Bellos, Dimitrios Andreas (2015) A case study at Lake Constance for water depth determination from hyperspectral data. Masterarbeit, TU München.
Dieses Archiv kann nicht den Volltext zur Verfügung stellen.
Kurzfassung
Hyperspectral remote sensing is an established technology for bathymetric mapping of shallow water areas. Inland lakes are highly endangered due to sedimentation and eutrophication, thus systematic research is essential. This study focuses on bathymetry retrieval based on a radiative-transfer model for the German lake Constance. Above water surface, underwater, on boat sample measurements and Sonar data were collected during a ship campaign at the areas of Langenargen and Rhine by using RAMSES sensors and a Tritech altimeter. An airborne campaign provided hyperspectral data using a recently calibrated at DLR, HySpex VNIR-1600 sensor. Two sediment and one macrophytes bottom albedos were derived from on boat sample measurements. The underwater measurements were synchronized with the Sonar data. Bottom reflectance was determined from above and underwater downwelling irradiance measurements by using inverse modelling and compared with remote sensing reflectance spectra derived from samples. Due to the unreliability of bottom reflectance spectra, bathymetry was derived from hyperspectral imagery through inverse modelling by using the representative sample-derived bottom albedos for each test site. From the resulted fit maps of fAi, scale factors were derived and applied on the bottom albedos for the correction of intensity errors. For the validation of the derived water depths, a combined Lidar-Multibeam echosounder dataset was used. Water depth could be determined up to 6 m for the Langenargen area and 5 - 6 m for the Rhine area with the RMSE ranging between 39 and 79 cm among the test sites. Two stations of the Rhine area revealed depths up to 46 m, therefore the inversion technique failed to determine bathymetry. At the Langenargen area, accuracies of 28 - 73 cm from 0 - 4 m and 44 - 105 cm from 4 - 6 m can be achieved if the systematic errors could be eliminated. Regarding the Rhine area, accuracies of 19 - 40 cm from 0 - 3 m and 89 cm from 3 - 4 m seem possible. Large statistical error was revealed due to data quality and inversion ambiguities, while errors in the derived bottom albedos were also expected.
elib-URL des Eintrags: | https://elib.dlr.de/98851/ | ||||||||
---|---|---|---|---|---|---|---|---|---|
Dokumentart: | Hochschulschrift (Masterarbeit) | ||||||||
Titel: | A case study at Lake Constance for water depth determination from hyperspectral data | ||||||||
Autoren: |
| ||||||||
Datum: | 1 September 2015 | ||||||||
Referierte Publikation: | Nein | ||||||||
Open Access: | Nein | ||||||||
Seitenanzahl: | 87 | ||||||||
Status: | veröffentlicht | ||||||||
Stichwörter: | Lake Constance, bottom re-Lake Constance, hyperspectral, RAMSES, inverse modelling, bottom reflectance, bathymetry, HySpex | ||||||||
Institution: | TU München | ||||||||
Abteilung: | School of Forest Science and Resource Management | ||||||||
HGF - Forschungsbereich: | Luftfahrt, Raumfahrt und Verkehr | ||||||||
HGF - Programm: | Raumfahrt | ||||||||
HGF - Programmthema: | Erdbeobachtung | ||||||||
DLR - Schwerpunkt: | Raumfahrt | ||||||||
DLR - Forschungsgebiet: | R EO - Erdbeobachtung | ||||||||
DLR - Teilgebiet (Projekt, Vorhaben): | R - Vorhaben Spektrometrische Verfahren und Konzepte der Fernerkundung (alt) | ||||||||
Standort: | Oberpfaffenhofen | ||||||||
Institute & Einrichtungen: | Institut für Methodik der Fernerkundung > Experimentelle Verfahren | ||||||||
Hinterlegt von: | Gege, Dr.rer.nat. Peter | ||||||||
Hinterlegt am: | 29 Okt 2015 13:06 | ||||||||
Letzte Änderung: | 22 Dez 2016 12:41 |
Nur für Mitarbeiter des Archivs: Kontrollseite des Eintrags