Carbon Intensity Model¶

The carbon intensity (CI) model converts electricity generation mix data into grams of CO2 per kilowatt-hour (gCO2/kWh) for each country where Gnosis Chain nodes operate. This is a critical input to the network-wide carbon footprint calculation.

Data Source: Ember Global Electricity Review¶

The model ingests data from the Ember Global Electricity Review, which provides:

Coverage: 200+ countries and territories
Granularity: Monthly electricity generation data
Generation types: Coal, gas, oil, nuclear, hydro, wind, solar, bioenergy, and other renewables
Update frequency: Monthly, with a typical lag of 1--2 months

Ember Data Availability

Ember publishes monthly generation mix data for most countries. For smaller territories or countries with incomplete reporting, the model falls back to regional or global averages. See Fallback Hierarchy below.

Country-Level Carbon Intensity Calculation¶

The carbon intensity for each country is computed as a weighted average of the emissions factors for each generation type, weighted by each type's share of total generation:

\[ CI_{\text{country}} = \sum_{g \in \text{gen types}} \left( \text{share}_g \times EF_g \right) \]

Where the emissions factors (gCO2/kWh) by generation type are:

Generation Type	Emissions Factor (gCO2/kWh)	Category
Coal	820--1,100	Fossil
Gas	350--500	Fossil
Oil	650--890	Fossil
Nuclear	5--15	Low-carbon
Hydro	4--30	Renewable
Wind	7--15	Renewable
Solar	20--50	Renewable
Bioenergy	50--250	Renewable

Grid-Type Uncertainty¶

The measurement uncertainty assigned to the carbon intensity estimate depends on the grid's overall CI level. Cleaner grids have higher relative uncertainty because small absolute changes represent larger relative swings.

CI Range (gCO2/kWh)	Grid Classification	Uncertainty (%)
< 100	Very clean	25
100 -- 300	Clean	20
300 -- 600	Mixed	15
> 600	Fossil-heavy	12

Seasonal Adjustments¶

Electricity generation mix varies with seasons due to heating/cooling demand and renewable availability. The model applies continent-level seasonal adjustment factors.

Hemisphere Convention

Seasons are defined by meteorological convention. Winter (W) refers to December--February in the Northern Hemisphere and June--August in the Southern Hemisphere. Summer (S) is the opposite. The adjustment factors below reflect each continent's dominant hemisphere pattern.

Continent	Winter Factor (W)	Summer Factor (S)	Notes
Europe	1.12	0.88	Higher heating demand increases fossil share in winter
N. America	1.08	0.95	Moderate seasonal swing, summer AC partially offsets
Asia	1.05	1.02	Smaller seasonal effect due to diverse climate zones
Oceania	0.95	1.05	Southern Hemisphere: reversed seasons
S. America	0.98	1.03	Mild seasonality; hydro-dependent grids
Africa	1.02	0.98	Minimal seasonal variation

Combined Uncertainty¶

The total uncertainty in the carbon intensity estimate combines the temporal (seasonal) uncertainty and the measurement uncertainty in quadrature:

\[ \sigma_{\text{total}} = \sqrt{\sigma_{\text{temporal}}^2 + \sigma_{\text{measurement}}^2} \]

Where:

\(\sigma_{\text{temporal}}\) is derived from the grid-type uncertainty table above
\(\sigma_{\text{measurement}} = 10\%\) (baseline measurement error in Ember data and emissions factors)

Worked Example

Given:

Country carbon intensity: CI = 250 gCO2/kWh (clean grid)
Temporal uncertainty: 20% (from CI range 100--300)
Measurement uncertainty: 10%

Calculation:

\[ \sigma_{\text{total}} = \sqrt{0.20^2 + 0.10^2} = \sqrt{0.04 + 0.01} = \sqrt{0.05} = 0.224 = 22.4\% \]

Result:

CI = 250 +/- 56 gCO2/kWh (1-sigma)
95% confidence interval: [140, 360] gCO2/kWh

Fallback Hierarchy¶

When country-level Ember data is unavailable, the model applies the following fallback chain:

flowchart TD
    A["Node country identified?"] -->|Yes| B["Ember data available for country?"]
    A -->|No| E["Use hardcoded fallback: 450 gCO2/kWh"]
    B -->|Yes| C["Use country-level CI from Ember"]
    B -->|No| D["Use world average CI (~440 gCO2/kWh)"]

Priority	Source	CI Value (gCO2/kWh)	When Used
1	Country-level Ember data	Varies	Default for all mapped countries
2	World average	~440	Country has no Ember data
3	Hardcoded fallback	450	Node country is unknown

Network Effective Carbon Intensity¶

The network-wide effective carbon intensity is the node-weighted average across all countries:

\[ CI_{\text{effective}} = \frac{\sum_{c} \left( N_c \times CI_c \right)}{\sum_{c} N_c} \]

Where:

\(N_c\) = estimated number of nodes in country \(c\)
\(CI_c\) = carbon intensity for country \(c\) (from Ember or fallback)

This produces a single network-level CI value used in headline reporting and cross-chain comparisons.

dbt Implementation¶

The carbon intensity pipeline is implemented as a series of dbt models across staging and intermediate layers.

Model Details

Staging Layer

stg_crawlers_data__ember_electricity_data: Ingests raw Ember CSV data. Standardizes country names to ISO 3166-1 codes, pivots generation shares from wide to long format, and filters to the most recent available month per country.

Intermediate Layer

int_esg_carbon_intensity_ensemble: Computes country-level carbon intensity with full uncertainty bands. Applies emissions factors to generation shares, calculates grid-type uncertainty, applies seasonal adjustments, and combines uncertainties in quadrature. Outputs CI mean, standard deviation, and upper/lower bounds per country per month.
int_esg_node_geographic_distribution: Maps observed nodes to countries using IP geolocation data from the crawler pipeline. Produces daily country-level node counts used to weight the network effective CI calculation. Handles VPN/proxy detection flags and applies geographic confidence scores.

-- Simplified CI ensemble logic
SELECT
    country_code,
    date,
    ci_mean,
    ci_mean * (1 + seasonal_factor) AS ci_adjusted,
    ci_mean * grid_uncertainty AS ci_std_temporal,
    ci_mean * 0.10 AS ci_std_measurement,
    sqrt(
        power(ci_mean * grid_uncertainty, 2)
        + power(ci_mean * 0.10, 2)
    ) AS ci_std_total
FROM intermediate_carbon_intensity