2 Methods
2.1 Design principles (Hyatt-aligned)
Hyatt (1997) intended estimate-type assignment to reflect (i) method properties, (ii) statistical properties, and (iii) documentation (Hyatt 1997). The updated guidance encodes these dimensions explicitly as decision gates, method-family checks, and final evidence requirements.
| Hyatt (1997) dimension | How the updated guidance encodes it |
|---|---|
| Method properties (survey + analysis) | Method families + property checks (coverage, effort, timing, visibility, cross-section coverage) |
| Statistical properties (units, accuracy, precision) | Type eligibility by method family and a final precision/accuracy gate (supports CV/SE where available) |
| Documentation | Documentation qualifier + final documentation enforcement |
2.2 Decision key design
The classification follows a property-first sequence:
- Data format gate: non-numeric presence/not-detected is classified as Type 6.
- Method known gate: if survey and analytical methods are not identified, the estimate is provisionally Type 5 and flagged as method-unknown.
- Method family selection: a primary method family scopes the applicable questions.
- Method-family checks: coverage, effort, visibility, timing, and related criteria drive conservative downgrades.
- Final checks: documentation and uncertainty evidence are evaluated; where evidence is missing, conservative downgrades apply.
This structure is intended to reduce subjective table interpretation by turning common qualifiers into explicit questions and recorded qualifier codes.
2.3 Software availability (implementation of the guidance)
The guidance is implemented in the SMN Escapement Estimates Toolkit (R Shiny application) (NuSEDS Escapement Estimates Toolkit Working Group 2026). The toolkit executes one canonical decision key and produces (i) a final estimate Type (1–6) and (ii) explicit qualifier codes explaining conservative downgrades.
In this report, the software is treated as a companion implementation rather than a primary publication artifact. To avoid over-emphasizing engineering details for the intended audience, the implementation is referenced only at a high level.
2.4 Handling common operational complications
The updated guidance treats several recurring complications as explicit qualifiers rather than implicit judgement calls.
2.4.1 Breaches/bypass and infilling
Breach/bypass events at counting sites, and periods of missing observation due to outages or missed visits, can materially affect interpretability unless the magnitude and correction method are documented. The updated guidance treats bypass/breach risk and incomplete coverage as explicit downgrade triggers and records whether defensible infilling/interpolation was required (Vélez-Espino et al. 2010; Holmes, Cronkite, and Enzenhofer 2005; See, Kinzer, and Ackerman 2021).
2.4.2 Survey timing relative to run timing
For survey-based indices, visit count alone is not sufficient: surveys must bracket the period when fish are present, and visibility constraints and observer effects can govern bias and comparability (K. R. Holt and Cox 2008; Jones, Quinn, and Van Alen 1998; Korman et al. 2002; Hill 1997). The updated guidance includes timing and visibility checks within method families where these factors are primary drivers of interpretability.
2.4.3 Combined-method estimates
Combined-method workflows (e.g., system-wide sonar with tributary visual apportionment, or fences supplemented during breach periods) should be recorded as explicit components. The updated guidance applies a conservative rule: where multiple components contribute to the final estimate, the assigned Type should not exceed that implied by the weakest component unless a documented integration method supports a higher classification (Parsons and Skalski 2010).
2.4.4 Calibration and historical revisions
Where historical values have been recalibrated or revised, the updated guidance treats calibration as an analysis layer that should be captured in metadata (calibration source, diagnostics, and revision history) rather than as a new estimate Type. This supports interpretation of time-series values by downstream users.
2.4.5 Quantitative uncertainty (CV/SE)
Hyatt (1997) distinguished higher-quality absolute-abundance estimates in part by qualified precision (variance evidence) (Hyatt 1997). Where available, reporting quantitative uncertainty (e.g., CV or SE) improves interpretability for downstream use. For example, AUC and peak-count approaches have established approaches for incorporating uncertainty and evaluating sensitivity to survey timing and frequency (English, Bocking, and Irvine 1992; Hilborn, Bue, and Sharr 1999; Hill 1997; Parken, Bailey, and Irvine 2003; Millar, McKechnie, and Jordan 2012). The updated guidance includes a final precision/accuracy check and recommends capturing quantitative uncertainty when available.
2.5 Recommended metadata elements for New SEDS/NuSEDS
The updated guidance can be expressed using existing NuSEDS fields in many cases, but several high-value metadata elements are needed to make estimate Types interpretable and reproducible for downstream users.
Note: The field names below are illustrative. The key point is that these elements are (i) recorded by the guidance as explicit qualifier codes and (ii) recommended for storage in New SEDS/NuSEDS so that Type assignment is transparent to data users.
| Metadata element | Why it matters for interpretation | Typical qualifier code(s) | Priority for upload |
|---|---|---|---|
| Enumeration method (field) | Scopes which property checks apply. | METHOD_UNKNOWN (if missing) | Required |
| Estimation method (analysis) | Distinguishes analysis pathways (e.g., mark–recapture, AUC, modelling). | METHOD_UNKNOWN (if missing) | Required |
| Primary method family (FS/V/A/S/T/R/P/M) | Controls eligibility rules and the set of required qualifiers. | METHOD_UNKNOWN (if missing) | Required |
| Run-window coverage evidence | Explains conservative downgrades due to missed start/end of run. | RUN_COVERAGE | Recommended |
| Device/site uptime evidence | Explains conservative downgrades due to outages/unobserved intervals. | UPTIME | Recommended |
| Breach/bypass context | Explains conservative downgrades where missed fish are plausible. | BREACH_BYPASS | Recommended |
| Infilling/interpolation method | Distinguishes defensible correction from undocumented gap-filling. | INFILL_METHOD | Recommended |
| Survey timing relative to run timing | A core driver of bias and comparability for survey indices. | TIMING | Recommended |
| Visibility/conditions | Affects detectability and comparability (especially for visual methods). | VISIBILITY / ENV_COND | Recommended |
| Documentation evidence | Determines whether a downstream user can independently interpret the estimate. | DOC | Required |
| Quantitative uncertainty (CV/SE) | Supports qualified precision for Type 2 and more transparent uncertainty. | PRECISION_ACCURACY | Recommended |
| Combined-method components | Prevents concealment of a weaker component within an overall estimate. | (varies; conservative component rule) | Recommended |
| Calibration source + diagnostics | Supports interpretation of recalibrated time series. | (varies) | Recommended |
| Revision history (what changed/when) | Supports transparency when historical values have been revised. | (varies) | Recommended |
2.6 NuSEDS data dictionary alignment
The NuSEDS data dictionary defines the database fields used to store enumeration methods, estimation methods, estimate classification (Types 1–6), and supporting metadata (e.g., inspections/effort and timing fields) (Fisheries and Oceans Canada 2025). The updated guidance is designed to be expressible using existing NuSEDS fields where possible, and to clearly identify gaps where additional metadata would improve interpretation and reproducibility.
| Field Name | Field Definition | Role in updated guidance | |
|---|---|---|---|
| 4 | ESTIMATE_CLASSIFICATION | This categorizes estimates based on their levels of accuracy and precision (Type-1 are the most accurate, Type-6 the least accurate). There… | Stores Type 1–6 estimate classification (includes some legacy non-Type labels) |
| 3 | ENUMERATION_METHODS | The enumeration method used to observe fish. The first method listed is the primary method. Values are: Bank Walk, Based on Angling Catch, … | Primary field method (enumeration) used to scope method-family checks |
| 5 | ESTIMATE_METHOD | There are several standard methods to chose from. Cumulative CPUE - Created for Stikine Sockeye Fixed Site Census - Combining one or more r… | Primary analysis method (estimation) and special cases (combined, calibrated, unknown) |
| 2 | ADULT_PRESENCE | Values are present if adults were observed, none observed if no adults were observed during the stream inspections, not inspected if adults… | Supports presence/not-detected pathways (Type 6 context) |
| 8 | JACK_PRESENCE | Values are present if jacks were observed, none observed if no jacks were observed during the stream inspections, not inspected if jacks we… | Supports presence/not-detected pathways (Type 6 context) |
| 9 | NO_INSPECTIONS_USED | This is the number of stream inspection logs that are linked to the SEN or were used in the analysis. E.g. 10 stream inspections and a fixe… | Supports effort/visit thresholds (VISITS and related downgrades) |
| 13 | START_DTT | This is the time stream inspections began e.g. 2000-10-15 means that the first inspection for this season’s estimate started on October 15 … | Supports timing/coverage interpretation (inspection start date) |
| 12 | RUN_TYPE | Run_Type indicates the run timing for different runs within the same season. In some cases, the runs may be well documented enough to label… | Supports timing context when multiple runs occur in a season |
| 7 | INDEX_YN | This indicates whether the estimates are for a portion of the population. This is usually due by purposely limiting enumeration to a portio… | Flags index (partial coverage) estimates (relative-abundance context) |
| 1 | ACCURACY | This is the ability of a measurement to match the actual value of the quantity being measured. Some historical estimates that were imported… | Legacy qualitative field; not a substitute for quantified uncertainty metadata |
| 10 | PRECISION | This is the ability of a measurement to be consistently reproduced, or put another way, the number of significant digits to which a value h… | Legacy qualitative field; not a substitute for quantified uncertainty metadata |
| 11 | RELIABILITY | This field was added for the inclusion of historical data from an external source. It is the level of reliability that the person placed in… | Legacy/import field (historical); not consistently present |
| 6 | ESTIMATE_STAGE | Preliminary SENs are the first drafts of summary estimate documents. Source data may be incomplete and their accuracy has not been verified… | QA/workflow stage (preliminary/near final/final); not a type determinant |
2.7 Method families
| Code | Method family | Best attainable type |
|---|---|---|
| FS | Fixed site census (manual or electronic) | 1 |
| V | Visual ground or snorkel count | 2 |
| A | Aerial survey count | 3 |
| S | Hydroacoustic sonar count (modelled) | 2 |
| T | Trap model (non-spanning) | 2 |
| R | Redd survey | 2 |
| P | Electrofishing CPUE index | 3 |
| M | Mark-recapture program | 2 |