NCGS-M002 — Stratified-sample estimate of tree state at a declared horizon
- Code: NCGS-M002
- Version: NCGS-2026-09-24
- Approval status: Approved
An estimate of the survival family from a sample of tree individuals drawn per stratum, made on monitoring visits at a declared observation horizon under a monitoring plan whose sampling design is sample: every sampled tree reached is observed as the census does, and the site’s totals and rates are estimated with stratum weights, a standard error and an interval. The alternative to NCGS-M001 for a planting too large to census.
underlying principles and assumptions
The tree individual is the unit of observation (ADR-0002); the sampling frame is the stratum allocation of the registered design and the sample is drawn before the visit; the plan declares the design and the strata sampled (ADR-0016); a horizon is a checkable window (ADR-0010); an unobserved sampled tree is unobserved, never dead; an estimate carries its uncertainty or it is not reported; rates roll up by weighted sums of estimated numerators and denominators, never by averaging.
methods and procedures
At design, declare a sample design with the strata sampled; before the visit, draw n_h tree individuals per sampled stratum h from the planting records by a stated random procedure and record the draw; at the site, observe each sampled tree as NCGS-M001 does; compute per-stratum proportions and weight them by the stratum’s planted count N_h; report the estimate with its standard error and a 95 percent interval.
data requirements and inputs
A registered design whose monitoring plan declares a sample design and the strata sampled, with the strata’s indicative counts; conformant planting records for the site; the recorded sample draw; monitoring visit records at a declared horizon whose activity applies NCGS-M002; one tree observation per sampled tree reached.
calculation and transformation rules
For each sampled stratum h with N_h trees planted and n_h sampled trees observed, a_h alive: p_h = a_h / n_h; estimated trees alive A = sum over h of N_h p_h; estimated survival rate S = 100 × A / sum over h of N_h; the same weighting gives trees by condition class, establishment at T3 or later and trees dead by cause; the standard error of S is the stratified estimator’s, sqrt(sum over h of (N_h / N)^2 p_h (1 - p_h) / (n_h - 1) × (1 - n_h / N_h)), and the interval S ± 1.96 SE; monitoring visits conducted is the count of visit records. A stratum the plan does not sample contributes nothing and is reported as unsampled.
outputs and units of measure
Estimated trees alive at horizon (count, NUM); estimated survival rate at horizon (PERCENT) with standard error and interval; estimated trees by condition class (count, NUM); estimated establishment rate at T3 or later (PERCENT); estimated trees dead by cause (count, NUM); monitoring visits conducted (count, NUM); each disaggregated by horizon, species, stratum, site and census place.
validation and verification approach
Verification as for the census: a reviewer meeting the KSR requirement in impact evaluation and estimation and in field survey and data capture, who is not the observer, checks that the plan declares the sample design, that every observed tree is in a sampled stratum and in the recorded draw, that the draw was random as stated, the offsets against the declared horizon and the photograph per tree, and recomputes the weights and the standard error.
limitations and applicability boundaries
An estimate, never a count: a funder wanting each tree’s fate needs the census. Precision falls with small n_h; a stratum with one sampled tree yields no standard error. The frame is the planting records, so owner-planted delivered-only trees are outside it. Defined for the monitoring visit activity type only.
rationale
KIA-Adapt’s A4 estimator weighted strata this way; a small planting can be censused, but the standard must not make the census the only admissible way to monitor a planting of a thousand sites, and the plan already declares the sampling design.
Applicability
- NCGS-M002: a sampled visit observes only trees in a stratum the plan samples; a tree outside the sampled strata is outside the frame.
- NCGS-M002: the estimator applies under a sample design declared on the monitoring plan the trees’ planting activity, or a delivered tree’s handover activity, runs under; a census plan is verified under NCGS-M001 (ADR-0016).
- NCGS-M002: a sampled visit declares a horizon.
Indicators it serves
Activity types
- Monitoring visit
Knowledge domains a reviewer needs
- Field survey and data capture
- Impact evaluation and estimation
Variables
| Variable | Role | Unit | Description |
|---|---|---|---|
| Estimated establishment rate (E) | DerivedOutput | PERCENT | E = 100 × (sum over h of N_h × thriving_h / n_h) / sum over h of N_h, at T3 or later. |
| Estimated survival rate (S) | DerivedOutput | PERCENT | S = 100 × A / N, reported with SE and the interval. |
| Estimated trees alive at horizon (A) | DerivedOutput | NUM | A = sum over sampled strata h of N_h × a_h / n_h. |
| Estimated trees by condition class (C_k) | DerivedOutput | NUM | C_k = sum over h of N_h × (sampled trees in class k in h) / n_h. |
| Estimated trees dead by cause (D_c) | DerivedOutput | NUM | D_c = sum over h of N_h × (sampled trees dead of cause c in h) / n_h. |
| Interval of the estimated survival rate (S ± 1.96 SE) | DerivedOutput | PERCENT | The 95 percent interval S ± 1.96 SE, reported with S. |
| Monitoring visits conducted (|V|) | DerivedOutput | NUM | |V| = count of sampled monitoring visit records at the site with at least one observation. |
| Standard error of the estimated survival rate (SE) | DerivedOutput | PERCENT | SE = 100 × sqrt(sum over h of W_h^2 × p_h (1 - p_h) / (n_h - 1) × (1 - n_h / N_h)); undefined for a stratum with n_h = 1. |
| Stratum weight (W_h) | DerivedOutput | UNITLESS | W_h = N_h / N, the stratum’s share of the trees planted at the site. |
| Sampled tree alive (a_i) | MeasuredInput | UNITLESS | |
| Sampled tree cause of death (cd_i) | MeasuredInput | UNITLESS | |
| Sampled tree condition (c_i) | MeasuredInput | UNITLESS | |
| Sampled tree stratum (h_i) | MeasuredInput | UNITLESS | |
| Strata sampled (H) | MeasuredInput | UNITLESS | |
| Trees planted in stratum (N_h) | MeasuredInput | NUM | The stratum’s planted count from the registered design’s stratum allocation, reconciled to the planting records. |
Monitoring requirements
- NCGS-M002 monitoring requirement: sampled visits at each planned horizon: One visit per site per horizon the plan reaches, observing every sampled tree reached; the draw recorded before the visit; an unobserved sampled tree reported as unobserved.