Carbon-smart farming: drones, AI, forecasts and IoT soil sensors for Bangladesh

In an article published on 30 May 2026 in the agriculture and nature section of Jagonews24, agriculture and environment specialist Samiron Biswas made the case for “carbon-smart agriculture”, or carbon technology in farming, in Bangladesh. Climate change, pollution, declining soil fertility and rising production costs had put farming in a difficult position, he wrote, and spray drones, artificial intelligence (AI), weather forecasting and Internet of Things (IoT) soil analysis used together could produce more food at lower cost and with less pollution. He added that the use of these technologies was slowly growing in Bangladesh.

In developing countries such as Bangladesh, the writer argued, agriculture is not only a way of producing food but is tied directly to the environment, the economy and people's lives, so it must become more efficient, technology-driven and environmentally friendly. This report sets out the four technologies he proposed, what international research and Bangladesh's government agencies say about them, and the obstacles to using them in the field.

Biswas described carbon-smart agriculture as a modern farming system that uses technology properly to cut carbon emissions from farming, protect natural resources and raise production at the same time; in his view it is not a single technology but an integrated form of sustainable farming. Conventional farming, he wrote, releases large amounts of carbon dioxide, methane and nitrous oxide through excessive fertiliser, pesticide, irrigation and fuel use, and these gases are among the causes of climate change.

The idea he described is known internationally as climate-smart agriculture. In its 2010 paper “Climate-Smart Agriculture: Policies, Practices and Financing for Food Security, Adaptation and Mitigation”, the Food and Agriculture Organization of the United Nations (FAO) defined it as agriculture that sustainably increases productivity and resilience (adaptation), reduces or removes greenhouse gases (mitigation), and enhances the achievement of national food security and development goals. The same paper said that, on population growth and consumption trends, agricultural production would need to rise by at least 70 percent to meet demand by 2050, while climate change was likely to reduce productivity, production stability and incomes in some areas.

The World Bank's climate-smart agriculture page (last updated in February 2024) calls it an integrated approach to managing landscapes (cropland, livestock, forests and fisheries) that addresses the interlinked challenges of food security and climate change. Its three aims are higher productivity, greater resilience and lower emissions, and it covers climate-resilient varieties, conservation agriculture, agroforestry, precision farming, water management and improved livestock management. The Bank adds that agrifood systems cause about one third of global greenhouse gas emissions yet receive 4 percent of climate finance, with only a fifth of that reaching smallholders.

Wikipedia's article on climate-smart agriculture also records criticism of the concept: some have called it a form of greenwashing for big business, and critics note that no universally accepted standard exists to judge whether those who call themselves climate-smart actually are. Whether a technology or project labelled “carbon-smart” really reduces emissions therefore remains a question to be measured.

The Intergovernmental Panel on Climate Change (IPCC), in the Summary for Policymakers of its 2019 Special Report on Climate Change and Land, said that agriculture, forestry and other land use accounted for around 13 percent of carbon dioxide, 44 percent of methane and 81 percent of nitrous oxide emissions from human activities in 2007-2016. Together that was about 12 billion tonnes of carbon dioxide equivalent a year, or 23 percent of total net human-caused greenhouse gas emissions. Counting pre- and post-production activities across the whole food system, the share rises to 21-37 percent.

The IPCC supports the writer's point about excess fertiliser. Human-caused nitrous oxide emissions from soils, it says, are primarily due to nitrogen application, including inefficiencies such as over-application or application poorly timed to crop demand. Cropland soils emitted around 3 million tonnes of nitrous oxide a year in 2007-2016. On methane, the report names ruminant livestock and the expansion of rice cultivation as important contributors to the rising atmospheric concentration.

The report also lists ways to cut emissions. Cropland practices that help both adaptation and mitigation include increasing soil organic matter, erosion control, improved fertiliser management, improved crop management such as paddy rice management, and heat- and drought-tolerant varieties. It estimates the technical mitigation potential of crop and livestock activities and agroforestry at 2.3-9.6 billion tonnes of carbon dioxide equivalent a year by 2050, and names integrated pest management, rainwater harvesting and precision agriculture among examples of sustainable land management.

Rice is at the centre of Bangladesh's farming, and paddy fields are a major source of emissions. FAO's 2010 paper says about three billion people, roughly half the world's population, eat rice every day, and about 144 million hectares are planted with rice each year. Waterlogged, warm paddy soils make rice a large emitter of methane. Research found that emissions occur mainly in the months when the ground is fully waterlogged, so managing irrigation and fertiliser together has been found to reduce them substantially.

One example is alternate wetting and drying (AWD). FAO says the International Rice Research Institute (IRRI) developed this water-saving technique with national research institutes, and that under the revised 2006 IPCC methodology such multiple aeration potentially cuts methane emissions by 48 percent compared with continuously flooded fields, while also saving irrigation water and raising productivity. Wikipedia's article on agriculture in Bangladesh, citing a report of the Ministry of Environment and Forests, says the Department of Agricultural Extension (DAE) introduced AWD to use water efficiently. The method usually relies on a PVC tube set in the field to watch the water depth, and it suits dry-season rice, when rainfall is low.

A second example is urea deep placement (UDP). In FAO's description, urea in the form of super granules or small briquettes is placed under the soil near the roots, out of the floodwater where it is easily lost. In Bangladesh, the paper says, the practice showed 50-60 percent savings in urea use and yield increases of about one tonne per hectare: a field-tested form of the “less fertiliser, more yield” the writer described.

Biswas called spray drones an important innovation: flying machines that apply fertiliser, pesticide or organic inputs quickly and precisely. Farmers used to walk the fields spraying, spending more time, labour and money and coming into direct contact with toxic chemicals. With drones, he wrote, large areas can be sprayed in a short time and only where needed, less chemical is wasted, farmers face fewer health risks, less fuel means lower emissions, and crops are not trampled.

FAO and the International Telecommunication Union (ITU) published “E-agriculture in Action: Drones for Agriculture” in Bangkok in 2018. It says drone crop spraying was first used in Japan in the 1980s, when unmanned helicopters fitted with spraying equipment and pesticide tanks treated fields. Typical modern spraying drones carry over ten litres of liquid pesticide and discharge over a litre a minute, covering a hectare in about ten minutes. Japan's remote-controlled Yamaha RMAX helicopter had logged over two million flight hours in agricultural spraying, with some 2,500 units treating more than 2.5 million acres a year in Japan.

The publication adds a condition: to use drones fully, spraying must be paired with imaging, image processing and automated analysis so that only affected areas or plants are treated. Done this way, dosage in affected areas improves and overall chemical use falls. Drones with infrared and multispectral sensors can build maps of a vegetation index known as NDVI to find stressed plants, because healthy leaves reflect much near-infrared light while dehydrated or stressed leaves reflect less.

The publication also covers rules. Dropping items from drones is illegal in many countries, it says, and spraying from drones and carrying dangerous goods is highly regulated or prohibited; privacy, cyber security, insurance and liability must also be settled before flights. Anyone planning drone spraying therefore needs the civil aviation authority's permission and rules, and the upazila agriculture officer is the right person to advise which pesticide a crop needs and in what amount.

The writer described AI as a tool for data-based decisions. He gave the example of Dr. Chashi, a smart farming platform that, in his account, was helping farmers with disease diagnosis, weather information, fertiliser management and crop care. AI works through mobile apps, image analysis, data collection and machine learning, he wrote; a farmer sends a photo of the crop and the AI quickly identifies the disease and suggests a solution, speeding diagnosis, reducing excess fertiliser and pesticide and saving cost, labour and pollution. Young farmers and agri-entrepreneurs were showing interest in digital farming, he added, and AI-based initiatives such as Dr. Chashi could take the country's agriculture to a new level.

Peer-reviewed research has tested the idea. In a 2016 study in Frontiers in Plant Science, Sharada Mohanty, David Hughes and Marcel Salathé trained a deep learning model on 54,306 images of healthy and diseased leaves from the PlantVillage project to identify 14 crop species and 26 diseases. It reached 99.35 percent accuracy on a held-out test set, correctly classifying 993 of every 1,000 images. Training took hours, but classifying an image took under a second, fast enough for a smartphone.

The same study set out the limits. Tested on images taken under conditions different from the training images, accuracy fell to just above 31 percent. The authors stressed that the approach supplements rather than replaces existing diagnosis: laboratory tests are always more reliable than diagnosis from visual symptoms, and early-stage disease is often hard to see. An app's advice is best checked with an agriculture officer.

FAO's 2019 briefing paper “Digital Technologies in Agriculture and Rural Areas” likewise says AI can improve resource use, support early decisions through predictive models and keep up round-the-clock monitoring, letting farmers use satellite or drone data instead of walking the whole farm.

Bangladesh's farming depends heavily on the weather, the writer noted, and erratic rain, drought, cyclones and extreme heat cause heavy losses. Forecasts tell farmers when rain will come, when to irrigate, when spraying is suitable and whether storms or floods are likely, which, he wrote, cuts unnecessary irrigation and pesticide waste, reduces crop losses, saves water and fuel and limits farmers' financial losses.

A government service for this already existed. The background page of the Department of Agricultural Extension's Bangladesh Agro-Meteorological Information System (BAMIS) says that although Bangladesh is among the countries most exposed to climate change and disasters, it lagged in getting weather, river and early-warning information to farmers. In 2012 the World Meteorological Organization held a technical seminar at which experts identified three needs: better agro-meteorological observation and seasonal forecasting by meteorologists and agronomists together, a fast system to deliver advice to the grassroots, and training so that the advice is acted on.

According to BAMIS, the Bangladesh Weather and Climate Services Regional Project, financed by the World Bank, was approved in 2017 for the Bangladesh Meteorological Department, the Bangladesh Water Development Board and the DAE. Under it the DAE ran the Agro-Meteorological Information Systems Development Project from July 2016 to June 2021. The BAMIS portal it built takes weather data from the Meteorological Department and river data from the Water Development Board, has a technical committee check and translate them, and supplies them to farmers in usable form.

The portal says agro-meteorological bulletins are updated twice a week for the 64 districts and once a week at national level, alongside information on crops' sensitivity to weather, the links between weather and pest and disease attacks and their control, a crop-weather calendar, extreme-weather alerts, and special advisories for livestock, poultry and fish. FAO's 2019 briefing paper similarly says timely weather-based agro-advisory messages can help farmers anticipate and respond to pest attacks, crop failures and climatic changes.

Soil health is vital, the writer said, yet farmers often apply too much fertiliser without knowing it, damaging the soil and adding to pollution. IoT soil analysis uses sensors to measure soil moisture, pH, nutrients and temperature, he wrote, showing the soil's real needs so that the right amount of fertiliser is applied, chemical use falls, fertility is kept, water is saved and output rises.

FAO's 2019 briefing paper describes precision agriculture as an application of the Internet of Things in farming. Guidance systems during planting and fertiliser application can save seed, fertiliser and tractor fuel and cut hours in the field, it says, and variable rate technologies and drones can reduce water and pesticide use as well as labour and resource costs. The FAO-ITU drone publication adds that precision farming combines sensor data and imaging with real-time analysis to map variation within a field, supporting soil health scans, irrigation scheduling and yield estimates.

The environmental case for the right dose is plain in the IPCC's findings: over-applied or badly timed nitrogen is the main source of nitrous oxide from soils. FAO's 2010 paper also says better management of soil fertility and organic matter, and more efficient use of nutrients, let farmers produce more with proportionally less fertiliser. Sensors are one way to do the arithmetic, but soil test results and the upazila agriculture office's recommendations remain the farmer's main guide to how much of which fertiliser a field needs.

Bangladesh is an agricultural country, Biswas wrote, but climate change has brought rising salinity, erratic rainfall, drought and floods, falling soil fertility and higher production costs. Carbon-smart agriculture is very important for meeting these challenges, he argued, and technology-based farming would not only raise output but also help protect the environment.

Wikipedia's article on agriculture in Bangladesh says farming is the country's largest employer: it made up 14.2 percent of GDP in 2017 and employed about 42.7 percent of the workforce, and in the 2022-23 financial year it contributed more than 12 percent of GDP. Citing a Ministry of Environment and Forests report, it says a one-degree rise in maximum temperature at the vegetative, reproductive and ripening stages reduced Aman rice production, that water salinity directly affects rice production in the coastal belt, and that the country would lose 12-16 percent of its land if the sea rose by one metre.

For adaptation, the article says, the Bangladesh Rice Research Institute (BRRI) introduced salt-tolerant varieties such as BR-11, BR-23, BRRI dhan28, BRRI dhan41, BRRI dhan47, BRRI dhan53 and BRRI dhan54, and measures such as resilient varieties, diversification, changed cropping patterns, mixed cropping, improved irrigation, soil conservation and agroforestry are practised in the country. The digital tools the writer proposed would sit alongside these.

The writer acknowledged the obstacles: farmers' limited knowledge of technology, high initial investment, poor rural internet, lack of training and a shortage of modern equipment. Government, research institutes and private initiatives working together could gradually overcome them, he wrote.

FAO's 2019 briefing paper gives the global picture. Although a billion new mobile subscribers joined between 2013 and 2018, 3.8 billion people remained offline, disproportionately in rural and remote areas, and only around a third of rural people in the least developed countries were covered by 3G networks. Affordability is the main barrier to smartphone ownership in those countries, where a basic mobile broadband plan still cost over 60 percent of gross national income per capita on average, the paper says, citing ITU data for 2017. FAO warns that if the benefits of digital agriculture are unequally shared, digital divides within and between countries could widen.

For drones, the FAO-ITU publication notes another challenge: advanced image analysis produces huge datasets, up to 140 GB for a single square kilometre at one-centimetre resolution, which need their own storage and management plans. FAO's 2010 paper adds that smallholders will need institutional and financial support to move to climate-smart agriculture, that institutions must be strengthened to spread climate-smart information across many farmers, and that considerable investment is needed to fill data and knowledge gaps and develop suitable varieties and technologies.

The writer closed by arguing that agriculture is the basis of people's lives, the environment and civilisation, and that properly applied, spray drones, AI, weather forecasts and IoT soil analysis could together deliver lower carbon emissions, healthier soil, water savings and higher output.

Photo: Vis M / Wikimedia Commons (CC BY-SA 4.0)

Sources: Jagonews24, Food and Agriculture Organization of the United Nations (FAO), International Telecommunication Union (ITU), Intergovernmental Panel on Climate Change (IPCC), World Bank, Frontiers in Plant Science, Bangladesh Agro-Meteorological Information System (BAMIS), Department of Agricultural Extension, Wikipedia