180 likes | 338 Views
In Situ Database and Sensor Characterisation Results Sam Lavender, UoP. Inputs from the GlobCOLOUR team Special thanks to Yaswant Pradhan, Julien Demaria, Gilbert Barrot & David Antoine. Characterisation review. In Situ Data / Database Level-2 Characterisation: In situ
E N D
In Situ Database and Sensor Characterisation Results Sam Lavender, UoP Inputs from the GlobCOLOUR team Special thanks to Yaswant Pradhan, Julien Demaria, Gilbert Barrot & David Antoine.
Characterisation review • In Situ Data / Database • Level-2 Characterisation: • In situ • Cross Characterisation • Level-3 Characterisation • Overall Conclusions
Types ofIn Situ Data • Chlorophyll (HPLC and fluorometric) • Diffuse attenuation coefficient at ~490nm (not shown in this presentation) • Fully normalised water-leaving radiance primarily at 412, 443, 490, 510, 531, 555, 620, 670 nm • Computed when Lw, Ed, Es, Chl, and solar and viewing geometry available.
Level-2 characterisation data sets BOUSSOLE NOMAD SeaBASS 2002 onwards
L2 (M)LAC In-situ meta Preparation/Generation L3-DDS Generator Extraction Statistics/Result In-situ data L3 DDS Match-up Result L3-DDS Reader GC in-situ Reader Exclude NO match-up Timediff <24 hrs Y Timediff 12 hrs Flagged pixels ( e.g. NO_MEAS) rejected Num Pix >= 13 CV = 0.15 DDS Match-Up N N Locationdiff <=0.02° Y NO match-up N Extract 5x5 kernel Level-2 Data Processing
Pre-merging sensor cross-characterization Statistics table for the global comparison for year 2003 (annual min/avg/max): To be compared to the one issued by the JRC team based on monthly data over 36 months (July 2002 – June 2005):
Level-3 characterisation data For GlobCOLOUR PPS months (July and October 2002, January and April 2003) a total of 124 in situ data points were available from the OBPG validation database:
Level-3 characterisation data As the validation using OBPG dataset for the PPS months was strongly influenced by coastal waters, it was decided that a validation would also be performed using the BOUSSOLE data; chlorophyll data from 2002-2003 and radiances from 2003-2006.
BOUSSOLE Level-3 Characterisation Higher median % differences than the OBPG results.
Chl-a Comparisons Target of less than 35%
Overall Conclusions • In Situ Database: data was converted to a consistent format that was easily distributed within the consortium. • Level-2 Characterisation: rather than correcting for the bias in the level-2 data, the results were reflected in the error bar estimates. • In situ: ideally more in situ data is needed, but the limited amount gave sufficient information for the merging process. • Cross Characterisation: The global analysis showed small biases between sensors, especially MERIS and MODIS, while the regional analysis showed that the differences between sensors fluctuates and is seasonally dependant.
Overall Conclusions • Level-3 Characterisation: are highly dependant on the in situ data. • Chlorophyll a: 35% met for OBPG results • OBPG results: smaller errors for the OBPG level-2 characterisation (PPS months only) than GlobCOLOUR unmerged level-3 data. For the merged data, the errors appeared to reduce; especially for the GSM model. • BOUSSOLE results: larger errors than the OBPG data despite it being a predominately Case 1 site, but it is influenced by Saharan dust and seasonal CDOM values. • nLw: The 5% error level is not yet met for either the level-2 or level-3 data (8.5-20.5% excluding 670nm for merged data).
Thank you for your attention