We are excited to share that Climate Connect Digital (CCD) hosted its webinar on the 24th of May. This impactful event was centred around Digital Measurement, Reporting, and Verification (DMRV). DMRV represents a paradigm shift in climate action, leveraging advanced technology to automate the precise measurement, reporting, and verification of greenhouse gas emissions and sustainability metrics. With the potential to significantly reduce the time it takes to verify carbon credits, DMRV has captured the attention of environmental enthusiasts and industry professionals alike.
During this informative session, Mustafa highlighted three important aspects of MRV: emissions tracking, mitigation actions, and support provision. All three can potentially be streamlined and scaled through digitalisation. He also identified three main barriers to implementing digital MRV (DMRV): namely high costs, data privacy concerns, and a lack of resources and capacity in developing countries.
Continuing on, Mustafa provided specific examples of how CCD in particular has applied digital MRV in the context of climate action. These included using satellites and drones for biomass estimation and forest fire detection, employing MRV tools for renewable energy projects to verify carbon claims, and automating the extraction of Sustainable Development Goals (SDGs) from carbon project documents using Natural Language Processing (NLP) capabilities.
He noted that many of these use cases also enable greater community engagement and participation in the DMRV process; for example, through open-access mobile applications.
CCD has developed ML-based models to estimate canopy height and above-ground biomass (AGB) over woodlands at sub-national scale. This allows for identification of forest deforestation and disturbance, and aids efforts to conserve biodiversity hotspots. How it works:
Our ML models use high-granularity optical and radar satellite data from the ESA and NASA (1984-present). This allows for high precision remote sensing around the globe, with a minimum accuracy of 80% for forest cover, and 70% for biomass and canopy height.
The ML models share processed and analysed data via web-GIS portals. The data contains 2D/3D geospatial visualisation, including variables which correlate to geographical coordinates. This can reveal temporal and spatial patterns.
Community data collection:
Digital data collection devices and tools measure social impact indicators.
Mustafa presented a case study in which this method was used remotely to identify and track fluctuations in biomass levels due to managed plantations and regular harvesting.
Forest fire detection and monitoring
CCD utilises 3D imaging data from the following sources to deliver its active forest fire detection and monitoring services:
This data is inputted into CCD’s proprietary ML algorithms along with geographical coordinates, to enable fire tracking in real time. Email and mobile alerts are then generated for quick response and fire prevention.
This ML-based ensemble boasts ~85% accuracy, and captures over 500 incidences of forest fires globally on a daily basis. Clients can utilise these services to create actionable alerts for project proponents, forecast fire growth and calculate resulting carbon credit losses, and better manage permanence risks and buffer pools for carbon registries.
DMRV for renewables
CCD offers forecasting and scheduling services for independent power producers.
SCADA systems or other third party sources; India’s Central Transmission Utility (CTU); daily generation reports (DGRs).
SCADA data is compared with other sources to understand the issuance of carbon credits. Real-time data is also shared daily with the Global Carbon Council (GCC), for both energy and carbon credits.
SCADA data is audited against other sources to identify discrepancies with meter data. This audit provides an optimised estimation of the issued carbon credits through SCADA data.
Mustafa pointed out that having an independent source of data (such as SCADA) to verify the amount of carbon credits being claimed by projects on a registry can significantly reduce the costs associated with verification. Another major benefit is that this can accelerate the issuance of carbon credits, allowing new issuances on a bi-annual or monthly basis rather than once a year.
Automated SDG extraction from project documents
Identifying which SDGs a project contributes to is a crucial task, and one that can be very time consuming for companies or project developers who rely solely on manual methods. However, AI can be used to intelligently extract SDGs relevant to a particular project based on the unstructured content contained in related documents (e.g. images and text). This method relies on Natural Language Processing (NLP).
CCD has conducted R&D and testing on more than 50 documents in order to develop its own NLP service. Currently, the company’s models operate with 85% accuracy, which is projected to reach 95% within the next two quarters. CCD utilises the following AI models for SDG extraction from project documents: