## Resolution Determination through Level of Aggregation Analysis

Title | Resolution Determination through Level of Aggregation Analysis |

Publication Type | Conference Proceedings |

Year | 2019 |

Authors | Calder, BR |

Conference Name | U.S. Hydrographic Conference (US HYDRO) |

Conference Dates | March 19-21 |

Publisher | The Hydrographic Society of America |

Conference Location | Biloxi, MS |

Keywords | Bathymetric Modeling, Bathymetric Processing, Digital Elevation Model, hydrography, Resolution Detemination, seabed 2030 |

In order to accommodate significantly varying depths within a survey area, and the consequent data density changes, variable-resolution depth modeling technologies are now being deployed. A core question for such technologies is how to determine the appropriate spatially-varying resolution at which to estimate or model the seafloor in a computationally efficient manner. Current methods include conversion from roughly-estimated depth or data density to resolution, or spatially-recursive sub-division (typically via a quadtree) with an appropriate similarity metric, typically working on a coarse-to-fine basis (i.e., starting with the whole survey area, and working to finer scales as the resolution is determined). All of these methods require a preliminary pass through the source data, and make various assumptions about its structure. Computational efficiency and level of assumptions are therefore important for implementation.
As an alternative to these techniques, this paper describes a fine-to-coarse method based on a "level of aggregation" metric which makes no assumptions about the structure of the data, allowing it to be used equally on acoustic, lidar, or random point data. This method is methodologically direct and simple, data adaptive, readily parallelized, and automatically determines both the rate at which resolution is changing and the final resolution within this structure.
The method is illustrated in the context of processing Riegl VQ-880-G high-resolution shallow lidar data, and mixed-sensor acoustic data from a NOAA survey, with particular attention to parallel and distributed implementation. A direct corollary of estimating resolution is the ability to assess whether a given data set can meet survey specifications, which effectively provides a measure of how "surveyed" an area is. This is illustrated on an archive collection of random data from the U.S. Atlantic Margin in the context of Seabed 2030. |