Drone-based tree identification and counting method in a carbon offsetting project
28/11/2025 , Audition Room - 2nd Floor
Langue: English

Concerns about institutional carbon emissions are increasing, driven by the need for transparency in reporting and mitigation of commitments. The Paris Agreement on Climate Change has prompted some countries to enforce institutional reporting of carbon emissions and mitigation actions. Emitters can mitigate their emissions by adopting low-emission technologies or investing in offsetting programs. In 2022, the Malawi Liverpool Wellcome Programme invested in a carbon offsetting program involving tree planting on Mulanje Mountain. This study, we expanded our remote sensing project on mapping plastic waste accumulating in the environment to aid monitoring of these trees and estimating their carbon storage potential. We conducted field surveys using transects to assess tree survival, collecting data with KoboCollect. Field-level pictures of the transects were captured to aid in the identification of vegetation species, which is being conducted by a professional ecologist. Drone images of the planted trees were collected using a DJI Mavic 3M, capturing both optical and multispectral reflectance (infrared and near-infrared). The images were processed with OpenDroneMap to produce a seamless orthomosaic map and a 3-D model of the planting plot. We are training an automatic classifier to identify the planted trees and invasive species from the drone imagery. Field-based monitoring revealed 73 planting stations, with 39 stations having surviving trees, representing a 54% survival rate. The monitored vegetation included grown trees, shrubs, pine trees, weeds, grass, and other invasive species. Drone images from two sites have been processed, and efforts are underway to annotate the data to identify tree planting stations and surviving cedar trees to quantify tree survival. The ongoing study has the potential to enhance the transparency of offsetting programs, particularly tree-planting campaigns. Once the classification model is finalized, it will improve field-based tree monitoring, providing rapid insights into tree survival and facilitating appropriate compensation or replacement actions. Future efforts will focus on estimating the carbon storage potential of the trees.

Patrick is a PhD student based at the Malawi Liverpool Wellcome Programme. His research focuses on developing methods to investigate the negative impacts of environmental waste accumulation on human health. Recognising the complexity of this issue, he believes his background in geospatial sciences provides valuable tools for uncovering these linkages and generating evidence to guide practical action especially in contexts where waste accumulation is not yet recognised as a public health concern, despite its potential risks. He is a strong advocate for progressive science, science that continually evolves as it is translated into tangible societal benefits, using implementation itself as a feedback loop. As an example, he has extended his PhD work to support efforts aimed at improving transparency in carbon mitigation within a typical African setting.