Post-procesado de información LIDAR mediante técnicas geoestadísticas

  1. Delgado García, J. 1
  2. Pérez García, J. L. 1
  3. Gómez Molina, A. 2
  4. Gómez Vidal, M. D. 1
  5. Soares, A. 3
  1. 1 Universidad de Jaén
    info

    Universidad de Jaén

    Jaén, España

    ROR https://ror.org/0122p5f64

  2. 2 Universidad Politécnica de Madrid
    info

    Universidad Politécnica de Madrid

    Madrid, España

    ROR https://ror.org/03n6nwv02

  3. 3 Universidade de Lisboa
    info

    Universidade de Lisboa

    Lisboa, Portugal

    ROR https://ror.org/01c27hj86

Book:
El acceso a la información espacial y las nuevas tecnologías geográficas
  1. M.T. Camacho Olmedo (ed. lit.)
  2. J.A. Cañete Pérez (ed. lit.)
  3. J.J. Lara Valle (ed. lit.)

Publisher: Universidad de Granada

ISBN: 84-338-3944-6

Year of publication: 2006

Pages: 105-118

Congress: Congreso Nacional de Tecnologías de la Información Geográfica (12. 2006. Granada)

Type: Conference paper

Abstract

Aerial LIDAR sensors are become actually a real alternative for data capture for digital elevation model (DEM) generation. The information provided by the sensor is composed by a very high dense XYZ point cloud (points can be located into the ground or in different objects that are situated on the ground –trees, vehicles …-) and the eco reflectivity of the laser pulse. These data have two main characteristics: in the first place, the points do not present a regular distribution and, secondly, the data includes points that present measuring errors -or they are located on elements without any interest-. These problems define the main objectives of this work that are discussed from a geostatistical point of view. Geostatistics presents the important advantage to include in modelling process parameters that they are derived of the own experimental information. The data geostatistical postprocessing begins with the spatial variability functions calculation. These functions will be basic in the following processes. Due to the terrain characteristics is usually to obtain non-stationary variogram so the proposed methodology is based in the drift estimation using a polynomial regression and the residual calculation (that have a stationary behaviour). Once the spatial variability functions are available, these functions will be used in a cross validation process that it is made that combine the terrain height and intensity information, providing a error anomalous data classification that can be used in order to debugging the experimental data. Once the data are debugged (and data with measuring errors located) it is possible to approach the digital elevation model generation that obtain the final model according a regular grid. This process has been made using a block residual kriging methodology that provides the most representative value of the block (that will be defined by the grid spacing) from the punctual measured residual data. The final model will be obtained adding to the residual estimates the drift values.