Application of Geostatistical Conflation Techniques to Improve the Accuracy of Digital Elevation Models Carlos A. Felgueiras, Jussara O. Ortiz 1 , Eduardo C. G. Camargo 1 1 Divisão de Processamento de Imagens - Instituto Nacional de Pesquisas Espaciais (INPE) - Caixa Postal 515 – 12201-970 – São José dos Campos – SP – Brazil {carlos,jussara,eduardo}@dpi.inpe.br Abstract. This short paper describes and analyzes the results of a methodology that allows to conflate existing Digital Elevation Models with a sample set of elevation points in order to obtain more accurate results on modeling elevation information. The set of elevation points has higher vertical accuracy than the DEM and it is used the geostatistical procedure, known as kriging with an external drift, to perform the conflation. An initial case study is presented integrating a Shuttle Radar Topography Mission - SRTM - data and a sample set of elevation points obtained from a region of Campinas, city of São Paulo in Brazil. Others procedures of estimations and simulations will be considered in the future to explore the potential of the conflation techniques using geostatistics. 1. Introduction Digital Elevation Models - DEMs -, and their derivative products, are very important information used as input for spatial models performed in Geographical Information Systems - GIS - environment [Burrough 1987]. From a DEM it is possible to derive slope and aspect maps, drainage networks, contour lines, profile and volume calculations, etc. Nowadays it is possible to find DEM information for free, without financial costs, of almost any region of the earth surface. Unfortunately, the vertical (the heights) accuracy of these free DEMs are not appropriated for some spatial models. On the other hand elevation information of the earth surface can also be obtained in a set of spatial locations, 3D points, sampled in a geographical region of interest. These samples can be collected with very high vertical accuracy using Global Positioning System - GPS - equipments, for example. Geostatistical tools has been used successfully to analyze and to model environmental attributes represented as a set of sample points of geophysical and geochemical indices, concentrations of soil elements, elevations, temperatures, etc. [Isaaks and Srivastava 1989, Goovaerts 1997]. Therefore, geostatistics can be applied to a sample set of elevation points, hereinafter referred to as "sample points", in order to create DEMs using estimations and simulations procedures. Also the geostatistical procedures allow performing conflations by integrating different sources of environmental attributes [Hengl et al 2008, Karkee et al 2008]. In the case of elevations there are kriging and simulation procedures that can be used to conflate existing DEMs