Geospatial QA/QC Workbench
QA/QC means checking a delivered dataset against its specification before anyone uses it. Pick the product you are checking. Each step shows what to check, how to do it in ArcGIS Pro, QGIS and code, the ISO 19157-1 quality element it tests, and the errors it catches.
Typical errors in this product
Click an error to see what it is, why it matters and which step catches it.
How many items to inspect?
For manual review you inspect a sample. Enter the number of items in the lot, for example map tiles, 1 km grid cells, buildings or DEM tiles.
Sampling plan
Single sampling, normal inspection (ISO 2859-1 Table 1 and Table 2-A). Pick the sample at random, for example with ArcGIS Pro Subset Features or QGIS Random selection. Confirm plans for contractual work against the standard.
RMSE, CE90 and LE90 from check points
One point per line: dx, dy or dx, dy, dz in metres (dataset minus check point). Export them from ArcGIS Pro Extract Values to Points or QGIS Sample raster values and paste here.
Results
CE90 ≈ 1.5175 × RMSEr, LE90 ≈ 1.6449 × RMSEz, assuming normal errors with no bias and similar RMSE in x and y. Use at least 20 independent, well-distributed check points.
Land cover confusion matrix
Rows: class on the map. Columns: class in the reference (what the checker saw). Enter the number of sample points in each cell.
Accuracy
User's accuracy: of the points mapped as a class, how many really are that class (reliability for map users). Producer's accuracy: of the real class, how much the map captured. In ArcGIS Pro, Compute Confusion Matrix produces the same table; in QGIS, the AcATaMa plugin.
Every check in the workflow tests one of the six ISO 19157-1 quality elements. Open a sub-element for a plain definition and an example from the four products.
Standards used
| Standard | Covers | Used for |
|---|---|---|
| ISO 19157-1 | Data quality: elements, measures, evaluation and reporting | Every check and the quality report |
| ISO 19131 | Data product specifications | The requirements each product is checked against |
| ISO 19110 | Feature cataloguing | Feature types, attributes and allowed values for vector data |
| ISO 19115-1 | Metadata | Checking delivered metadata and recording quality results |
| ISO 19158 | Quality assurance of data supply | Assessing a producer's process, not just one delivery |
| ISO 2859-1 | Sampling by attributes (AQL) | Sample sizes and accept/reject numbers for manual review |
Tolerances marked "example" are illustrations. Real values come from each product's specification. This page describes general practice, not the procedure of any particular organisation.
Amin Adnan Gasim · coffeemap.dev