Summary

Between 2023 and 2026, Northwild restored 6,200 hectares of drained blanket bog in Scotland’s Flow Country. Its carbon credits had been valued with a default peatland model that carried ±38% uncertainty, so buyers discounted them heavily. Terrapulse installed a network of flux towers, water-table loggers and soil plots to measure what the restoration actually did. After three years, the measured benefit was 41,000 tCO₂e at ±5% uncertainty — 28% more than the model predicted — and the credits sold at a 32% premium.

The site

The Flow Country is the largest expanse of blanket bog in Europe. Peat here has built up over 8,000 years, locking carbon into waterlogged soil where dead plants never fully decompose. From the 1970s, large areas were ploughed and drained for forestry. Drainage lowers the water table, oxygen reaches the peat, and stored carbon escapes as CO₂ — a drained bog can emit 10–20 tonnes of CO₂ per hectare every year.

Northwild’s 6,200 hectares span six catchments: two former forestry plots where the trees had been felled, and four areas of open bog crossed by 412 km of drainage ditches.

The problem: a number nobody trusted

Restoring peat is well understood: block the ditches, raise the water table and let sphagnum moss return. Proving how much carbon that saves is much harder. Registry default models estimate emissions from broad condition categories — “drained” or “rewetted” — rather than measuring them, so they come with wide uncertainty bands.

For Northwild, that meant every credit carried a large discount. Buyers priced in the worst case, and the project was selling credits for roughly 60% of what comparable measured credits fetched.

We knew the restoration was working — you could see the bog coming back. We just couldn’t prove it in a way a buyer would accept.

— Elena Marsh, Project Director, Northwild

What we measured

Before a single ditch was blocked, our field team installed a baseline network. Stations were placed by stratified sampling, so every land type was represented in proportion to its area:

• 12 flux towers — eddy-covariance systems measuring CO₂ and methane exchange every 30 minutes

• 340 water-table loggers — recording depth below the surface every hour

• 420 soil cores — sampled to one metre and repeated each autumn

• 18 acoustic monitors — tracking birds and insects as an indicator of habitat recovery

Open bog in the eastern catchment, one year after rewetting.

All data streams to the Terrapulse platform, where it is quality-checked, gap-filled with our open-source pipeline (v2.4) and linked to satellite imagery to scale results across the whole site.

What the data showed

Water tables responded fast. Within six weeks of ditch blocking, mean depth rose from 34 cm below the surface to 9 cm — close to the level of an intact bog. By the second summer, net CO₂ emissions across the rewetted areas had fallen by 74%.

Over three years, the site avoided and removed 41,000 tCO₂e, against a default-model estimate of 32,000. Our figure carries ±5% uncertainty; the model’s carried ±38%.

Methane was the biggest surprise. Rewetted peat produces some methane, and default models assume a large increase. Our chambers measured a much smaller rise — 18% rather than 40% — because water tables settled just below the surface instead of flooding it.

What didn’t go to plan

Two of the 340 loggers failed in the first winter, and a storm in January 2024 damaged a flux tower in the western catchment, leaving a five-week gap. We filled it using neighbouring towers and widened the uncertainty for that period. Both issues are disclosed in the verification report.

The former forestry plots also recovered more slowly than open bog. Felled brash left on site kept water tables lower for longer, so Northwild has since mulched those areas.

The result

An accredited verifier audited the raw flux data, processing code and calculations in early 2025 and closed with no major findings. With measured, low-uncertainty results, Northwild sold its first issuance at a 32% premium to modelled credits, and two buyers signed forward agreements for the next five years.

What’s next

Monitoring continues for the full 30-year crediting period. In 2026 Northwild is extending the network to a further 2,400 hectares, and its data is helping recalibrate the regional default peatland model — so other projects benefit from measured results too.