Please describe your proposed solution.
Overall challenge
The urgent need for sustainable solutions is driving a rapid adoption of smart environmental monitoring and diagnostics technologies that utilise spaceborne Earth Observation (EO) and climate data. Target applications encompass a wide range of commercial domains, such as natural resource management, disaster management, weather forecasting, precision agriculture, forestry, mineralogical prospection and global trade. Furthermore, the emerging market for carbon credits and other sustainability services, such as direct carbon capture, regenerative agriculture, reforestation and afforestation require careful monitoring and validation to achieve the desired level of credibility and trust.
Unfortunately however, multiple major technological bottlenecks continue to impede wide adoption and commercial utility of environmental sensing technologies beyond the limited scope of scientific research. We have conducted an extensive survey of multiple organisations that aim to promote and accelerate reforestation, afforestation and soil carbon restoration, including for example Open Forest Protocol (<https://www.openforestprotocol.org/>), Restore (https://restor.eco), The Pond Foundation (<https://thepondfoundation.org/>), Landano (<https://www.landano.io/>), Forest Conservation Fund (<https://www.fundforests.org/>) and Agora Carbon Alliance (<https://agorocarbonalliance.com>).
Our findings show that at this time the vast majority of monitoring and validation for such projects relies on labour-intense, expensive, time-consuming and often unreliable ground samples and manual inspection. Significant efforts have been invested into automation of monitoring and validation using remote sensing technology, but the scalability and efficiency of such methodology is currently limited by the following factors:
- Commercial high-resolution EO imagery is far too expensive for most environmental monitoring projects and applications, while the spatial resolution of the low-cost EO imagery is too coarse for most such applications.
- Challenging atmospheric conditions, such as obstruction by clouds and cloud shadows, make the availability of high-quality data highly irregular and temporarily incoherent.
- Monitoring often requires high-quality historical imaging data that is often unavailable, or too expensive.
- Each EO system is designed to address a specific segment of use cases using a specific combination of technical characteristics, such as revisit time, number of spectral bands, spectral range and spatial resolution, thus making automated assimilation of data from multiple such systems difficult, or impossible.
- Radiometric calibration, and thus the stability and reliability of collected data varies drastically between different systems and continues to present a major and largely unsolved technological challenge.
Significant attention has been recently raised by the Dynamic World by Google (https://www.dynamicworld.app/) – “A near real-time land cover dataset for our constantly changing planet”. This system clearly demonstrates the immense potential of remote sensing for environmental monitoring applications. The system addresses the very important challenge of global coarse land-cover classification into nine major types, however it does not provide the means to understand the local environmental dynamics for a specific project or use case.
Our vision
The proposed project will develop and deploy an Environmental Oracle that will offer a consolidated source of regularised multi-modal multi-scale (spatial, spectral and temporal) environmental data required for implementation of low-cost, scalable monitoring and validation solutions for carbon capture and ecological restoration projects and DApps. The proposed data structure will include a fusion of spatial, spectral and temporal features specifically designed and optimised for training of analytical and predictive ML models.
The information aggregated by Environmental Oracle will act as a form Environmental Passport for any target plot of land that will both record all changes of status, as well as provide a source of objective data for the analysis of dynamics, performance and health indicators.
Figure 1. H3: Uber’s Hexagonal Hierarchical Spatial Index
The proposed Environmental Oracle will utilise Uber’s Hexagonal Hierarchical Index (H3) for multi-scale indexing and access to consolidated environmental data. The system will provide access to the following data types in the form of regularised and consolidated multi-layer multi-modal data cubes:
- Instant
- Radiometrically calibrated, cloud-free multi spectral satellite imaging data at up to 2m/px spatial resolution
- Short-term cycle – 12-month time series at 5 days revisit rate
- Reflectance spectrum
- Climate data
- Temperature (max/min/5-day average)
- Precipitation (5-day average)
- Cloud cover (5-day average)
- Long-term cycle – 12-year time series at 60 days revisit rate
- Vegetation index
- Climate data
- Temperature (max/min/60-day average)
- Precipitation (60-day average)
- Cloud cover (60-day average)
Figure 2. Ecomandala visualisation of the information features constituting the Environmental Oracle multi-modal spatial-spectral-temporal dataset.
The resultant consolidated source of environmental data will be deployed on Cardano blockchain and will be made available through a standardised API which will significantly streamline and simplify access to such data for developers of environmental monitoring and validation applications, as well as a wide variety of sustainability focused DApps. Some examples of such DApps could include DeFi support of regenerative agriculture, agriculture supply chains, forest conservations and tree planting projects.
As part of our future work, we are planning the development and release of an Ecomandala NFT that is a dynamic visualisation of the aggregated and regularised environmental data detailed in Figure 4. Ecomandala will represent the historical record and the current state of any given plot of land in an information-rich and visually compelling form allowing for interactive and engaging exploration of environmental data.
Figure 3. Original Sentinel-2 L2A scene (left) and processed scene with the clouds and cloud shadows removed using the proposed data fusion method (right).
The proposed project will invoke the latest developments in the area of Artificial Intelligence and Machine Learning to enable coherent fusion of multiple sources of environmental data, including European Space Agency (ESA) Copernicus Sentinel 1 and 2, as well as the corresponding climate data in order to obtain regularised, cloud and cloud shadow-free time series of satellite imaging data.
The satellite imaging data will be further super-resolved to the spatial resolution of 2 m/px, significantly expanding the range of target monitoring and verification applications.
Figure 4. x10 super-resolved images generated by Gamma Earth S2DR model. Sentinel-2 RGB 10 m/px (left), S2DR RGB 1 m/px (center) and Google Maps RGB 30 cm/px (right).
Project Work Packages
The proposed Environmental Oracle will incorporate the following major Work Packages
- Ingestion and assimilation of multi-date Copernicus Sentinel 1 and 2 data
- Detection, masking and removal of clouds
- Super-resolution of EO imagery to 2 m/px spatial resolution
- Fusion of EO imaging data and climate data into single multilayer spatial map
- Implementation of Environmental Oracle
- Design and development of Web portal for data visualisation and access
- Development of Python API
Importantly, Work Packages 1-3 will constitute the direct extension of the previously attained results and only a minor refactoring and improvements is envisioned within the scope of the proposed project. The current proposal will be mainly focused on the implementation of Work Packages 4 - 7.
Please describe how your proposed solution will address the Challenge that you have submitted it in.
The envisioned Environmental Oracle will help resolve a major technological bottleneck associated with the challenge of sourcing, accessing and processing relevant environmental data required for monitoring, auditing, and validation of carbon capture and ecological restoration projects. It will enable the development and deployment of a new class of high impact DApps on Cardano blockchain positioning Cardano as a leader in environmental monitoring applications.
Environmental Oracle will constitute a platform and an integration layer enabling further development and deployment of monitoring and validation solutions for specific DApps on Cardano ecosystem. We therefore inspire to
- Showcase the methodology of low-cost scalable monitoring and validation of ecological restoration projects
- Standardise, streamline and simplify access to consolidated environmental data
- Attract the developers of monitoring and validation solutions for carbon capture and ecological restoration projects to Cardano ecosystem
Our solution will further promote collaboration, synergy and interoperability between Cardano projects by leveraging the achievements of the recent Fund 8 “Oracle Development Portal” project: https://cardano.ideascale.com/c/idea/400771 (which is not a related project but we will be happy to collaborate in order to maximise the impact of both projects).
Examples of funded and submitted Cardano Ideascale projects with direct synergy potential to the proposed Environmental Oracle
- 21st century Agri supply chain <https://cardano.ideascale.com/c/idea/403695>
- Oracle Developer Portal <https://cardano.ideascale.com/c/idea/400771>
- Indigenous Land Rematriation <https://cardano.ideascale.com/c/idea/398750>
- Open ledger for agricultural land <https://cardano.ideascale.com/c/idea/400812>
- Protecting wildlife & Maasai, Kenya <https://cardano.ideascale.com/c/idea/399158>
- Landano: Cardano land registry Dapp <https://cardano.ideascale.com/c/idea/381957>
- Community validation Dapp <https://cardano.ideascale.com/c/idea/421218>
What are the main risks that could prevent you from delivering the project successfully and please explain how you will mitigate each risk?
Type of risk: Technological
Description of risk: The envisioned system implies ingestion and assimilation of heterogeneous multi-modal environmental data from multiple sources and data providers. The respective end points may be subject to change over time, thus making the system prone to outages.
Effect of risk: Reliability constitutes one of the major advantages and selling points of the proposed system. Consequently, any reliability issues can have significant negative impact on long term adoption.
Mitigation methods: The system will require careful planning of performance monitoring and maintenance mechanisms. Furthermore, system robustness mechanisms will be implemented that will allow the system to remain online even when some of the input signals may be temporarily unavailable.
Type of risk: Economic viability
Description of risk: The sourcing, processing of data will have significant cloud infrastructure and computing costs
Effect of risk: As is the case with any technological system, the long term success will be put at risk if the economic viability is not realised from an early stage.
Mitigation methods: Careful attention will be given to the cost management and computational efficiency of the system. Furthermore, we will continue to continuously monitor the cost-benefit balance of the proposed solution and will seek to reach a break even point at the earliest possible stage.