Ground survey sampling practices to construct a constrained Delaunay triangulated irregular network digital terrain model for the purpose of large scale map applications

Abstract

The validation of a digital terrain model’s (DTM) accuracy is often defined by de facto standards that do not address data acquisition sampling practices or the reconstruction methods to create the digital terrain. The testing of DTM accuracy for large scale mapping products is rarely performed because completeness in the dataset is the highest value, the raw data is collected with high precision, and the DTM is expected to function during the application process. The standardized root mean square error statistical analysis method that is used to test final DTM accuracy quality and categorically certify final DTM products is not a useful test for the data model provider. Validation metrics need to be developed for use earlier in the DTM process that focus on the data model provider’s workflow during the verification phase. These undeveloped quality metrics during the verification phase has led to model contractors defining project directives to the data model provider to include terminology based on derived model products, levels of sampling resolution, and expected raw data accuracies that do not relate to the testing standards of the DTM. This research examines the current DTM quality validation standard format and the possibility to develop relevant quality standards based on prediction by production to be applied during the data verification phase for large scale mapping products prepared by in situ, heavily biased sampling, and constructed DTMs. Ground survey methods of instrumentation and sampling are presented to identify a best practice method of repeatable survey strategy. The method of reconstruction of the raw data into a digital terrain model is that of a constrained Delaunay triangulated irregular network (CDT). The American Society of Photogrammetry and Remote Sensing ASPRS Positional Accuracy Standards for Digital Geospatial Data, 2014, for testing a DTM are reviewed and presented in a familiar unit and scale factor for ground survey providers. The only accepted verification method for DTM quality by the model contractor is “prediction by production”. By analyzing metrics from 61 large scale mapping projects collected using these recommended practices and constructed as a CDT, the criteria for analyzing the verification phase of digital terrain modeling can begin to be identified. Within the 61 DTMs studied, there is no correlation between eight scale groups using a simple resolution of survey points per DTM planar area and each scale group needs to be analyzed separately. Because the DTM quality benefits from interpolation derived from survey sampling strategy and CDT construction methods, additional factors of sampling efficiency must be developed and applied to the data of the 61 DTMs being analyzed. A DTM that is of high quality functionality can be assumed as statistically confident and metrics of mass point resolution, planimetric interval spacing, number of triangle facets and edges, and the CDT geometries can be used to test for completeness and accuracy in the raw dataset. However, DTMs with very large scales of 1”=5’ or 1”=10’ require unrealistic resolutions to pass a significant confidence level regardless of efficiency factors applied. Finally, in the era of digital models deriving computer drafted mapping products, model contractors can stop using antiquated mapping standards of hard published scales to define contour intervals, planimetric accuracies, and mass point resolutions to the data model provider and the data model provider can certify to applicable quality categories.