Center for Satellite Applications and Research (STAR) Review 09 – 11 March 2010 Altimetric Bathymetry Model Excels Global bathymetry model combines depths predicted from satellite gravity with ship measurements Histogram at left shows that depths predicted from altimetry match measured depths well. There is no systematic bias or scaling of errors with depth. Can we improve on this? Science Challenges: Develop systematic method to assess errors Document heritage/legacy of multiple versions Improve bathymetry models & algorithms Next Steps: Make new estimate, improving algorithm with knowledge gained in this study. Transition Path: Disseminate results to end users through NGA/Navy MOA, GEBCO and Google. Error Analysis of the Altimetric Bathymetry Models used by GEBCO and Google Earth K. M. Marks and W. H. F. Smith (GOVERNMENT PRINCIPAL INVESTIGATOR) NOAA/NESDIS/STAR Requirement: Primary NOAA Goal- Commerce & Transportation: -Explore, develop, and transition emerging technologies and techniques to enhance marine navigational safety and efficiency -Research focuses on developing and evaluating a wide range of state-of-the-art hydrographic and ocean mapping technologies and applications Science: Can error analyses lead to improved altimetric bathymetry models? Benefit: Navy Submarine Navigation Improvement Program [SNIP] International [IHO & IOC] ocean mapping [GEBCO] NOAA Weather & Water Info Goal [marine spatial planning; Outreach & Education, etc.] NOAA Understanding Climate Goal [bottom shape and roughness steer currents & cause mixing] NOAA Ecosystems Goal [bathymetry features (seamounts) create habitat] Google Earth Abstract. We analyze errors in global bathymetry models that combine satellite altimetry with in situ soundings and shorelines to estimate depths. Various versions of these models have been incorporated into Google Earth and the General Bathymetric Chart of the Ocean (GEBCO). We seek to understand how software changes, changes in in situ data, and other effects have propagated error into these products. We use JAMSTEC (Japan Agency for Marine Earth Science and Technology) multibeam surveys not previously incorporated into the models as “ground truth”. We find two problems. One is a bias in the scaling from gravity in milliGals to topography in meters that affected the km wavelength band in versions 9 thru 11. We also find that regionally averaged (>160 km wavelength) depths were accumulating error over successive versions of the model. These problems were not in version 8.2 but crept in subsequently, and have been mostly mitigated in version Errors in version 12.1 show no systematic correlation with depth. However, it appears that some special features of the algorithm used only in version 8.2 are superior to those in the algorithm used subsequently, and we are investigating how these features can be revived in the future. Bathymetry Model Table Our altimetric bathymetry products are used worldwide in Google Earth, GEBCO, and throughout the scientific community. Our effort produces the only altimetric bathymetry model, all others are based on it We are tracking the downstream products derived from each model version. Smooth Seafloor Rough Seafloor Regional Errors Error Analysis leads to Model Improvements As the bathymetry model evolved over the years, new depth observations were added and prediction algorithms were improved. Altimetry research at STAR has reduced the RMS error in altimetric gravity from 4.1 to 2.0 mGal. We found a gravity->topography scaling error in versions 9 through 11, and this was fixed in v12, bringing RMS depth error down to 46 m from 80 m. We find that an older model, version 8.2, had the smallest errors and did the best job in smooth areas. It used a non-linear thresholding approach. We will incorporate features of the 8.2 algorithm in future versions. Models do well over Rough Seafloor Satellite gravity anomalies over rough seafloor are correctly ingested into bathymetry model Older version 8.2 has smallest errors, filtering correction discovered by our analysis improves latest version 12.1 Future versions can be further improved Regional Errors Reduced In version 11 (left), the impact of new data (red tracks) was allowed to diffuse over long distances, while in version 12 (right), it was not. The long-wavelength component of version 12 is based on S2004, which traces heritage to v8.2. This reduced regional-scale errors in version 12.1 The next generation model will employ better data diffusion. Selected Publications K. M. Marks, W. H. F. Smith, and D. T. Sandwell, Error Analysis of the altimetric bathymetry models used by GEBCO and Google Earth, manuscript in preparation, to be submitted to Marine Geophysical Researches. K. M. Marks and W. H. F. Smith, An uncertainty model for deep ocean single beam and multibeam echo sounder data, Marine Geophysical Researches, Vol. 29, No. 4, , Dec Published online: 23 January 2009, DOI: /s y. K. M. Marks and W. H. F. Smith, Some remarks on resolving seamounts in satellite gravity, Geophysical Research Letters, Vol. 34, 2007; L03307, doi: /2006GL K. M. Marks and W. H. F. Smith, An evaluation of publicly available global bathymetry grids, Marine Geophysical Researches, Vol. 27, No. 1, 19-34, K. M. Marks and W. H. F. Smith, 2500 m Isobath from satellite bathymetry: Accuracy in light of IHO S-44 standards, The International Hydrographic Review, Vol. 6, No. 2, , August Sandwell DT, Smith WHF (2005) Retracking ERS-1 altimeter waveforms for optimal gravity field recovery, Geophys J Int. 163: doi: /j X x Smith WHF, Sandwell DT (1997), Global sea floor topography from satellite altimetry and ship depth soundings. Science 277: doi: /science Smith WHF, Sandwell DT (1994) Bathymetric prediction from dense satellite altimetry and sparse shipboard bathymetry, J Geophys Res. 99(B11): doi: JB00988 Center for Satellite Applications and Research (STAR) Review March 2010