How every figureis calculated.In full, and versioned.

This page is the method our engine runs, written down. Every report names the version it used, so a figure from last year can still be checked against the rules that made it.

1 · The formula

Emissions are worked out one hour at a time. For each hour, every energy source’s reading is multiplied by the emission factor that applies to that source at that hour, and the results are added up.

E_hQ_s,hEF_s,h

E_h: emissions in hour h, kg CO₂e
Q_s,h: quantity from source s in hour h
EF_s,h: factor for source s, valid at hour h

One real hour from the sample14 Mar 2026, 11:00 · Grid · 178.8 kWh × 0.4316 kg/kWh = 77.17 kg CO₂e

2 · Sources and scopes

Snowleaf covers what a site measures directly at the meter.

Source	Measured in	Counted as
Grid electricity	kWh	Scope 2, location-based
Diesel generator	litres	Scope 1
Gas generator	m³	Scope 1
On-site solar	kWh	No operational emissions (factor 0)

3 · Emission factors

These factors are placeholders while we verify published sources. They are labelled that way in every report that uses them. When a verified factor replaces one, the method version goes up and older reports keep the factor they used.

Source	Factor	Source of the factor
Grid electricity	0.4316 kg CO₂e/kWh	Illustrative, pending verification · 2026
Diesel generator	2.68 kg CO₂e/litre	Illustrative, pending verification · 2026
Gas generator	Not set yet	No factor yet; until one is added, a run with gas readings stops
On-site solar	0 kg CO₂e/kWh	Snowleaf method: no operational emissions

The rules the engine follows

Each factor is valid for a set period. Exactly one factor must cover each source in each hour.

If no factor covers an hour, or two do, the run stops with an error. It never falls back to a default.

A factor’s unit must match the reading’s unit, or the run stops.

4 · Data quality

Every reading carries one of four labels, and the label travels with it into the report. Nothing is relabelled to look better than it is.

MeasuredRead from a meter for that hour.
ManualEntered by a person, for example from a logbook.
EstimatedWorked out from other data rather than read directly.
Gap-filledA missing hour filled in so the year adds up, and shown as such.

5 · Completeness

Every run reports what share of the period’s hours it has readings for, and grades it.

Baobab Site A (sample), 20268,760 of 8,760 hours · 100% · OK

The two marks are 90% and 98%. Below 90%: critical · 90–98%: warning · 98% and above: OK.

6 · PUE

Power usage effectiveness is the facility’s total energy divided by the energy used by the IT equipment, over the report period. The sample site’s is 1.46. If either meter has no readings, PUE isn’t shown at all: never a zero, never a guess.

PUE =facility energy, kWhIT equipment energy, kWh

7 · Versions

When the method changes, its version number goes up and the change is listed here. A report always names the version that produced it.

0.1.0: First version: hourly calculation for grid, generator and solar, data-quality labels, completeness and PUE.