Validation and Quality Assurance
Quality assurance is not optional - it's a core part of every mapper's responsibility. Before uploading any data, you must validate your work using JOSM's built-in tools and manual inspection. Poor quality mapping creates more work for validators and undermines humanitarian efforts.
JOSM includes powerful validation tools that catch common errors automatically. Never upload without running validation and fixing all errors and warnings.
Part 1: JOSM Validation Tool
The validation panel identifies errors, warnings, and potential issues in your mapping data.
Running Validation
- Before upload: Select all your mapped data (Ctrl+A)
- Open validation: Window → Validation Results (or Shift+V)
- Run validation: Click the 'Validation' button in the panel
- Review results: Errors appear in red, warnings in yellow
- Click each issue to zoom to problem area
- Fix errors before upload - zero errors required
- Address warnings when possible, document if unfixable
Non-Negotiable: You MUST fix all validation errors before uploading. Uploading data with errors violates mapping protocols and may result in removal from projects.
Part 2: Common Validation Errors
Understanding frequent errors helps you avoid them while mapping.
Most Common Errors and Fixes
- Duplicate nodes: Two nodes at same location - Select both → Tools → Merge nodes
- Unconnected roads: Roads cross without shared node - Move node to snap to road
- Overlapping buildings: Buildings shouldn't overlap - Adjust boundaries
- Unclosed ways: Building missing final connecting node - Complete the shape
- Missing tags: Object without required tags - Add building=yes or highway= tag
- Crossing ways: Buildings overlapping incorrectly - Separate or fix boundaries
- Nodes close together: Nodes <0.5m apart - Delete unnecessary middle node
Part 3: Manual Quality Checks
Automated validation can't catch everything. Perform these manual inspections.
Manual Inspection Checklist
- ✓ All buildings are squared (press Q if not)
- ✓ Buildings align accurately with imagery
- ✓ No buildings mapped on shadows or vehicles
- ✓ Roads connect properly at intersections
- ✓ Road types match visible characteristics
- ✓ All features have appropriate tags
- ✓ No duplicate features (same building mapped twice)
- ✓ Task edges: Features continuing into adjacent tasks are complete to boundary
- ✓ Checked against project instructions for compliance
Pro Tip: Zoom out to see your entire task area at once. This reveals patterns, duplicate features, and alignment issues not visible when zoomed in.
Part 4: Preparing for Upload
Before clicking upload, prepare your changeset with proper documentation.
Upload Checklist
- Validation complete: Zero errors, warnings addressed
- Manual checks: Reviewed entire task area
- Changeset comment: Describe what you mapped
- Hashtags: Include project hashtag (e.g., #hotosm-project-12345)
- Source tag: Document imagery used (e.g., source=Bing)
- Review: Final scan before clicking Upload to OSM server
Good Changeset Comments
- ✓ 'Mapped 47 buildings in residential area #hotosm-project-12345'
- ✓ 'Added primary and residential roads #missingmaps'
- ✓ 'Digitized buildings and paths in rural village'
- ✗ 'mapping' (too vague)
- ✗ 'buildings' (no context)
- ✗ '' (empty - absolutely never!)
Part 5: Responding to Validation Feedback
When a validator invalidates your task, treat it as valuable learning, not criticism.
Handling Invalidation
- Read validator comments carefully - they're educational
- Look at validator's fixes to understand issues
- Ask questions if feedback is unclear (use project chat)
- Apply lessons learned to future tasks
- Don't be discouraged - invalidation is normal in early tasks
- Track your validation rate - aim for >85% validation rate
Important: Repeated invalidation for the same issues indicates you're not learning from feedback. If invalidated 3+ times for similar problems, pause mapping and seek additional training.
Key Takeaways
- Run JOSM validation (Shift+V) before every upload - zero errors required
- Manual quality checks catch issues automation misses
- Proper changeset comments are mandatory, not optional
- Include project hashtags in all uploads
- Treat invalidation as learning opportunity, not failure
- Aim for >85% validation rate on completed tasks
- Never upload data with validation errors