We use AI, drones, science and community knowledge to help farmers become more climate resilient

Nearly 30% of India's land is now considered degraded - a problem that affects over 500 million people who are dependant directly or indirectly on agriculture and forestry for their livelihoods in India. Land degradation is a result of

1. Excessive inorganic fertilizer and pesticide use
2. Deforestation 
3. Diversion of healthy soil to other activities like construction 

While India is struggling with land degradation, it is also struggling with climate change. Agrarian and indigenous communities in India are amongst those worst affected by climate change - they are facing declining yields and incomes because of erratic rainfall, frequent pest attacks, forest fires, depleting soil fertility and groundwater tables.

At the same time declining soil fertility has meant that the food that we are growing is lower in nutrition and quality. Monocultures of cereal crops, soybeans and sugarcane have reduced biodiversity in our soil and also in our diets. One study from America estimated a 38% decline in the nutritional value of several vegetables since the 1950 (Davis, 2004). 

Sources:

1. https://india.mongabay.com/202...
2. https://pubmed.ncbi.nlm.nih.go...
3. https://pulitzercenter.org/sto...

We help farmers and indigenous communities transition to multi-layer agroforestry and use the drone and AI-based quantification of this change to raise funds for the transition from philanthropy, government and the carbon markets. 

Farmers understand that agroforestry is better for the soil, requires less inputs and is more climate-resilient than traditional monoculture farming. However they lack access to finance to make the transition, the know-how on what to plant and what to intercrop it with and information on what to do incase of a pest attack or declining yields. 

That's where Farmers for Forests and our technology comes in. In the first step we use satellite data, farmer surveys and on-ground soil samples to work with the community on what to plant as the main tree species and the intercrop. 

We then help communities procure high quality saplings, seeds, biofertilizers and drip irrigation at a subsidised rate, remove the financial barriers they face for transition. The use of drip irrigation systems and biofertilizers means farmers now need 80% less water to grow their crops. 

Our agroforestry plots are typically 3 times more biodiverse and sequester 4 times more carbon than a plot of traditional agriculture. They also help rejuvenate the soil through the accumulation of leaf litter, organic fertiliser and farm made fertilisers like jeev amrut. 

Once the plantation is done, we carry our frequent drone surveys, including with multi-spectral drones. We are then able to analyze the drone information using neural networks to tell us: 

1. How many trees have been planted? 
2. What is the geo-location of each tree?
3. How many metres of drip irrigation systems have been deployed? 
4. What is the species of trees that have been planted?
5. How healthy these trees and crops are? Are there any specific trees that are in distress and need intervention?
6. What is the height of these trees?
7. What is their diameter? 
8. What is their crown size? 

We are then able to use this information to:

1. Provide advisory to farmers transitioning to agroforestry for the first time and ensure higher tree survival rates and healthier farms
2. Provide financial support to farmers in the form of cash incentives or Payments for Ecosystem Services (PES) 
3. Provide regular and detailed data right down to the tree level on biodiversity, carbon sequestration and health to philanthropists, government and carbon market buyers that are funding this work to increase fund flow back to the projects to help more farmers transition to agroforestry 

We are currently in the process of piloting a few other interventions using the same base level data we collect: 

1. Easier decision making for installation of food processing units based on the geospatial coverage using species and yield data that we are able to predict 
2. Farm to fork traceability 
3. Provision of better quality loans and insurance schemes to farmers based on quantification of their natural assets, yield and carbon sequestration predictions

We work with small and medium sized farmers in 2 states in India. On an average the farmers in our programs have land holding sizes of 1 to 2 Hectares and household incomes of about $2500 / year. 

Participants in our program can get reliable, income support of between  in addition to being able to reap the economic and nutritional benefits from the variety of different fruits, medicinal herbs and agroforestry produce that these trees bear.

Our work also creates numerous green jobs in the communities, especially for women that work primarily as tree planters. More trees on farms also improve groundwater recharge in the village, increase biodiversity and attract a wide range of pollinators from insects to birds to mammals. Several studies point to the beneficial effects that increased biodiversity and pollinators have on crop yields as well

An interesting thing we found was that many of the farmers we work with often practice organic and regenerative practices for the their own kitchen gardens where they grow a variety of different fruits and vegetables, but for their larger farms they mention an inability to do this because of a lack of finance, knowledge and no price premiums available in the market for biodiverse, regeneratively grown food, since technology is not being deployed to independently and regularly verify some of these practices to bring trust in the minds and hearts of consumers and financiers. 

Through our program, we are able to: 

1. Help farmers transition to agroforestry on a portion of their land 
2. Provide regular advisory services and finance in the first 4 years when farmers need it the most 
3. Provide confidence to buyers and funders through a birds-eye of what is happening on the ground 

Our senior leadership team has 2 co-founders, one COO and a team of 7 Senior Managers. Every single person in our senior leadership team is either from the areas where we work, come from a farming family or has at a minimum at least 8 years of experience working in rural India. 

Members in our team have former work experience across a wide variety of sectors that they bring to the organisation: drones & technology, agriculture, finance, forestry and monitoring and evaluation. 

We have a field team of nearly 40 people full-time and often hire about 60 people or so on a part time basis. All of these people are from the communities where we work and have had lived experience with the problem we are trying to solve. 

Our programs are carried out in close consultation with the farmers and indigenous communities we work with. We do this through village level meetings, WhatsApp groups and one on one trainings with farmers. We often take farmers in our programs for exposure visits to meet other farmers in the program, so that there is free flow of information and farmers are inspired to take up new models of agroforestry and share learnings with one-another and us. 

In the course of our work over the last 4.5 years we have found that, for the long-term success of our interventions community participation and knowledge is critical. We can only provide intensive support to farmers for the first 4 years, after that we move to a low-touch model and so farmer and community participation and training has been absolutely critical in ensuring sustainability. 

In first 2.5 year of our program, we carried out several pilots both in terms of technology and what works on the ground that farmers are finding useful. We were able to find product market fit about 2 years ago - both from a farmer perspective and from the perspective of a carbon credit buyer. 

Our program is at the stage where farmers, see tremendous value in our program and are willing to a pay a subsidised rate to be a part of the program and actively seek us out. Our technology is now performing with close to 90% accuracy for tree detection, tree height and tree health and close to 75% accuracy for species detection, biodiversity and carbon measurement. Our methodology is vetted and we are increasing training data to be able to train algorithms to perform at higher accuracy for the bits where accuracy is 75%. 

While overall we are working with over 22,000 farmers across various programs, our drone-measured agroforestry solution is currently being implemented on the ground with nearly 1800 farmers. Over the next 2 years we have secured the funding to scale this program up to nearly 35,000 farmers.  

We are applying for the following reasons:

1. We want to be able to use the Solve network to connect with mentors, peers and company sustainability heads and sustainability enthusiasts across the world!
2. To make a dent in the climate-change and farmer problem in India we need to get multiple partner organisation and governments on board, customising our model to implement in their geographies and communities and using our technology. eBeing a part of the Solve network gives an increased amount of legitimacy and authenticity to our work. 
3. We want to connect with mentors and peers who have been in the same scale journey as us, to help us put in the place the systems - tech, operational, financial and administrative to be really able to implement with high quality at large scale 
4. We want to be able to use the Solve network to partner with professors and researchers doing Geospatial Imaging work, soil biology research, botany and carbon sequestration measurements

Most trees don't show up on even very high resolution satellite data (0.5m) until they are about 4 or 5 year old. But high resolution satellite data is expensive! Meaning that most forestry, agroforestry and restoration projects rely on freely available or lower resolution satellite data that is unable to give farmers they inputs they need for on-ground restoration or plantation work. The first 4-5 years are the most crucial for any project - this is when the trees are most susceptible to mortality and disease. 

Our drone and AI based solution is able to give farmers, donors and carbon credit buyers useful information like location, health, height, species, carbon sequestration. The other solution would have been to use manual monitoring which is time consuming, prone to error and expensive or high resolution satellite imagery which is currently prohibitively expensive. 

We are able to get accurate location and height information for even trees that are as small as 4 feet in height. The image below shows the algorithm detecting trees:

In addition to locating every tree in our program, identify a few different types of species, we are also able to measure height and crown size and use this data to estimate. Something that is hard to do through satellite data: 

By having eyes on the ground across all of our program plots, we are also able to intervene and spot problems early. Additionally multispectral data allows us to give information to the farmers on trees that are showing signs of stress, like the trees on the boundary in the image below:

And we are able to do this at a fraction of the cost it would require high resolution satellite data or manual monitoring. Once high resolution satellite data becomes affordable over the next few years, the same algorithms could be employed on that data, reducing the need of drones and making scale easier. 

Ecological Indicators

1. Carbon Sequestration - as measured through drones and independent auditors
2. Land Use Land Change Patters - as measured through satellite data - primarily PlanetScope and Sentinel 
3. Tree Health / Plot Health - as measured through various indices that rely on the spectral signatures of Red, Blue, Green, Red Edge and Near Infrared Light
4. Biodiversity - as measured through species surveys and biodiversity registers
5. Soil Health - as measured through baseline and periodic soil surveys

Socioeconomic Indicators

1. Income & Asset Levels - as measured through asset rosters and self reported income data 
2. Spending on Nutrition & Nutrient Intake - as measured through self reported surveys 
3. Climate-change awareness & adaptability - as measured through self reported surveys 

1. Our field data, including feedback from farmers and details on their knowledge of DIY organic fertilizers (like Jeev Amrut), intercropping practices, species preferences, demographic information is collected on Mobile Phones using Kobo through our call centre or through in-person visits
2. Our drone data is collected using DJI Mavic Multispectral drones
3. The field data is analysed in R and Python and displayed in an app coded in Angular to our field team and the farmers 
4. Our coding software stack includes R, Python, Angular, PHP, Java Scrip, React and QGIS
5. The drone data is analyzed using Pix4D and Python. Our base algorithm comes from the open source algorithm DeepForest. We have built on this to be able to create further analyses 
6. The data is further analysed in QGIS 
7. We use local cloud storage but are in the process of transitioning to AWS for our dashboard and Detection Interfaces

Full time: 57

Part time: 60+ community members we hire during primary plantation season

We have been working on this specific solution since August 2019 - nearly 4.5 years. 

Both Arti and Krutika (co-founders) have been working in development and development related sectors since 2009. 

All of our co-founders are women. In addition to this several members in our team belong to groups deemed as scheduled castes and scheduled tribes, as recognised by the constitution of India. 

Over 95% of our team comes from agrarian and indigenous families. 

One thing that we are working on correcting is our gender ratio. While we have several women in leadership positions in our team, we are trying to increase the number of women in our field teams. 

We are a hybrid social enterprise and thus use multiple pools of finance to carry out our work. 

There is significant demand for carbon credits from agroforestry and forestry projects given all the co-benefits these projects have, but buyers are often hesitant to invest large amounts upfront in creating supply because of the numerous natural, governmental and land ownership risks that exist. Additionally, buyers find it hard to get aggregated data on the thousands of smallholder farmers that are a part of these projects. 

That's where our innovative finance model comes in! We are using philanthropy, government schemes and equity based financing to fund these projects for the first 4 years and collect drone and other field data on their progress. 

The four years of ground progress coupled with detailed and regular data from drone surveys and analysed by AI is able to significantly de-risk the project - allowing carbon finance to come in and invest support the project and the farmers over the next 20 to 25 years. 

In total we have so far been able to raise about $2 million of philanthropic capital through funders like Accenture, Meta, the Nudge Foundation, Mulago, Fast Forward, CISCO, Rohini Nilekani Philanthropies, Opus Software Solutions, J-PAL, Mount Holyoke College, Rainmatter Foundation and ACT Grants to name a few. 

This philanthropic funding has allowed us to get product market fit, develop technology and also serve thousands of smallholder farmers. 

We have raised a significantly larger amount through equity and carbon finance, the details of which we are unable to reveal because of Non Disclosure Agreements signed, but through this funding we are now able to scale our work over the next 2 years to 35,000 farmers across 15,000 Hectares. We also collect 10-15% of the overall agroforestry implementation costs from farmers. The amount is high enough, that only farmers who are genuinely interested in agroforestry come into the program but also low enough to not exclude smallholder farmers. It provides an additional source of revenue to us and ensures deep investment from farmers in the work. 

As of the Financial Year April 2023 to March 2024, we are profitable.