In order to prevent catastrophic destabilization of the planet’s climate system and to maintain the planet’s biosphere, human society must limit its greenhouse gas emissions before the global ‘carbon budget’ is exceeded. According to the Paris Agreement (PA), countries define their own contributions towards emissions and periodically revise these towards more ambitious targets, till such levels are achieved. Policymakers use climate policy instruments such as carbon taxes or offsets to create the incentives required, which require an accurate and consistent price for carbon. Today, carbon markets are highly fragmented and there is high variance in the prices set in different jurisdictions, ranging from less than $1 (in Mexico) to $119 (in Sweden) per ton of carbon dioxide. The reason for this large discrepancy is multidimensional and complex, but their effect is simple: an untrustworthy supply chain for one of the most important information goods in the world, carbon pricing.
To this end, we have designed the decentralized infrastructure required to produce this pricing from diverse data-sources in realtime, an ongoing challenge in carbon pricing markets. Within this project we will develop the oracle infrastructure that can monitor, verify and report real-time climate variables from trusted sources, such as the global carbon dioxide concentration. Additionally, we will derive the carbon budget and immutably log both the outcomes (realtime co2 concentration and remaining budget) and the parameters that led to the outcome, rendering the whole supply chain transparent. This system will serve as a building block for a greater system that can derive carbon pricing in a similar manner.
The carbon prices produced themselves act as a canonical or shadow price, by which other markets and actors can implement cap-and-trade, carbon taxation, or internal pricing strategies.
This project was conceived during a UC Berkeley class, CS294-177:DeFi, in the fall of 2021. It is being carried out with guidance from Open Earth Foundation and is being updated to integrate an IAM into the system in the Spring and Summer of 2022.
This project's goals are:
- Deliver realtime carbon dioxide (CO2) concentration data, aggregated from National Oceanic and Atmospheric Administration’s (NOAA) Global Monitoring Lab, to any smart contract on the Ethereum network
- Compute the remaining carbon budget from this data point
- Immutably log the inputs and outputs that led to this outcome to an on-chain database
- Allow any web3 client to query this database
Additionally
- Be the building blocks for generating the Social Cost of Carbon via an Integrated Assessment Model, like FUND.
This in effect acts as a framework for illuminating carbon pricing supply chains. The next step is to extend this framework to do carbon pricing and logging as well. The longer term goals of this project are to enable open access carbon pricing to anyone (web or web3 clients), with a corresponding log to a public blockchain, and subsequent retrieval.
The software architcecture is as show below:
The de facto source of atmospheric carbon dioxide concentration is from the National Oceanic and Atmo- spheric Administration’s (NOAA) Global Monitoring Lab, which publishes greenhouse gas concentration readings from 87 stations around the world, at intervals ranging from hourly to monthly, with some unpredictable lag. Originally, we planned to create a model that would predict the realtime (hourly) outcome from the daily and weekly values, which would account for the seasonal nature of CO2 concentration change1, but we found that NOAA already publishes two sets of predicted values on a daily basis that model the global average CO2 concentration (’smoothed’), and control for the seasonal differences (’trend’). These were accurate and high frequency enough for our purposes. However NOAA does not provide an API for this data, but simply a text file download. To fill this gap, we created a simple CO2 server, built on Flask, that would scrape this data from NOAA on a daily basis, perform some data cleaning and store these values in a database. This data, along with it’s generated timestamp were exposed through an API that allows anyone to read the most recent CO2 concentration values (both smoothed and trend) along with an epoch timestamp. This API would enable this data to be brought on chain.
All implementation details for this Flask server can be found below:
As specified earlier, the SCs have a dual function: to fetch the cost of carbon from the CO2 Oracle and derive the carbon budget, and create an immutable entry to the underlying blockchain with the timestamp, and all input parameters. This schema encapsulates all the constituents that led to this particular price, rendering the supply chain transparent. To fetch the realtime data from the CO2 Oracle, we used the request-response mechanism, as required by Chainlink’s hybrid smart contracts. They require two separate calls to the SC: the first specifying the API endpoint and JSON hierarchy, and a second to read the retrieved data. The second call is also used to compute the carbon budget on-chain. In order to immutably log this schema, we created a limited functionality CRUD database (without the update and delete functions), that allows for append as well as sequential and random reads. We adapted Rob Hitchen’s CRUD framework2 to use a private Mappings as the hash table, with the timestamp as the unique key, and all the input parameters and generated output within a struct, as the storage. In order to allow web3 clients to iterate over the keys and retrieve any element without loosening the security of the hash table, we add a public dynamic array to store the keys:
mapping ( uint256 => ledgerStruct ) private co2Ledger ; uint256 [] public keyList ;
To enable any web3 client to iterate over the entire database and retrieve any or all entries, we expose the following functions:
function getEntityCount () public returns ( uint entityCount ) function getEntity(uint256 entityTimestamp) public returns (...)
function getKeyAtIndex(uint index) public returns(uint256 key)
We need to convert the current atmospheric CO2 concentration in parts per million (ppm) to Gigatons of CO2; this done by the fomula:
1ppm = 2.13 GTC = 7.81 GtCO2
We need to subtract this number from the total carbon budget available. The IPCC publishes this remaining CO2 budget with associated probabilities. The latest report, AR6 says:
For a 66% chance of success -> 360GtCO2 For a 50% chance of success -> 460GtCO2
For a 66% chance of success -> 1110GtCO2 For a 66% chance of success -> 1310GtCO2
On Jan 1, 2021, the CO2 conc -> 414.24ppm 2020 Avg : Source NOAA -> 3235 GtCO2
Therefore :
| Scenario | Probability | Remaining Budget(2021-01-01) GtCO2 | Total Budget GtCO2 |
|---|---|---|---|
| 1.5C | 66% | 360 | 3,595 |
| 1.5C | 50% | 460 | 3,695 |
| 2C | 66% | 1110 | 4,345 |
| 2C | 50% | 1310 | 4,545 |
Current CO2 conc (ppm): p Scenario : S
For S, total budget : T(S)
Remaining (approximate) CO2 budget (GTCO2) = T(s) - p X 7.81
In the contract APICO2Ledger.sol, we only model the remaining budget with the 1.5C-66% scenario.
The smart contract implementation and testing was done using the Brownie Frame- work which is a Python-based development and testing framework for smart contracts on Ethereum. Smart contracts were written in Solidity (v0.8). This framework enabled us to write sophisticated web3 client-side interaction scripts. This smart contract has been deployed to the Kovan testnet at 0xD56f9457961fE6C02B791c666Fba2830a510DD78. All code and instructions to replicate this system are available on Github. The CO2 concentration API can be accessed at https://co2ppm.herokuapp.com/trend.
The Brownie code for this implementation and details to recreate it can be found below:
This project is licensed under the MIT license.
This project is supported by a grant from the Haas School of Business, UC Berkeley, with funding from Ripple's University Blockchain Research Initiative. It is being advised by Prof. John Chuang at UC Berkeley.
Footnotes
-
This seasonal variation is because during the northern hemisphere summer, the vegetation around the equator and in the northern hemisphere absorb large amounts of CO2 during photosynthesis, and during the northern hemisphere winter, this activity shifts southwards where there are relatively fewer photosynthesizing agents. ↩
-
Rob Hitchens. Solidity CRUD part 1 URL: https://medium.com/robhitchens/solidity-crud-part-1-824ffa69509a. ↩
