Validacion en el sur de Europa de modelos digitales del terreno globales y productos derivados

Validación en el sur de Europa de modelos digitales del terreno globales y sus productos derivados

Validation in southern Europe of global digital elevation models and their derived products

ABTRACT

Digital elevation models are essential tools used in geosciences; there exist a wide range of products from satellites that cover practically the entire earth surface which are freely distributed. All DEMs have errors and their use is conditioned by the accuracy that is needed. Hence, it is essential to develop studies that validate and limit their errors.

The present study has been focused on the altimetric errors determinated by statistical methods of the SRTM models (C-BAND, with 90 m pixel) and ASTER (GDEM, with 30 m pixel) in an area of 10,611 Km2, located in the northern area of the Iberian Peninsula (Southern Europe). To evaluate the errors, 90 geodesic points with a 10cm planimetric accuracy and an altimetric accuracy oscillating between 5 and 20cm have been used, furthermore a DEMs of 5m of high accuracy has also been used to compare the models derived from DEMs such as slopes, aspect, drainage basins and flow lines.

In the study area, the SRTM model has lower altimetric errors than ASTER, with average errors of -6.001m and -13.538 m and SDs of 27.16m and 29.90m, respectively. The relationship between the errors and the elevation or slope has been observed.

The work also presents a compilation and comparison of the errors of these models in different earth areas.

KEYWORDS: DEMs evaluations, SRTM, ASTER GDEM, elevation accuracy

INTRODUCCIÓN

INTRODUCTION

En cualquier ámbito de la geociencia los modelos digitales de elevación (DEMs) imprescindible tanto para realizar análisis de la superficie de la Tierra en multitud de ámbitos de la ciencia, desde estudios climatológicos a análisis de cuencas, estudios de erosión, relieve, riesgo, etc, También son adecuados para modelizar zonas que no corresponden exclusivamente a la superficiales  como las asociadas a la actividad tectónica (Montgomery et al., 2001), etc.

In geoscience, digital elevation models (DEMs) are essential tools to for carry out analyzes of the Earth's surface in many areas, from climatological studies to basin analysis, erosion studies, relief, risk, etc. They are also suitable for modeling areas that do not correspond exclusively to surface areas such as those associated with tectonic activity (Montgomery et al., 2001), etc.

La disponibilidad de modelos precisos provenientes de la observación con técnicas GPS, LIDAR, fotogrametría, etc está más restringida a países de mayor desarrollo tecnológico y económico. Es por ello que los productos provenientes de misiones con satélites y con cobertura global son cada vez más utilizados. Normalmente éstos son Modelos Digitales de Superficie (DSM) y las diferencias que existen entre ellos se deben entre otras cosas a las diferencias en cuanto a la técnica de adquisición de datos, de los algoritmos utilizados para la creación de los modelos y características de cada región. Todo ello origina variedad de errores en los productos finales (Smith and Sandwell, 2003; Fisher and Tate, 2006; Nuth and Kääb, 2011) que deben ser valorados en las distintas regiones de la Tierra.

The availability of precise models made by observation with GPS techniques, LIDAR, photogrammetry, etc. is restricted to countries with greater technological and economic development. That is why the products coming from missions with satellites and with global coverage are increasingly used. Normally these are Digital Surface Models (DSM) and the differences that exist between them are due, among other things, to the differences in the technique of data acquisition, the algorithms used to create the models and characteristics of each region. All this leads to a variety of errors in the final products (Smith and Sandwell, 2003, Fisher and Tate, 2006, Nuth and Kääb, 2011) that should be valued in the different regions of the Earth.

El error de los modelos es debido a la incertidumbre asociada a tres factores, a saber: parámetros del sistema durante la adquisición de datos (posición y rango de inclinación de la atena, fase, etc), procesamiento de los datos en bruto y características del territorio como climatología, vegetación, orografía, etc. (Smith and Sandwell, 2003; Fisher and Tate, 2006; Nuth and Kääb, 2011). Otro aspecto importante a considerar es el datum vertical al que se refiere cada modelo. Como el número de variables que intervienen en el error final de un modelo es muy elevado el estudio de la propagación de errores se hace prácticamente inviable. Por esta razón el estudio la validación de los modelos se realiza por medio de una comparación con datos de mayor precisión.

The model error is due to the uncertainty associated to three factors: parameters of the system during the acquisition of data (position and tilt of the antenna, phase, etc), processing of the raw data and characteristics of the territory such as climatology, vegetation, orography, etc. (Smith and Sandwell, 2003, Fisher and Tate, 2006, Nuth and Kääb, 2011). Another important aspect to consider is the vertical datum to which each model is refered. As the number of variables that involve in the final model error is very high, the study of the propagation of errors becomes practically unviable. That is why the validation study of the models is carried out through a comparison with more accurate data.

Las especificaciones para el modelo SRTM y ASTER proporcionan unas precisiones de +16m y + 20m con un nivel de confianza del 95%, respectivamente. En los últimos cinco años se han incrementado trabajos regionales cuyo objetivo es validar los DEM globales, tanto distribuidos gratuitamente como de pago. Con la evaluación del error en cota, se ha podido observar que los resultados varían notablemente de unas zonas a otras, así el error máximo para ambos modelos se localizó en China donde ascienden a 26m y 23m para los modelos ASTER y SRTM, respectivamente (P.Li et al, 2013)

The specifications for the SRTM and ASTER model provide accuracies of + 16m and + 20m with a confidence level of 95%, respectively. In the last five years, regional works have increased to validate global DEMs, both free and paid. With the evaluation of the altimetric error, it has been observed that the results vary considerably from one area to another. The maximum error for both models was located in China where they amount to 26 m and 23 m for the ASTER and SRTM models, respectively (P .Li et al, 2013)

...