[Top] [Prev] [Next] [Bottom]

3.3 Attribute Accuracy

Accuracy assessment for measures on a continuous scale must be performed using procedures similar to those used for positional accuracy (providing a numerical estimate of expected discrepancies). The report of a test of attribute accuracy must include the date of the test and the dates of the materials used. In the case of different dates, the report must describe the rates of change in the phenomena classified. Spatial variations in attribute accuracy may be reported in a quality overlay.

Accuracy tests for categorical attributes may be performed by one of the following methods. All methods must make reference to map scale in interpreting classifications.

3.3.1 Deductive Estimate.

Any estimate, even a guess based on experience, is permitted. The basis for the deduction must be explained. Statements such as "good" or "poor" should be explained in as quantitative a manner as possible.

3.3.2 Tests Based on Independent Samples.

A misclassification matrix must be reported as counts of sample units crosstabulated by the categories of the sample and of the tested material. The sampling procedure and the location of sample units must be described.

3.3.3 Tests Based on Polygon Overlay.

A misclassification matrix must be reported as areas. The relationship between the two maps must be explained; as far as possible, the two sources should be independent and one should have higher accuracy.