Automatic extraction of shorelines from Landsat TM and ETM+ multi-temporal images with subpixel precision
A high precision geometric method for automated shoreline detection from Landsat TM and ETM+ imagery is presented. The methodology is based on the application of an algorithm that ensures accurate image geometric registration and the use of a new algorithm for sub-pixel shoreline extraction, both at...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2012 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/58901 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/58901 |
| Access Level: | acceso abierto |
| Palabra clave: | Beach management Coastal processes Landsat images Shoreline subpixel detection Automatic extraction Coastal process Correction models Descriptors Error assessment Geometric method High precision High-gain High-resolution techniques Land-cover types Landsat imagery LANDSAT TM Maximum error Mean errors Multi-temporal image Sub pixels Subpixel detection Subpixel precision Algorithms Normal distribution Errors Accuracy assessment Beach nourishment Error correction Geomorphological response Image resolution Land cover Landsat thematic mapper Pixel Precision Probability Reflectance Satellite imagery Shoreline INGENIERIA CARTOGRAFICA, GEODESIA Y FOTOGRAMETRIA |
| Sumario: | A high precision geometric method for automated shoreline detection from Landsat TM and ETM+ imagery is presented. The methodology is based on the application of an algorithm that ensures accurate image geometric registration and the use of a new algorithm for sub-pixel shoreline extraction, both at the sub-pixel level. The analysis of the initial errors shows the influence that differences in reflectance of land cover types have over shoreline detection, allowing us to create a model to substantially reduce these errors. Three correction models were defined according to the type of gain used in the acquisition of the original Landsat images. Error assessment tests were applied on three artificially stabilised coastal segments that have a constant and well-defined land-water boundary. A testing set of 45 images (28 TM, 10 ETM high-gain and 7 ETM low-gain) was used. The mean error obtained in shoreline location ranges from 1.22 to 1.63. m, and the RMSE from 4.69 to 5.47. m. Since the errors follow a normal distribution, then the maximum error at a given probability can be estimated. The results confirm that the use of Landsat imagery for detection of instantaneous coastlines yields accuracy comparable to high-resolution techniques, showing the potential of Landsat TM and ETM images in those applications where the instantaneous lines are a good geomorphological descriptor. © 2012 Elsevier Inc. |
|---|