Crop forecasting gets smarter with machine learning
The Institute for Scientific and Technological Information of the Council for Scientific and Industrial Research (CSIR-INSTI), has started research into the use of machine learning to explain the characteristics of crops, based on environmental and soil conditions of the specificlocation.
The outcome of the research will have an impact on food security and sustainability in Ghana.
The research is being conducted by a team of 10 technical staff of the Institute, and will employ machine learning models to extract physical patterns that affect yield, stress tolerance, and other agronomic indicators.
The team will also explore the potentials of combining sensor data collection with machine learning techniques for physicaltrait identification, classification, quantification, and prediction, ultimately promoting climate-smart and sustainable agriculture.
According to Dr. Ing. Michael Wilson, team lead, the team has deployed sensors to collect soil information on samples from various locations in Ghana. The next phase of the project will involve the use of cameras to observe the physical traits of crops grown on these soils.
The physical traits will be superimposed onto the real-time soil information obtained, and serve as a basis for future predictions of crop behaviours and possible soil conditions.
“By leveraging AI, our research team will also analyse this data and provide farmers with predictive analytics,”Dr. Ing. Wilson said. “This information, even in areas without internet access, could be delivered through SMS text messages, audio messages in local languages, or local radio broadcasts.”
Dr Ing Wilson explained that as part of this research, local sensor kits are being developed for deployment in local farms to assist precision in extension delivery. The potential impact of this research is significant especially for rural farmers, who often lack access to advanced technologies and weather information services.
“The data obtained will be useful for the development of agro-advisory tools and improvement in plant breeding programmes,” he said.
Mr John Awotwi, a team member, said the benefits of machine learning in the agricultural space are numerous.
“The beauty is that machine learning can be tailored to suit agricultural conditions peculiar to Ghana,” Mr Awotwi said.
The CSIR-INSTI is mandated to make scientific information available to stakeholders and the general population.
Over the years, CSIR-INSTI has carved a name for itself in the development of digital solutions for the both the public and private sectors. Most of these applications are hosted on the Ag-hub platform
www.ghanaaghub.com.This is in line with its mandate of appropriately packaging and making available scientific information to the public, which CSIR-INSTI is committed to.
Kuafo marketplace – an e-commerce platform, Agritech Advisor – a real-time question and answer platform and a cropping calendar which provides advice on the planting schedule and activities for maize and cowpea, and a weather station are some solutions recently provided by CSIR-INSTI in collaboration with the private sector.
Additional notable technologies developed by CSIR-INSTI include CSIR space, an online research publications repository; an online promotions system for research scientists of the CSIR and an online appraisal system for the CSIR.
The writers : Tracy Adjeley Sackey is senior Technical Officer while Samiratu Abdulai Mamah is an Assistant Research Scientific Officer – CSIR-INSTI
BY TRACY ADJELEY SACKEY AND SAMIRATU ABDULAI MAMAH
