Popia Spatial Audit

-- Check if spatial data can be joined to residential parcels with owner info
SELECT COUNT(*) FROM imported_data d
JOIN cadastral_parcels p ON ST_Within(d.geom, p.geom)
WHERE p.land_use IN ('RESIDENTIAL', 'RESIDENTIAL_MIXED');

Risk levels:

Step 2 — Check Flight Path → Individual Tracking Potential

For aviation/drone data:

• Private aircraft registration → owner identification (HIGH risk)
• Repeated flight patterns → personal movement inference (HIGH risk)
• Commercial airline flights → no individual tracking (LOW risk)

function assessFlightPrivacy(track: FlightTrack): RiskLevel {
  if (track.callsign?.match(/^(SAA|FA|MN|KQ)/)) return 'LOW';    // Commercial
  if (track.isPrivate && track.repeated > 3) return 'HIGH';        // Pattern
  return 'MEDIUM';
}

Step 3 — Verify Aggregate-Only Display for Guest Mode

Guest users must NEVER see:

• Individual property owner details
• Individual movement tracks
• Point-level data at residential addresses
• Any data linkable to a specific person

function enforceGuestAggregation(features: Feature[], userRole: Role): Feature[] {
  if (userRole === 'GUEST') {
    return aggregateToSuburb(features); // Aggregate points to suburb polygons
  }
  return features;
}

Step 4 — Check Temporal Data for Movement Pattern Inference

Temporal + spatial data creates movement profiles:

Risk Matrix:
  Spatial precision HIGH + Temporal precision HIGH = CRITICAL RISK
  Spatial precision HIGH + Temporal precision LOW  = HIGH RISK
  Spatial precision LOW  + Temporal precision HIGH = MEDIUM RISK
  Spatial precision LOW  + Temporal precision LOW  = LOW RISK

Mitigations:

• Snap timestamps to 1-hour intervals for guest display
• Reduce spatial precision to suburb centroid for aggregate views
• Never store individual movement sequences without explicit consent

Step 5 — Ensure Spatial Data Retention Policies

-- Enforce retention via scheduled cleanup
DELETE FROM api_cache WHERE expires_at < NOW();
DELETE FROM flight_tracks WHERE created_at < NOW() - INTERVAL '90 days';

Step 6 — Validate Consent for Location-Based Personal Data

Required consent checks:

interface SpatialConsentChecklist {
  dataContainsLocationPII: boolean;      // Can location identify a person?
  consentObtained: boolean;              // Explicit opt-in?
  consentScope: string;                  // What was consented to?
  purposeLimitation: string;             // Specific use case
  rightToErasure: boolean;               // Can we delete on request?
  crossBorderTransfer: boolean;          // Data leaving SA?
  dataMinimisation: boolean;             // Only necessary precision?
}

Output

POPIA Spatial Audit Report:

═══════════════════════════════════════
 POPIA SPATIAL AUDIT REPORT
 Dataset: [name]
 Date: [ISO 8601]
 Auditor: [agent]
═══════════════════════════════════════

 Risk Score: [LOW | MEDIUM | HIGH | CRITICAL]

 ☐ Residential parcel linkage:  [PASS | FAIL | N/A]
 ☐ Individual tracking potential: [PASS | FAIL | N/A]
 ☐ Guest mode aggregation:      [PASS | FAIL | N/A]
 ☐ Movement pattern inference:   [PASS | FAIL | N/A]
 ☐ Retention policy compliance:  [PASS | FAIL | N/A]
 ☐ Consent validation:           [PASS | FAIL | N/A]

 Required Annotations: [list of files needing POPIA blocks]
 Recommended Actions:  [specific mitigations]
═══════════════════════════════════════

When NOT to Use This Skill

• Non-personal geographic data (zoning boundaries, terrain elevation, land cover)
• Aggregate statistical data with no individual linkage
• Public infrastructure data (roads, utilities) with no owner information
• Mock/test data in public/mock/ with synthetic coordinates