Spatial pattern alterations from JPEG2000 lossy compression of remote sensing images: Massive variogram analysis in high performance computing

Pesquer L., Pons X., Cortes A., Serral I. (2013) Spatial pattern alterations from JPEG2000 lossy compression of remote sensing images: Massive variogram analysis in high performance computing. Journal of Applied Remote Sensing. 7: 0-0.
Enllaç
Doi: 10.1117/1.JRS.7.073595

Resum:

We evaluate the implications of JPEG2000 lossy compression of remote sensing images for spatial analytical purposes. The main issue is to identify which cases and conditions in geostatistical studies are suitable for using lossy compressed images. For these purposes, an extensive test using Landsat, compact airborne spectrographic imager (CASI), and moderate resolution imaging spectroradiometer (MODIS) image series has been analyzed, through applying and comparing two-dimensional and three-dimensional (spectral and time domains) compression methods with a wide range of compression ratios for several dates, different landscape regions, and spectral bands. Due to the massive test bed and consequently to the high time consuming executions, a parallel solution was specifically developed. Variogram analyses showed that all the compression ratios maintain the variogram shapes, but high compression ratios (>20:1) degrade the spatial patterns of the remote sensing images. These alterations are lower for the three-dimensional compression method, which was a considerable improvement (25%) on the two-dimensional method for large three-dimensional series (CASI, MODIS). However, the two methods behave similarly in the Landsat case. Finally, the parallel solution in a distributed environment demonstrates that high performance computing offers a suitable scientific platform for highly demanding time execution applications, such as geostatistical analyses of remote sensing images. © The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License.

Llegeix més

Rubric-Q: Adding quality-related elements to the GEOSS clearinghouse datasets

Zabala A., Riverola A., Serral I., Diaz P., Lush V., Maso J., Pons X., Habermann T. (2013) Rubric-Q: Adding quality-related elements to the GEOSS clearinghouse datasets. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 6: 1676-1687.
Enllaç
Doi: 10.1109/JSTARS.2013.2259580

Resum:

Geospatial data have become a crucial input for the scientific community for understanding the environment and developing environmental management policies. The Global Earth Observation System of Systems (GEOSS) Clearinghouse is a catalogue and search engine that provides access to the Earth Observation metadata. However, metadata are often not easily understood by users, especially when presented in ISO XML encoding. Data quality included in the metadata is basic for users to select datasets suitable for them. This work aims to help users to understand the quality information held in metadata records and to provide the results to geospatial users in an understandable and comparable way. Thus, we have developed an enhanced tool (Rubric-Q) for visually assessing the metadata quality information and quantifying the degree of metadata population. Rubric-Q is an extension of a previous NOAA Rubric tool used as a metadata training and improvement instrument. The paper also presents a thorough assessment of the quality information by applying the Rubric-Q to all dataset metadata records available in the GEOSS Clearinghouse. The results reveal that just 8.7% of the datasets have some quality element described in the metadata, 63.4% have some lineage element documented, and merely 1.2% has some usage element described. © 2013 IEEE.

Llegeix més