Evaluation of Geostatistical Procedures for Accuracy Improvement of Digital Elevation Models

• Grant number: 15/24676-9
• Support Opportunities: Regular Research Grants
• Duration: March 01, 2016 - February 28, 2018
• Field of knowledge: Physical Sciences and Mathematics - Geosciences
• Principal Investigator: Carlos Alberto Felgueiras
• Grantee: Carlos Alberto Felgueiras
• Host Institution: Instituto Nacional de Pesquisas Espaciais (INPE). Ministério da Ciência, Tecnologia e Inovações (Brasil). São José dos Campos , SP, Brazil
• Associated researchers: Antonio Miguel Vieira Monteiro ; Camilo Daleles Rennó ; Eduardo Celso Gerbi Camargo ; João Ricardo de Freitas Oliveira ; Jussara de Oliveira Ortiz ; LAERCIO MASSARU NAMIKAWA ; Sergio Rosim ; Sidnei João Siqueira Sant'Anna

Abstract

This research aims to evaluate, investigate, develop and validate, computational methods and methodologies, based on geostatistical procedures, to improve the accuracy of the altimetry attribute of Digital Elevation Models (DEMs). The DEMs are important components in different applications as, for example, in the elaboration of susceptibility to landslide maps, in the extraction of drainage networks, in the construction of pipelines and roads, in watershed delimitations, and several others. To address this demand of researches is essential that the DEMs have a minimum accuracy so that the resulting products can guide better intervention activities and decision-makings. In this sense, the generation of DEMs requires the use of more accurate data and appropriate computational techniques. To meet these requirements this research proposes the use of linear and nonlinear geostatistical procedures. Geostatistics can be seen as a tool that comprises a set of stochastic methods which enable variable predictions that are distributed over the geographical area, simulations for construction of new scenarios and generations of uncertainty maps related to the predictions and simulations, that are relevant to qualify the obtained results. Numerous researches in Brazil often make use of altimetry data from free sources but do not employ adequate modeling procedures using these data. In general, these data are generated from satellite image processing and subsequently made available via the Internet. Although these data are dense and with good spatial distribution, they are poor in the accuracy issue. For many applications the accuracy of this information may be unacceptable. A way to mitigate this problem is to aggregate these altimetry data in the database and use them along with other information, that come from reliable sources and better quality, in the modeling process. Generally, high accuracy and precision altimetry data are derived from the data collection works at points located inside the investigated region. This is a costly and necessary task, involving visits and field works and making use of high-tech equipment support. In this context, this work aims to integrate different sources of elevation data through geostatistical methods to obtain better quality altimetry information. Specific softwares will be explored such as R, Gstat, GSLIB, SPRING, etc., and it will also be developed specific functions according to the needs of the researches. Validations are made using high-accuracy information, obtained from global positioning equipments in field works and from remote sensing data. Also validations will be performed with data resulting from characteristic extractions of DEMs as, for example, drainage networks, watershed delimitations, contour maps, slope maps, etc .. The methods developed in this research work will be applied to case studies in Southeast geographical regions of Brazil. (AU)