Spatial analysis for baseline surveys

Spatial analysis connects households and livelihoods with the places and processes through which impacts occur. Use it to investigate sampling-frame coverage, define exposure and assess access and change. Interpret locations alongside verified resource use, tenure, household evidence and consultation.

The geospatial assessment method explains how to select a spatial method for an appraisal question. This guide develops survey applications for Maluku and Nepal. The workflows are study-design proposals; the linked country and project discussions identify the evidence available for specific applications.

Use the dataset curation guide to connect demographic tables, administrative layers and source documentation before mapping them. The migration spatial worked example demonstrates travel routes, settlement demand and service capacity with synthetic data, including the distinction between a local service shortfall and a pooled regional total.

Assemble layers according to the question

Begin with the decision and impact pathway. For access to care, assemble settlements, facilities, usable transport links and barriers. For land acquisition, assemble the project design, parcels or use areas, and verified rights and livelihood relationships. Define the study area through these mechanisms, with any screening buffers justified by the relevant process.

Choose data with suitable dates, resolution, definitions, and reuse permissions. WorldPop's methods resources describe modeled population approaches; use the documentation for the specific product. A population raster can assist scoping or allocation, but it is not a household listing or proof of occupancy (references).

Check geometry, units, and dates

Validate geometries and coordinate reference systems before calculating. Latitude and longitude are angular coordinates; a planar buffer of 500 in a geographic CRS is not a 500-meter buffer. Use a suitable projected CRS for a limited study area or an appropriate geodesic method. Document the choice, particularly when the analysis spans multiple zones or islands.

Retain project design version and observation dates. A footprint from an early design may not represent the approved alignment. A facility layer can persist after a service closes. Survey coordinates can be imprecise, absent, displaced for confidentiality, or recorded at an interview location different from the asset or residence.

For raster summaries, identify whether cells contain counts, densities, classes, or probabilities. Sum population counts with appropriate handling of partial cells; multiply density by cell area before summing. Use resampling suitable for the variable. Bilinear interpolation changes category codes and is inappropriate for categorical land cover.

Use exposure to improve sampling

Create draft exposure strata from mechanisms such as permanent acquisition, temporary access restriction, and broader construction influence. Field-verify the classification, then freeze the frame and strata used for selection. Retain later corrections so the analysis can account for them.

For a coastal study in Maluku, examine links among residences, landing places, fishing grounds, gardens and markets. For a Nepal corridor study, examine paths and resource access that construction could interrupt, including routes used by households farther from the alignment. Record residence and activity locations separately, with their source and observation period.

If field verification, the study population or the project footprint changes after selection, document the revised classification and treatment of earlier observations. Preserve selection probabilities and the basis for any change in the analysis population.

Calculate a defined exposure measure

For household using parcels , a proposed area exposure measure is

The numerator and denominator must refer to the same period and type of use. Avoid double-counting overlapping geometries within the household's used land. Shared or customary areas require an explicit allocation or a separate relational measure; assigning the entire shared area to every user can inflate aggregate totals.

An exposure measure of 0.25 describes a quarter of the defined used area. Estimating livelihood loss also requires productivity, access, timing, replacement opportunities and mitigation. Analyze exposure alongside activity records and field evidence, retaining separate measures for distinct pathways.

Change the restriction width to inspect the intersection calculation. The example keeps the household's three used parcels and their total area constant. A 40-metre corridor overlaps 0.40 hectares of 1.72 hectares used, giving 23.3% area exposure. A restriction's duration, season and effect on access would be additional variables in a livelihood assessment.

For access, compare observed or validated baseline travel time with modeled travel time under the proposed design. State mode, destination, season, route assumptions, and restrictions. Straight-line distance offers a screening measure; it does not capture steep terrain, boat connections, road conditions, or affordability.

Integrate statistics at their own geographic level

Attach a district statistic as contextual information, with district ID, year and source. Household poverty or eligibility requires evidence at the corresponding unit. Preserve that distinction in the data model and map legend to avoid ecological inference errors.

For boundary changes, use a documented correspondence and suitable common geography. Keep observed, aggregated, and modeled values distinct. In the Nepal index, paired records sharing older references demonstrate why documentary and geographic units need separate identifiers.

For Nepal, begin with the historical district source register, then select current official statistical geography and the appropriate table. Keep a join audit with matched, unmatched, duplicated, split, and uncertain records. A join that produces no row does not establish zero population. Name-normalization rules can help identify candidates, but accepted matches need a recorded basis.

For Himalayan land cover, distinguish the data's represented year from its upload date and the report's publication date. Identify the national report, downloadable raster and map service separately, recording the edition relationships between them. The Nepal and Bhutan source workflow connects these records to demographic context.

The Bhutan geospatial notebook connects directly to ICIMOD's HTTPS services, and the Bhutan geospatial source discussion retains the historical boundary comparisons. Both identify source editions and load detailed layers on request. The earlier ICIMOD Nepal land-cover notebook remains a historical source lead whose service connections require checking before reuse. Use the synthetic migration example above to explore the maintained access-model demonstration.

Measure change using consistent definitions and geography. A land-cover difference may indicate environmental change but does not alone show which household lost access or whether the project caused it. Triangulate imagery, timing, design records, resource-use observations, and interviews. A change in satellite appearance and a change in welfare are separate findings until a relationship is demonstrated.

Protect locations and explain the map

Exact household locations, contested claims, sensitive identity, and culturally important sites can identify people or expose information they did not agree to share. Keep precise locations in controlled analytical records. Prepare public outputs with suitable aggregation, suppression, or permitted generalized geometry. Adding random displacement does not guarantee anonymity and can invalidate fine-scale spatial inference.

Every published map needs a clear analytical unit, legend, scale where appropriate, dates, sources, and description of modeled quantities. Use labels or patterns alongside color; choose an ordered scale for ordered quantities. Provide a readable table or text equivalent. Show unavailable areas explicitly so readers do not interpret absence of data as absence of people or risk.

Publish a compact analytical view: the selected geography, permitted indicators, transformation recipe and source citations. Retain detailed originals in their managed repository, with documented access terms and disclosure decisions for derived outputs.

The Bhutan demographic and risk discussion applies these principles to national statistics, boundaries, terrain and hazard evidence. The spatial analysis topic links related assessment methods and country notebooks. The map-sharing discussion connects the Maluku and Nepal maps with their source records and publication choices.