is to stop burning on 500,000 farms in the next five years through alignment. Aligning incentives. Aligning practices. Aligning science with culture, and technology with trust.

What makes Mitti’s work powerful isn’t just its precision. It’s the fact that we connect three of the most urgent challenges of our time — climate, food, and equity — and offer a pathway forward that’s practical, measurable, and hopeful.

The Mitti Labs team is deeply grateful to NASA for our pioneering partnership towards promoting science-backed Climate-smart stubble management practices. This is what tech for good looks like: grounded in nature, guided by science, and built with — not for — the communities it serves.