Spatial Assignment Workflows

Overview

Spatial assignment is the process of linking point-based events to spatial analysis units.

In riskworkflowr, this usually means assigning point events to:

• administrative boundaries
• custom polygons
• operational areas
• H3 or hexagonal grids

This step is important because all downstream counts, rates, SMRs, and maps depend on the assignment logic used.

Why spatial assignment matters

Spatial joins can appear simple, but different assignment choices can produce different analytical outputs.

For example:

• a point may fall inside a polygon
• a point may fall just outside a polygon due to boundary or geocoding uncertainty
• a point may need to be assigned to the nearest unit
• a point may require an auditable fallback rule

riskworkflowr supports transparent assignment logic so that analysts can understand how events were allocated.

Supported join modes

The main spatial assignment function is:

sj_join()

It supports three join modes:

intersect
nearest
intersect_nearest

Intersect assignment

Intersect assignment links each point to the polygon it falls within.

This is usually appropriate when:

• point locations are accurate
• polygon boundaries are meaningful
• events should only be assigned if they fall inside a unit

sj_join(
  points = event_points,
  polygons = analysis_units,
  id_col = "unit_id",
  join = "intersect"
)

Nearest assignment

Nearest assignment links each point to the closest polygon.

This may be useful when:

• points are near but not inside a polygon
• boundary precision is imperfect
• geocoding uncertainty exists
• every event must be assigned to a unit

sj_join(
  points = event_points,
  polygons = analysis_units,
  id_col = "unit_id",
  join = "nearest"
)

Intersect-nearest assignment

The intersect-nearest workflow first attempts an intersect join and then uses nearest assignment as a fallback.

This is often useful in operational spatial risk workflows because it provides a practical balance between strict spatial containment and complete event allocation.

sj_join(
  points = event_points,
  polygons = analysis_units,
  id_col = "unit_id",
  join = "intersect_nearest"
)

Join auditability

Spatial assignment should be auditable.

Where applicable, riskworkflowr records fields such as:

join_method
join_distance_m

These fields help distinguish events assigned by direct intersection from events assigned using a nearest fallback.

This is important for review, QA, and communication.

Assumptions

Spatial assignment assumes that:

• the point geometry is suitable for analysis
• the analysis unit geometry is valid for the processing context
• the coordinate reference system is appropriate for distance calculations
• the selected join rule matches the analytical question

Limitations and pitfalls

Spatial assignment may be affected by:

• geocoding uncertainty
• boundary misalignment
• invalid or complex geometry
• points located near polygon edges
• use of geographic CRS for distance calculations
• inappropriate fallback assignment rules

Nearest assignment should be used carefully where distance or boundary meaning is important.

Alternative approaches

Other valid approaches may include:

• manual review of uncertain points
• distance thresholds
• buffer-based assignment
• probabilistic assignment
• hierarchical assignment rules
• exclusion of uncertain locations
• sensitivity testing across multiple assignment methods

The appropriate method depends on the data, spatial scale, uncertainty, and decision context.

Future development

Future versions of riskworkflowr may include additional support for:

• distance thresholds for nearest joins
• clearer QA summaries of fallback assignments
• assignment uncertainty flags
• sensitivity testing across join modes
• richer diagnostics for boundary and CRS issues