Curating demographic and spatial baseline datasets
A baseline brings several kinds of evidence together: official population counts, household survey estimates, historical district descriptions, maps, and project fieldwork. A useful dataset collection preserves the differences between them. It records what each source measures, where and when it applies, how it can be joined, and which uses are permitted.
Use this workflow to maintain the evidence behind the Maluku and Nepal survey designs. The spatial baseline guide develops geographic linkage and exposure analysis; the questionnaire guide explains how source definitions inform project questions.
Start with the question and the unit
Write the decision before choosing files. A question about changes in household access to water needs different evidence from a district population comparison. Specify the population, observation period, geographic unit, outcome, and comparison required by the decision.
| Evidence | Suitable contribution | Limit to preserve |
|---|---|---|
| Published census tables | Population and household context for the published geography | Aggregates do not identify affected households or establish resource rights |
| Household living standards survey | Welfare estimates for the domains supported by its sampling design | A small project area is not automatically a valid estimation domain |
| Historical district profile | Earlier local conditions, institutions, terminology, and source leads | Publication year, observation year, and historical boundaries may differ |
| Administrative boundary | Geographic linkage and reporting | A matching place name does not prove a matching statistical unit |
| Land cover, roads, and services | Environmental context and explicit access models | Classification or proximity does not establish actual use, tenure, or causation |
| Project household and participatory evidence | Verified losses, access, priorities, and lived experience | Detailed records need consent, controlled access, and disclosure review |
Use a collection plan to identify evidence gaps and arrange fieldwork or source requests. Establish project household rosters through an appropriate listing and consent process.
Nepal statistical sources
The NSO census results portal and its ward download catalogue provide starting points for the 2021 census. Retain the table title, reference date, indicator definition, denominator, and geographic identifiers with every extract. Keep missing, suppressed, and zero values distinct.
The official administrative code table is a reference for linking statistical units. Store codes as strings. A ward number needs its parent local level and district; names alone are insufficient. A crosswalk should record the source code system, the boundary edition, and any split, merge, or uncertain correspondence.
The Nepal Living Standards Survey IV catalogue documents household welfare evidence collected in 2022–2023. Preserve weights, design variables and reporting domains alongside the estimates. Record NSO's access conditions and the written permission required for any redistribution of microdata.
The older district profiles in the searchable Nepal index remain valuable as historical evidence. Preserve the original title and district coverage. Where a source covers an earlier combined district, record that relationship rather than silently treating it as two independent current-district observations.
Bhutan and comparative mountain research
Bhutan's National Statistics Bureau maintains the population and housing census catalogue, living standards survey catalogue, and statistical geographic code resources. Select a specific edition and its documentation before making a comparison. A census count, survey estimate, and population projection describe different evidence.
Nepal districts and Bhutan dzongkhags are administrative contexts, not equivalent statistical samples. Comparisons should align the outcome definition, denominator, price basis, survey or census period, geographic coverage, and uncertainty. Report the remaining differences rather than hiding them in a single combined indicator.
The Bhutan agriculture statistics notebook and plotting discussion offer historical examples of organizing and visualizing socioeconomic tables. Check their source definitions, geographic codes and file dependencies before adapting the calculations. The Bhutan demographic and risk discussion identifies further national and spatial sources.
Mountain access research also needs seasonal assumptions. Roads, river crossings, slope and service availability can change the routes people use. The migration spatial example demonstrates the calculations with fictional settlements and seasonal network closures. Its results describe the synthetic scenario.
Build a source record that can be reused
Create one curated record for an identified dataset edition. Link it wherever that edition is used, and retain alternative copies as archive candidates. A filename or an S3 folder is a discovery lead; neither establishes the source, observation date, or licence.
| Record field | What to retain |
|---|---|
| Identity | Provider, original title, edition, official landing page, persistent identifier where available |
| Time | Observation or fieldwork period, publication date, retrieval date, and any revision date |
| Geography | Country, administrative level, code system, boundary version, extent, and coordinate reference system |
| Measurement | Unit of observation, indicator definitions, denominator, missing-value codes, and survey design |
| Files | Format, byte count, checksum or object version, companion files, and a recovery reference |
| Access | Who may inspect the original, reuse the data, see an output, or download a copy |
| Lineage | Transformations, exclusions, join keys, scripts, validation results, and articles using the result |
Use checksums to establish byte identity and source documentation to establish provenance, definitions and reuse conditions. Retain licence evidence, including terms for derivatives or commercial use.
Connect statistics and maps explicitly
Begin with a small table of geographic keys and the indicators needed for the analysis. Validate uniqueness on both sides of the join. Save unmatched rows, duplicate keys and their resolutions as an audit output.
Keep the statistical geography separate from the project footprint. A project buffer intersecting a ward does not show how many households lose access or assets. If an areal allocation or population model is necessary, label the allocation rule, preserve the original totals, and test how conclusions change with the assumptions.
For an access analysis, retain the road and crossing version, travel mode, seasonal restrictions, destination availability, and assumed travel speeds. Check important routes against local knowledge. Combine those results with household and participatory evidence before making a finding about impact or mitigation.
The spatial baseline article develops these safeguards. The quantitative lab complements them with synthetic examples of weighting, sampling uncertainty, price adjustment, and follow-up response.
Keep originals remote and working views small
Retain the original edition in controlled storage and record its identity before creating a working extract. Select the required rows, columns, geographic area, and resolution. Preserve the script and the relationship between the original and the extract so the result can be reproduced without keeping every large file on a laptop.
A shapefile requires its companion files and projection information. A cloud optimized raster can support range reads where the service and reader permissions allow them. Choose the smallest useful analytical view: an aggregate table, spatial subset or reviewed map. The map-sharing workflow explains how to reuse OpenGeo outputs or retain a local rendering recipe alongside the source archive.
Curate and attach in the site editor
Authors can use AWS sources to browse or search a connected repository. Connect a selected object to Source inventory and record its provider, edition, dates, geography, definitions and reuse terms before attaching it to a notebook. Review alternative archive copies against that record, preserving the selected edition and any reason for preferring it.
Reuse the source record in each article that depends on it. Manage access to the original and delivery of restricted notebook cells separately. Record disclosure and reuse decisions for public outputs as part of their provenance.
Public readers should receive the official source link and reviewed discussion. Signed-in readers can receive an archived original only when both their permissions and the source's recorded reuse terms allow it. The reference guide supplies the live bibliography and the managed archive table for this series.