Hourly carbon accounting? If you want to be accurate, then it’s going to be the new standard!

GRaphic ex Canva to illustrate blog post on high resolution carbon accounting

A recent scientific paper from the US finds that the traditional approach to carbon accounting using annualised emission values is becoming increasingly inaccurate for electricity as intermittently-generated renewables enter the grid. The recommended remedy is moving to hourly values for emission calculations instead of yearly ones. Wattwatchers has begun testing these findings for Australia using highly-granular electricity monitoring data we collect every five minutes.


We’re calling it high-resolution carbon accounting. Thanks to the energy transition, and surging electrification, we think it will inevitably become a new standard for accurately calculating greenhouse gas emissions from electricity at buildings and sites of all kinds.

Put simply, it means moving from broad brush annualised and averaged values (or factors) for the emissions intensity of electricity to high-resolution hourly ones, with a focus on what the international protocols define as ‘Scope 2’ emissions – which cover use of energy and include electricity as a major component.

This is the first in a series of blog posts about high-resolution carbon accounting. Subsequent posts will cover the following topics in more detail:

The Wattwatchers team has been prompted to work on this emerging opportunity after seeing a scientific paper from the US, published in April 2022, titled: Hourly accounting of carbon emissions from electricity consumption.

We see this research as being highly relevant to Australia, where new renewables like solar and wind are surging, because the abstract for the paper says:

Our results show that annual-average accounting can over- or under-estimate carbon inventories as much as 35% in certain settings but result in effectively no bias in others. Bias will be greater in regions with high variation in carbon intensity, and for end-users with high variation in their electricity consumption across hours and seasons. As variation in carbon intensity continues to grow with growing shares of variable and intermittent renewable generation, these biases will only continue to worsen in the future.

As you’ll see below, our initial analysis strongly aligns with what the US research team found and reported, and further work is required to understand this trend better.

Why is this important?

For starters, the US research team makes the point that accuracy is a fundamental requirement for counting carbon around the world, saying:

‘Accuracy is one of the fundamental GHG accounting and reporting principles described by The GHG Protocol. As noted in the Protocol’s Corporate Accounting and Reporting Standard, ‘data should be sufficiently precise to enable intended users to make decisions with reasonable assurance that the reported information is credible. GHG measurements, estimates, or calculations should be systemically neither over nor under the actual emissions value, as far as can be judged, and that uncertainties are reduced as far as practicable’.

At a lay person level, the importance is straightforward: if carbon accounting is becoming more and more inaccurate at a systemic level, due to steadily rising levels of new, intermittent,  renewables in the grid, then the methodologies being used need to be reassessed and fixed.

Otherwise, some participants in the market will be unfairly advantaged, and others will be unfairly disadvantaged. We wouldn’t accept this for our bank accounting, and we can’t endorse it for our carbon accounting either.

If new renewables are making the overall job of reaching net zero for electricity even a little bit easier, and the data supports this, then this should be reflected in the calculation protocols. The benefits of accuracy at the model level will filter through to everyone, from everyday households trying to calculate and potentially offset their emissions footprint, through to the high finance game of global carbon trading.

How do you model it?

The model for doing high-resolution carbon accounting for grid-supplied electricity, which we’ve begun testing with our datasets, takes three key things into account: emissions intensity per generation type (e.g. black coal, brown coal, gas, hydro, wind and solar); the generation type ‘mix’ per measured time interval; and finally, using interval data from on-site monitoring or metering, the actual imports of electricity from the grid per interval.

For our initial testing, we’ve found it convenient to use 30-minute intervals rather than hourly ones, but we don’t think this materially changes the outcomes. 

The main external sources we used for our data modelling are the National Greenhouse Accounts Factors 2021 and data from the OpenNEM website^  to calculate the emissions intensity per generation type on a state-by-state basis; and OpenNEM for 30-minute interval data on the generation ‘mix’.

^OpenNEM is an open platform for National Electricity Market (NEM) data created through the Energy Transition Hub at the University of Melbourne, with funding support from Energy Consumers Australia (ECA). It sources its energy data from the Australian Energy Market Operator (AEMO) and the Australian PV Institute (APVI).

What does our preliminary analysis add to the discussion?

Our preliminary analysis shows that emissions intensities fluctuate significantly over a 24-hour period (see Figure 1 below). Intensities are far lower during the day when solar makes up a significant portion of generation. Different states also have very different emissions intensity profiles. For example, South Australia – the state with the highest penetration of wind and solar –  has a much less predictable profile than the more coal-dependent states like Queensland, NSW and Victoria.

Figure 1: Emissions intensity over time by state

It’s important to point out that this preliminary analysis only takes generation within the individual states into account. Imports from other states will be included in the next iteration of the model. This will affect the emissions profiles. For example, South Australia will have a higher average emissions intensity when imports are included because the energy it imports is generally more carbon intensive than the energy generated within the state. Nevertheless, the values calculated in this initial analysis can help us get a sense of the impact that high resolution carbon accounting can have in Australia.

Comparing average annual accounting and high resolution accounting

We used our analysis of emissions intensities to calculate emissions from grid imports for a set of 744 residential sites. We calculated emissions both using the annual average emissions intensity (i.e. current practice) and emissions intensities at 30-minute intervals (high resolution carbon accounting), and compared the results of both calculations.

What we found is in line with what the US research paper reported: annual-average accounting can over- or under-estimate carbon emissions substantially for certain sites, but result in effectively no bias in others. For the sites we analysed, annual-average accounting tends to overestimate emissions in all states except South Australia (see Figure 2). This makes sense when you think about it: residential sites tend to use most of their electricity during the day (when the emissions intensity of the grid is lower) and very little overnight (when the emissions intensity of the grid is higher).

It must be noted that this preliminary analysis only comprises 19 days of data. A proper comparison of annual average accounting and high resolution accounting requires a full year of data, to account for seasonal effects.

Figure 2: Percentage by which annual average accounting over- or underestimates carbon emissions for 744 sites—positive values indicate overestimations.

The next blog post in this series will cover our modelling and analysis in more detail.

What could this mean for your carbon calculations?

High-resolution carbon accounting may sound like making an already complex chore even harder.

But the emissions goal posts for the electricity system are changing, as more and more intermittent renewables enter our grids, and as more and more emissions-intensive fossil fuel-powered generation is phased out of the electricity grid mix.

Accuracy, however, is important – for your sites, for the wider economy, and for the planet.

So hourly or even 30-minute carbon accounting will be more accurate than the traditional yearly approach, and will become increasingly so as renewables proceed to dominate grids (AEMO, the Australian Energy Market Operator, is preparing for 100% penetration of the grid by renewables at some times of the day as early as 2025, just a few years off). 

As well as being good in its own right, greater accuracy may save money too, for site or fleet owners, when the results translate to reducing emissions footprint for electricity use thanks to the high-resolution methodology. (This also may be increasingly material in an era of expanding electrification, including electric vehicle charging, and increasing penetration of renewables into our grids.)

It also opens up an opportunity to manage time of use of electricity for emissions intensity, as well as for other priorities such as efficiency and peak demand. The third blog post in this series will explore the topic of carbon efficiency in more detail.

The recommendations in the US paper (section 5) include a range of indicative scenarios where using hourly versus ‘annual-average values’ for emissions intensity may have practical implications for managing, regulating and incentivising carbon emissions and their impact.

Next steps

Beyond the initial testing and analysis outlined in this article, Wattwatchers is continuing to develop the model and inputs to it, and will apply the results to additional datasets including from commercial sites and non-Wattwatchers meters.

We welcome interest from our partners and customers, and from other external stakeholders, and are open to collaborating on further developing this concept of high-resolution carbon accounting.

Adriaan Stellingwerff is a Senior Software and Data Engineer at Wattwatchers. Murray Hogarth is Head of Impact and Communications.

FOOTNOTE: In our initial testing of high-resolution carbon accounting in Australia, Wattwatchers used data from 744 homes across several states where we have monitoring installed through our My Energy Marketplace (MEM) project. The MEM, a $10 million project, is supported by $2.7 million in grant funding from the Australian Renewable Energy Agency (ARENA)*.


The My Energy Marketplace project is receiving funding from ARENA as part of ARENA’s Advancing Renewables Program. The views expressed herein are not necessarily the views of the Australian Government, and the Australian Government does not accept responsibility for any information or advice contained herein.