The lack of timely, reliable, and cost-effective methods for measuring the carbon sequestration capacity of land at a large scale presents a challenge in leveraging the world’s vast agricultural fields – an estimated 5 billion hectares – to combat climate change. Without a more efficient method of measuring the carbon sequestration capacity of large fields, it is impossible for farmers to optimize their practices, such as tillage, over-cropping, fertilization, and watering, to maximize the carbon sequestration potential of their land.
Metaspectral is pioneering carbon management for agriculture using hyperspectral imagery.
Metaspectral is charting a greener future by developing a more accurate and cost-effective tool that will quantify and monitor carbon sequestration/emissions for various industries, including agriculture. This will be done on a larger scale than current methodologies.
Hyperspectral imagery enables a vast amount of information to be discovered (up to 300 different spectral bands per pixel), including the detection of chemical composition of substances present in each pixel of an image. This imagery, when combined with ground truth data, has the potential to measure amounts of greenhouse gasses more accurately – specifically CO2 and CH4, but also N2O.
This project focuses on developing a machine learning model to ingest hyperspectral imagery and then predict the concentration, or flux values, of CO2, CH4 and N2O. These amounts will then be used to create an atmospheric model deriving the carbon sequestration/emission level at every captured location of interest.
