Smarter Farms: Watson Decision Platform for Agriculture

Share this post:

Bringing the power of Watson to farmers

agricultureAgriculture, a $2.4 trillion industry, is a foundation of economies worldwide. Factors such as climate change, population growth and food security concerns have propelled the industry into seeking more innovative approaches to protecting and improving crop yield. As a result, artificial intelligence is steadily emerging as part of the industry’s technological evolution.

Together with IBM Watson and The Weather Company, teams from IBM Research-Brazil and IBM Research-India designed and built a suite of agribusiness tools and solutions to help the agriculture industry use the power of AI to make more informed decisions about their crops – the Watson Decision Platform for Agriculture.

Underpinning the platform, IBM PAIRS GEOSCOPE processes some of the satellite data and serves as storage component in the current system architecture. By aggregating and analyzing terabytes of multi-layer geospatial data using machine learning and advanced analytics, PAIRS allows us to store and run queries on the geo-referenced data.

Four of the APIs included in this new platform come from our global labs.

Yield History and Forecast for Corn
This API uses big data and machine learning to predict yield for corn crops two to three months in advance with only a limited amount of data and computing power. Our system enables high-speed yield forecasts at a very high resolution (20 meters), generating personalized insights for farmers. The models  can also be used to determine yields for past growing seasons — critical for validation of agriculture insurance claims and risk, optimizing supply-and-demand chain logistics and predicting commodity prices.

Disease & Pest Indicators for Corn
This API service predicts the risks in corn production, leveraging hyper-local weather forecast details (temperature, relative humidity, precipitation, etc.) from The Weather Company and crop specific inputs (sowing date, growth stage, etc.) to model the outbreak probability of various pests and diseases. It also considers transport of the spore that triggers the disease. The advance notice for disease could help farmers reduce pesticide usage and take preventive or curative measures to avoid any unexpected yield loss.

High Definition Normalized Difference Vegetation Index (HD-NDVI) for Crop Health Monitoring
HD-NDVI uses geospatial and satellite data to identify crop type and crop growth stage at a high resolution, 30 meters. The insights from this API could be used to assess crop health, determine fertilizer, pesticide and irrigation schedules, validate crop insurance clams, predict yield, and reduce risk in commodity trading. With this level of insight, farmers could take preventive actions (pesticide application, fertilizer or nutrient application, etc) to preserve and improve the health of their crops.

High Definition Soil Moisture (HD-SM)
HD-SM is a high resolution, real-time measurement tool that monitors soil moisture at multiple depths (up to one meter) using a combination of AI algorithms and physical models along with several satellite and weather model data sets. Satellite data is combined with terrain data (such as land type, vegetation type, atmospheric parameters and solar radiation) from land surface models which is used to simulate changes in soil moisture.

Our work on this project is just one of many instances where IBM Research is helping to transform the agriculture industry.agriculture Teams in Africa are helping to build and test a blockchain-enabled finance lending platform for Twiga Foods. Our team in India has developed a suite of capabilities for small farms, and has partnered with the think-tank NITI Aayog to develop a pest and disease prediction model using artificial intelligence to provide real time advisory to farmers. And researchers in Brazil just built a prototype, the AgroPad, which provides simple, real-time analysis of the chemical makeup of soil or water.

Supporting Papers
Development of a High Resolution Soil Moisture for Precision Agriculture in India, International Conference on Precision Agriculture

Crop-identification using Sentinel-1 and Sentinel-2 data for Indian region, IEEE IGARSS

A Scalable Machine Learning System for Pre-Season Agriculture Yield Forecast, IEEE eScience

Director, IBM Research-Brazil

Sriram Raghavan

VP, IBM Research AI

More Agriculture stories

IBM Research at SIGMOD 2020

ACM SIGMOD/PODS 2020 like many other events impacted by COVID-19 pandemic will be taking place virtually from June 14 through June 19. The focus of work at SIGMOD 2020 ranges from adding graph querying to relational databases, to natural language interfaces to data, to operationalizing data for new AI workloads. Results to be presented includes work done at our IBM Research-Almaden and IBM Research-India labs, as well as by our summer interns from universities and our partners in other IBM units.

Continue reading

Largest Dataset for Document Layout Analysis Used to Ingest COVID-19 Data

Documents in Portable Document Format (PDF) are ubiquitous with over 2.5 trillion available from insurance documents to medical files to peer-review scientific articles. It represents one of the main sources of knowledge both online and offline. For example, just recently The White House made available the COVID-19 Open Research Dataset, which included links to 45,826 papers […]

Continue reading

IBM Research AI at ICASSP 2020

The 45th International Conference on Acoustics, Speech, and Signal Processing is taking place virtually from May 4-8. IBM Research AI is pleased to support the conference as a bronze patron and to share our latest research results, described in nine papers that will be presented at the conference.

Continue reading