Field notes, methods and honest results.

Method

Why survival rate is the only planting number worth quoting

The number almost every organisation reports is trees planted. It is the easiest number to produce and the least informative one available. A sapling in the ground is an input. It tells you what was spent, not what was achieved.

The number that matters is how many of those saplings are alive after their third summer. Three years is roughly when a native sapling on a restored site has either established a root system that can find water on its own, or has not. Before that, survival is mostly a measure of how much watering happened. After it, survival is a measure of whether the planting was ecologically sensible in the first place — right species, right spacing, right site.

Why the gap between the two numbers is so large

Planting is a single event with a photograph attached. Aftercare is dozens of unphotogenic visits spread over years, in seasons when nobody wants to be in a field. The economics of attention push every organisation toward the first and away from the second. That is not usually dishonesty; it is drift.

What we do instead

• Every planting is recorded with location, species, count and date at the time it happens
• Sites go onto an aftercare schedule before the drive is considered complete
• Survival is counted on return visits, not estimated from the planting figure
• Where survival is poor, we publish that and change the species mix

The Tree Health Monitor exists to make step three affordable. Walking every plot every season does not scale for a volunteer organisation; estimating canopy vigour from repeat imagery and then sending people to the plots that look worst does.

The honest consequence is that our published numbers will be smaller than a planting count would be. That is the point of them.

Back to the top

Field note

Cleaning the same beach twice is a data problem

If a site needs cleaning again six months after you cleaned it, the cleanup was a maintenance activity, not a fix. That is sometimes the correct thing to do — some sites will always need maintenance. But it should be a decision, not a surprise.

Repeat sites are telling you something specific: waste is entering the system at a predictable point, in a predictable way. A drain that floods. A market with nowhere to put its packaging. A stretch of road with no bin for two kilometres. Until that entry point changes, the site regenerates its own rubbish.

Composition beats weight

Most cleanups report bags collected or kilograms removed. Neither tells you where the waste came from. Composition does. A shoreline dominated by single-use bottles has a different upstream cause than one dominated by fishing gear or construction debris, and a different fix.

This is what the AI Waste Detection work is for: turning the photographs volunteers already take into a composition record, without asking anyone to fill in a form at the end of a long, hot morning.

The practical change

• Log the site, not just the drive
• Record composition, not only volume
• Flag any site cleaned more than once in twelve months
• For flagged sites, look upstream before scheduling another clean

None of this is technically difficult. It is just a different habit from the one most cleanup culture has built.

Back to the top

Explainer

What a carbon credit actually requires

There is a widespread impression that planting trees produces carbon credits. It does not. Planting is the first step of a long process, and most planting projects never complete it — often for good reasons.

The chain, in order

1. A methodology

   A published, approved set of rules for this project type. You do not invent your own.

2. A baseline

   What would have happened at this site anyway. Credits only count the difference.

3. Additionality

   Evidence the project would not have happened without the finance. This is where many projects fail.

4. Leakage

   Whether protecting this forest simply moved the pressure to the next one.

5. Permanence

   How long the carbon stays. A forest that burns in year eight did not sequester carbon; it borrowed it.

6. Monitoring

   Repeated measurement over years, to an agreed standard.

7. Validation, then verification

   An independent third party checks the design, and later checks the reported results. Not the project developer.

8. Issuance

   Only then does a registry issue anything.

Where an AI tool honestly fits

Our Green Carbon Calculator sits at the estimation stage, and only there. It can produce an indicative figure from species, count, age and site conditions. An indicative figure is a planning tool. It is not a credit, it is not tradeable, and describing it as one would be misleading regardless of how good the model is.

What we will say is this: environmental action, measured and monitored properly, may become eligible for carbon credits after appropriate validation and verification. That sentence is longer and less exciting than the alternative, and it is the only one that is true.

Back to the top