A single-file web tool that walks an operator through a UReDI v3 dataset assessment and produces a reproducible Priority score. No build step, no dependencies, no server. Open the HTML file and score.
Live tool: https://sdiggs.github.io/uredi-calculator/
Tom Karl's UReDI framework (Urgency, Recoverability Difficulty, Importance) attempts to quantify how much a dataset is worth and how hard it would be to recover. It's one of three frameworks in an ensemble approach to climate data preservation:
- UReDI answers how much a dataset is worth and how hard it would be to recover.
- R* answers how likely we are to lose it.
- The Keeling Curve Foundation's Long-Term Observations inventory answers which measurement series must keep running.
Value, risk, and continuity are three different questions. Run together they catch what any one framework alone would miss.
This calculator is the companion to the R* Calculator and follows the same single-file, zero-dependency design.
The tool is built for non-expert operators. Someone who is not a climate data specialist should be able to run a defensible UReDI assessment by answering a sequence of plain-language questions, each backed by a concrete 1 to 5 rubric.
The design goal is consistency:
- Repeat-run consistency. The same operator scoring the same dataset twice should land in the same place.
- Cross-operator consistency. Two operators of different experience levels scoring the same dataset should land close enough to agree on the result, even if neither score is "optimal."
Every component is framed as an explicit question. The five rubric levels are the answers. Objective components (storage size, record length, spatial extent) leave little room for divergence; the more judgment-based components are where operator experience shows, which is why the tool captures per-component reasoning and an operator experience self-rating with every run.
UReDI scores 15 sub-components, each on a 1 to 5 ordinal rubric, organized under three dimensions.
Urgency (U) — a single direct score for near-term risk of loss.
Recoverability Difficulty (RD) — how hard the dataset would be to recreate:
Rs— structural uniqueness (carries two-thirds of the RD weight)ETC— expertise costSC— storage costPC— processing cost
Importance (I) — how critical the dataset is across ten components:
SAS,PCS,PRS— scientific, public-sector, and private-sector usageTL,TR— temporal length and resolutionSE,SR— spatial extent and resolutionI_int— integration into analytical systemsD— downstream dependenciesQ— data quality
RDc = ( ETC + SC + PC ) / 3
RD = (2/3 x Rs) + (1/3 x RDc) RD* = round( RD )
I = SAS + PCS + PRS + TL + TR + SE + SR + I_int + D + Q
I* = round( I / 10 )
Priority = ( U x RD* x I* ) / 125 x 100
Importance sums additively (raw range 10 to 50, normalized to a 1 to 5 score). The final Priority is a multiplicative roll-up of the three dimension scores, producing a 0 to 100 result. Because it multiplies rather than averages, a low score on any single dimension pulls the whole Priority down. That is intended behavior: it concentrates attention on datasets that are simultaneously valuable, vulnerable, and hard to bring back.
Tiers are provisional, calibrated against the six worked examples in UReDI v3 Appendix B and meant to be re-calibrated once enough real datasets have been run.
| Tier | Priority |
|---|---|
| Critical | 65 to 100 |
| High | 40 to 64 |
| Moderate | 20 to 39 |
| Low | 0 to 19 |
- Open the tool. All sections start collapsed.
- Dataset & Curator — record the dataset name, its persistent identifier (DOI, ARK, or handle), and the primary curator. The PID field does light type-detection so the same dataset is logged identically across operators.
- Operator — optional but strongly encouraged. The experience self-rating in particular is what lets cross-operator variance be measured later.
- Score the 15 components under Urgency, Recoverability Difficulty, and Importance. Pick the one rubric level that best fits, and use the reasoning box to record judgment calls.
- Read the Priority score, tier, and breakdown in the results panel as you go.
- Export the finished assessment as JSON, copy it to the clipboard, or print to PDF.
The Worked Examples panel holds six datasets scored in UReDI v3, with the expert score for every component. Use them to calibrate your own judgment before or during an assessment.
The JSON export is the canonical record of an assessment. It captures the dataset and curator metadata, the operator metadata and experience rating, all 15 component scores with their selected rubric text and per-component reasoning, and the computed results (RDc, RD, RD*, I-raw, I*, the UReDI triplet, raw product, Priority, and tier). The structured format is what makes later cross-operator and cross-run analysis possible.
This is a single, self-contained HTML file. To deploy on GitHub Pages, place the app at the repository root as index.html and enable Pages for the main branch. No build pipeline is required.
uredi-calculator/
├── index.html # the application (single file)
└── README.md
Working build, under active road-testing and iteration. Known items for upcoming passes:
- Reference links. Most per-component "References & resources" entries are placeholders. They will be populated with web-accessible sources where a component genuinely benefits from one.
- Worked examples. The examples panel currently shows score tables and a short interpretation. The full per-component justifications from UReDI v3 Appendix B are slated to be added.
- Tier calibration. The four tier bands are provisional and will be revisited once a body of real assessments exists.
- SE rubric. Polar and basin-scale coverage sits awkwardly between the Regional and Continental rubric levels. Flagged for clarification.
- R* Calculator — the vulnerability framework in the ensemble
- The Ensemble Framework for Climate Data Preservation, which sets out how UReDI, R*, and the Keeling Curve Foundation's Long-Term Observations inventory work together
- UReDI framework (v3): Thomas Karl
- R* Composite Priority Index: Steve Diggs
- Long-Term Observations inventory: Keeling Curve Foundation
- Tool design and implementation: Steve Diggs
Set a license before public release. MIT is a reasonable default for the tool code; note separately that the UReDI framework itself is attributable to Thomas Karl.