Measure more. Or decide now?
Lakeside spruce · 32 ha · What could another survey change?
Survey may be worthwhile
A higher-yield signal favours selective thinning; a lower-yield signal favours current plan. The survey can change your action, not just reduce uncertainty.
Two signals. Two decisions.
Bayes’ ruleCurrent plan
50% chance of signal
15%updated chance of higher yieldSelective thinning
50% chance of signal
85%updated chance of higher yield- Best expected NPV now · Selective thinning
- €278,012
- Expected best NPV after survey
- €285,953
- Expected value of sample information (EVSI)
- €7,940
- Perfect-information upper bound
- €21,663
Survey assumptions
Illustrative, not a quote32 ha × €70/ha + €250 mobilisation. Information is assumed relevant to all 32 ha of this stand.
Inspect the teaching model
All three properties, stand boundaries and tree samples are fictional. Last illustrative observation: 15 September 2026. Areas reconcile to property totals. No cadastral or LiDAR data is used.
Demonstration calculations & units
20 annual periods. Starting volume 245 m³/ha; annual growth 5.8 m³/ha/year. Mortality 0.6% of starting annual volume. Volume end = start + growth − mortality − harvest, constrained to nonnegative values.
Living-tree carbon = volume × 0.42 t dry stem biomass/m³ × 1.3 biomass expansion × 0.5 t C/t dry biomass = 0.273 t C/m³. Carbon is reported in tonnes of carbon (t C), not tonnes CO₂. Soil, harvested-wood products and credits are excluded.
Timber €55/m³ × price multiplier. Harvest cost €18/m³ + €120/ha mobilisation per harvest. Annual management €12/ha; habitat retention adds €8/ha/year. On measurement-sensitive stands, selective thinning includes an explicitly synthetic €1,700/ha intervention cost in year 1; this teaches state-dependent trade-offs, not a commercial price. Harvest fractions: current/selective 28%, carbon 8.4%, habitat 15.4%. Harvest years: 10 / 1 / 15 / 12 respectively.
Post-harvest growth multiplier: 1.08 (current/carbon/habitat); selective 1.1 on stable stands. Sensitive stands: max(0, 1 + 1.8 × (yield factor − 0.6)). This synthetic response intentionally demonstrates decision switching.
NPV = Σ cashₜ / (1 + r)ᵗ + terminal value / (1 + r)²⁰. Terminal value = year-20 volume × max(0, €55 × price multiplier − €18). Included once; not immediate income. Zero discount means undiscounted sums.
Deadwood: standing end = 94% of previous standing + 60% mortality; fallen end = 96% previous fallen + 6% previous standing + 40% mortality. Habitat trees retained: current 80%, selective 65%, retention 100%, rounded per hectare. These are individual unvalidated proxies.
Which input changes what?
Discount changes NPV only. Timber-price multiplier changes harvest revenue and terminal value only. Thinning intensity changes harvest, future volume, carbon, mortality and financial results. Carbon minimum (% initial living stock) and habitat minimum (trees/ha) filter alternatives; they do not modify forecasts.
Low/central/high sensitivity is 0.45× / 1× / 1.55× annual growth, including the explicit release response. These are sensitivity assumptions, not statistical confidence intervals.
Information-value teaching model
Two states: lower yield 0.45×, higher yield 1.55×. Prior is the chance of higher yield. Symmetric survey accuracy is P(high signal | high yield) = P(low signal | low yield). Bayes updates each signal posterior. Actions must satisfy your constraints in both states. The survey can affect management of the entire selected stand; cost uses surveyed hectares only.
EVSI is the signal-weighted best expected NPV minus the best expected NPV now. Ecological benefits are not priced. No proprietary R-TIM matching or LiDAR processing is implemented.
Future integrations
Real tree observations, owner-authorised boundaries, validated growth models and professional assessments are future integrations. No email, booking, authentication or external account is connected.