Google+

ASPRS Standards

By: David Lundeen, CTO

In November 2014, the ASPRS released updated standards for geospatial data accuracy. The ASPRS Positional Accuracy Standards for Digital Geospatial Data replaces the ASPRS Accuracy Standards for Large-Scale Maps standard that was released in 1990. The new standard is more comprehensive and is more suitable for modern mapping capabilities.

The 2014 standard is fundamentally different than the 1990 standard. Instead of being based on map resolution, the accuracy requirements for each class of map are based on RMSE values for various measurements.

The chart below[1], taken from the 1990 guidelines, summarizes the horizontal accuracy requirements for a Class 1 map.

 

When the 1990 standard was released, scale was the primary indicator of map resolution. A number of factors contributed to the common use of scale. Map scale is based on film camera functionality and design. Aerial film cameras typically used the same size film (9 x 9 inches) and had the same focal length (6 inches)[2]. Traditional image processing standards and flying altitudes, combined with standard camera design, made it natural to use scale to indicate map resolution. Map accuracy standards were based on scale because scale was a universally important map attribute.

Most modern remote sensing projects use Ground Sampling Distance (GSD) to indicate map resolution.  GSD is based on the sensor focal length, sensor pixel size, and height above ground. Modern digital mapping technology makes it practical to use GSD as an indicator of map resolution.

Unlike the previous standard however, the 2014 standard does not base accuracy requirements on map resolution. Instead, the requirements are based on RMSE thresholds. The 2014 standard documentation includes an appendix with charts that relate GSD and scale to the new standards, but the accuracy requirements are independent of map resolution.

The new standard defines accuracy standards for a number of components that are used in producing geospatial data. Among other things, it includes accuracy requirements for aerial triangulation and seamlines, as well as the more typical horizontal and vertical accuracy requirements. Currently there are no accuracy standards applicable to oblique images.

Calculation of horizontal accuracy requirements, defined by RMSE thresholds, is summarized in the table[3] below.

 

The RMSEx and RMSEy values must be less than or equal to the RMSE that defines the class. The radial RMSE must be less than or equal to the RMSE multiplied by 1.414. A new standard, specifying seamline accuracy, requires the seamlines to be mismatched by no more than the RMSE multiplied by 2.

For example, if the horizontal accuracy class is 10cm, the RMSE for x and y values must be no more than 10. The radial RMSE must be no more than 14.14, and seamline mismatches must be no more than 20 cm. Use of an RMSE threshold to define accuracy classes makes it easy to determine accuracy requirements for any class.

Vertical accuracy of elevation data is also categorized by RMSE. Vertical accuracy class names are based directly on the required RMSE.  For example, a 6 cm vertical accuracy class would require a product with an RMSEz of no greater than 6 cm. The accuracy assessment is based on the type of terrain. Vertical accuracy of areas of non-vegetated terrain would be assessed as RMSEz while vertical accuracy of areas of vegetated terrain would be assessed as a 95th percentile value.

Horizontal accuracy requirements for elevation data depend on how the data is acquired.  For elevation data acquired through photogrammetric processes, the horizontal accuracy requirements are dictated by the horizontal accuracy requirements of the final product. For Lidar acquired elevation data, a calculation that takes into account altitude, GNSS error, IMU error, and other factors is used to determine the class requirements.

The new standard includes guidelines for the accuracy of aerial triangulation. In general, the AT RMSEx and RMSEy must be less than or equal to ½ the map RMSEx and RMSEy. The AT RMSEz value can be twice the AT RMSEx or RMSEy values (equal to the map RMSEx and RMSEy values).

The new standard documentation includes recommendations for the use of each class. Table B5 on page A14 identifies Class I as suitable for high accuracy work, Class II as suitable for standard mapping work, and Class III for products that don’t require a high degree of accuracy. The recommendations will make it easier for data suppliers and customers to agree on project specifications.

The addition of AT and seamline specifications are valuable additions. AT accuracy specifications are good indicators of the quality of intermediate processing steps. That makes it easier to identify problems before the product is complete. Seamline specifications are valuable because now there is a shared standard instead of reliance on individual assessment.

The 2014 standards are more comprehensive than the previous standards and remove subjectivity from several important areas of map production. Basing the standards on RMSE thresholds instead of map resolution simplifies defining project requirements and removes assumptions about mapping technology. The addition of aerial triangulation accuracy specifications makes it easier to assess product quality in the intermediate steps. Seamline specifications are useful because now there is a shared standard instead of reliance on arbitrary standards. The 2014 standards will bring significant benefits to the GIS community.