1 / 22

Algorithm Verification for AATSR: Level 2 Verification

Algorithm Verification for AATSR: Level 2 Verification. Andrew R. Birks Rutherford Appleton Laboratory Chilton, Didcot Oxfordshire OX11 0QX UK ENVISAT Validation Workshop, ESRIN, Frascati 9 - 13 December 2002. Objectives and Approach. To verify that

afyffe
Download Presentation

Algorithm Verification for AATSR: Level 2 Verification

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Algorithm Verification for AATSR:Level 2 Verification Andrew R. Birks Rutherford Appleton Laboratory Chilton, Didcot Oxfordshire OX11 0QX UK ENVISAT Validation Workshop, ESRIN, Frascati9 - 13 December 2002

  2. Objectives and Approach • To verify that • the algorithms used by the OP work correctly when presented with real AATSR data • the AATSR products are being correctly generated • To verify, and if necessary regenerate, the auxiliary data files used by the AATSR Operational Processor • Algorithm Verification is distinct from Product Validation, and aims to verify the processor, not the geophysical products • Verification is primarily based on scrutiny of AATSR Products • Verification Plan does not assume availability of RP data • Processor breakpoints not routinely available • Plan provides for use of ATSR-2 data if available

  3. Data for Algorithm Verification

  4. Format verification • Format verification of the Level 2 products was reported at the Calibration Review • IDL tool verifies that the logical format of the products is correct • Format verification was applied to all Level 2 products received • No errors were reported, and the calculated file size matched that given by the directory listing in each case

  5. Header Verification • Verification criteria: with the exception of certain product specific fields: • The MPH and SPH of the Level 2 Gridded (GST) product should be identical to those of the Level 1B product from which it was derived • The SPH of the Averaged (AST) and Meteo products should be identical, and differ from that of the GST product only by the omission of the tie point specifications, which apply only to the gridded product • These tests were applied to all the Level 2 Verification Products

  6. Results of Header Verification • MPH • No differences were found in any Level 2 product between the MPH and that of equivalent Level 1B product other than the expected differences in the product dependent fields (e.g. product name, product size, etc.) • In these cases the field values were correct • SPH • No material differences were found between the SPH of the GST product and that of equivalent Level 1B (GBTR) product other than the SPH descriptor field • Apart from the SPH descriptor field and the omission of the tie point specifications, the SPH of the AST and Meteo products were identical to the GST SPH

  7. Effects of Level 1B Problem Reports • Several minor problems with the Level 1B SPH were identified during Level 1B verification. It follows that the same problems are present in the Level 2 SPH • This does not invalidate the verification of the Level 2 SPH • The Level 2 SPH is derived from the Level 1B product SPH (there is no other source of the information). Thus it is logically correct to require that the Level 2 SPH be identical to that at Level 1b, even if the latter is wrong • The Level 1B errors will be corrected in the next release of the processor, and the Level 2 headers should automatically be corrected as a result. No correction to the Level 2 processor is required

  8. Verification of ADS • With the exception of the Summary Quality ADS (SQADS), the ADS of the Gridded (GST) Product should be identical to the corresponding ADS of the Level 1B product from which the GST product is derived • The Summary Quality ADS (SQADS) differs by the addition of four fields of Level 2 quality information. These represent percentages of cloudy and invalid pixels in a frame, and should be positive numbers in the range 0 to 10000 (in units of 0.01%)

  9. Results of ADS Verification • For each ADS in the Level 2 (GST) product, an automated comparison has been made with the corresponding ADS of the Level 1B product • This was done for each verification orbit • No differences were found in ADS #1 to ADS #6 inclusive • In the case of the SQADS, the only differences appear in the Level 2 quality fields. In these fields: • The values lie in the required range 0 to 10000 • The numerical values show sensible correlation with image properties

  10. Gridded (GST) Product Measurement Data Set • Full resolution image product ATS_NR_2P (GST) • Visual inspection, and comparison with the equivalent PP product (i.e. the product obtained by running the Level 2 PP with the Level 1B Verification Product as input) • Visual agreement between images and histograms of pixel values good • Surface type flagging matched the superimposed coastline; coastline is based on Geolocation ADS, so this is a test of consistency • Further quantitative scrutiny desirable

  11. Averaged (AST) Product Structure

  12. Relationships between Averaged Data Sets

  13. Verification of Averaged (AST) Product MDS • Verification criteria can be formulated as follows. • AST sea cells: • The averaged SST should be correctly related to the corresponding BT / TOA values • The dual view and nadir only SST values should be consistent; their difference should show a realistic range of values • AST land cells: • NDVI should be correctly related to corresponding averaged reflectance values from the BT / TOA records • LST should match the corresponding ABT value • All Data Sets. For each surface type and resolution separately: • The 30 arcmin [50 km] average surface temperature / NDVI should equal the average of corresponding 10 arcmin [17 km] cell values • The 30 arcmin [50 km] average BT / reflectance should equal the average of the component 10 arcmin [17 km] cell values

  14. AST Verification Tests • Both manual checks of selected records, and automated checks of a range of records (using IDL scripts) have been used • Manual checks are laborious, but a manual check on a single record is sufficient to identify a problem • An automated check of a range of records extends the range of conditions checked, and reduces the probability of unrecognised errors

  15. Retrieved Temperature MDS [17 / 50 km] • Manual checks on 17 / 50 km sea records have shown that: • 17 km AST confidence word is anomalous • SPR issued. OP has been corrected • 17 km nadir only SST agrees with independent IDL retrieval using infrared BTs from corresponding TOA / BT record • 17 km dual view SST on ascending arcs differs from corresponding nadir SST by an unrealistically large amount, and is clearly invalid • BT differences (nadir - dual) between 5.41 and 19.68 K • Health warning issued to PIs. • SPR issued. The problem is under investigation, and may now have been solved • Both nadir and dual view averaged SST in 50 km cells agree with average of component 17 km values • Standard deviations in 50 km cells also agree

  16. Retrieved Temperature MDS [10 / 30 arcmin] • Manual checks on 10 / 30 arc min sea records have shown that: • In 10 arcmin records both nadir only and dual view SST retrievals agree with independent IDL retrieval using infrared BTs from corresponding TOA / BT record • Both nadir and dual view averaged SST in 30 arcmin cells agree with average of component 10 arc minute values • Standard deviations in 10 arc minute cells also agree • These tests can be automated using IDL scripts, and this should be done to extend the range of the verification • Equivalent tests on land records pending

  17. 10 arc minute sub-cell co-ordinates • The latitude and/or longitude of the 10 arc minute sub-cells are calculated incorrectly when either is negative • The sub-cell should be identified by the co-ordinates of its south-west corner. In these cases the co-ordinates of a different corner are computed • Two adjacent sub-cells may be assigned the same co-ordinates when their common boundary is either the equator or the meridian • Maximum error of a single sub-cell: 10 arc min in each co-ordinate • This will also affect the Meteo product • SPR raised

  18. Brightness Temperature / Reflectance MDS • Initial spot check on 10 / 30 arc min sea records verified that average clear sea BTs in 30 arc minute cells agree with average of BT values in component 10 arc minute sub-cells • Automated tests have extended this to a wider range of records and to cloudy sea values. No discrepancies found • Equivalent check on 50 / 17 km MDS and Land MDS pending

  19. Verification of Meteo Product • Each Meteo product record contains a copy of the SST record for a 10 arc minute sub-cell plus the averaged clear sea infrared brightness temperatures for the same sub-cell • Verification criteria for the Meteo product: • Each infra-red brightness temperature field should equal the corresponding field in the AST product • The retrieved SST and related values should equal the corresponding field of the AST product • Checks have found no discrepancies in the BT and SST fields • The two halves of the confidence word appear to be transposed with respect to the confidence word in the AST product • SPR raised. Impact on the BUFR form of the product must be established when this is corrected

  20. Comparisons with ATSR-2 Data • Although not part of the original verification plan, a comparison of the AATSR Averaged (AST) Product with the equivalent ATSR-2 data is now planned for 4 orbits • Comparison of averaged products eliminates problems (different gridding schemes, changing cloud field) that would affect a comparison of gridded products • Status • AATSR Orbits selected • AATSR data available • ATSR-2 data product requests outstanding

  21. Auxiliary File Tuning • Level 2 Processor Configuration Data File ATS_PC2_AX • Miscellaneous threshold and geometrical parameters • No update of this file is currently envisaged • SST Retrieval Coefficient Data File ATS_SST_AX • Derived from radiative transfer modelling calculation; not subject to empirical adjustment • No update envisaged as a result of the algorithm verification activity • Were AATSR Validation to suggest that different modelling assumptions were appropriate, an updated modelling calculation might be made

  22. Conclusions • Verification tests have been applied to the L2 data products GST, AST and Meteo • No errors in the format, header or ADS have been identified (other than those associated with already known Level 1B SPRs) • No errors have been identified in the GST product – further scrutiny desirable • 4 Software Problem Reports raised against the AST and Meteo products: corrective action in hand • Checks of the Land MDS are still outstanding, and automated testing should be used to extend the range of conditions tested (including tests on Level 2 products now being distributed)

More Related