1 / 39

AN INITIAL LOOK AT THE IMPACT OF THE NEW ANTENNA LOSS MODEL Joe Tenerelli

AN INITIAL LOOK AT THE IMPACT OF THE NEW ANTENNA LOSS MODEL Joe Tenerelli SMOS QUALITY WORKING GROUP #4 7-9 March 2011. OVERVIEW.

vine
Download Presentation

AN INITIAL LOOK AT THE IMPACT OF THE NEW ANTENNA LOSS MODEL Joe Tenerelli

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. AN INITIAL LOOK AT THE IMPACT OF THE NEW ANTENNA LOSS MODEL Joe Tenerelli SMOS QUALITY WORKING GROUP #4 7-9 March 2011

  2. OVERVIEW • The Level 1 Team has developed a new antenna loss model that should help to correct for the impact of variations of the antenna physical temperatures on the reconstructed brightness temperatures. • In the following slides we will: • Review the bias problems in the SMOS brightness temperatures over the ocean; • Show that the direct sun correction algorithm may induce significant latitudinal variations in the bias between model and data over the open ocean; • Show that the direct sun correction may not behave as expected when the sun is in the back half-space of the array (Jerome’s presentation). • Look at the impact of the new loss model on a small set of Pacific Ocean orbits reprocessed by Roger Oliva (drift with latitude and temporal evolution). • Caveat: Given potential impact of the direct sun correction on the bias variations, we must use caution in interpreting the results.

  3. 10-DAY MEAN SSS BIAS: DESCENDING PASSES Problem 1: The differences between SMOS reconstructed brightness temperatures produced by DPGS and those predicted by our ‘best’ forward scene model drift with time. Certainly, a portion of this drift may be related to problems in the model (sun glint, galactic noise reflection). Another portion may originate with the instrument and Level 1 processing.

  4. 10-DAY MEAN SSS BIAS: DESCENDING PASSES Problem 1: The differences between SMOS reconstructed brightness temperatures produced by DPGS and those predicted by our ‘best’ forward scene model drift with time. Certainly, a portion of this drift may be related to problems in the model (sun glint, galactic noise reflection). Another portion may originate with the instrument and Level 1 processing. +0.5 K (Tx+Ty)/2 -1 psu

  5. Sensitivity of the first Stokes parameter to a 1 psuincrease in SSS: Roughly 1 K/psu in (Tx+Ty) in warm water Roughly 0.5 K/psu in (Tx+Ty) in cold water Best case (warm water): +.5 K (Tx+Ty)/2 -1 psu

  6. 10-DAY MEAN SSS BIAS: ASCENDING MINUS DESCENDING PASSES MID-MAY MID-AUGUST Differences in SSS retrieved from descending and ascending passes can reach more than 1 psu and change (even in sign) with time. END AUGUST MID-NOVEMBER

  7. HOVMOLLER PLOTS

  8. BIAS TRENDS OVERVIEW Several months ago it was noticed that, in terms of descending-ascending pass differences, there is a qualitative correlation between Tp7 and SMOS-model bias variations as a function of latitude and time: MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  9. BIAS TRENDS We have also seen discontinuities in the bias trends that seem to correlate with the sun passing between the front and back half-spaces. More on this later… MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  10. BIAS TRENDS The discontinuities in the bias as well as the correlation between Tp7 and the bias trends remain in the reprocessed data: MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  11. BIAS TRENDS However, the correspondence between Tp7 and the bias is less clear when we apply the simplified reconstruction, with no direct sun correction, to the reprocessed Level 1A files: MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  12. TP7 VARIATIONS AND SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  13. TP7 VARIATIONS AND SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  14. TP7 VARIATIONS AND SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  15. LATITUDINAL BIAS DRIFT We have also seen bias drifts with latitude, especially in November 2010 SSS bias decrease with latitude for descending passes is associated with (Tx+Ty)/2 increase by about 1 K (red curve).

  16. BIAS TRENDS IN TERMS OF SSS BIAS For the November descending passes, the DPGS SSS bias drift with latitude is consistent that the drift in AF-FOV bias in (Tx+Ty)/2: DPGS DESC-ASC JRECON DESC-ASC DPGS (Tx+Ty)/2 increase With latitude

  17. BIAS TRENDS DPGS – JRECON DESC-ASC FOR NOVEMBER 2010: POSSIBLE IMPACT OF DIRECT SUN CORRECTION

  18. IMPACT OF NEW LOSS MODEL Roger Oliva has reprocessed several orbits using the new antenna loss model implemented by DEIMOS. Results suggest that latitudinal drift in (Tx+Ty)/2 is reduced using the new loss model with the L1PP image reconstructions. Nevertheless, the drift is reduced further when we switch to the simplified reconstruction with no direct sun correction: Original loss model New loss model Extension to LICEFs November 9 Pacific Descending (red) vs Ascending (blue) Passes Fixed-Cal Reprocessing solutions (includes NIR AB)

  19. IMPACT OF NEW LOSS MODEL A similar result on November 10: Original loss model New loss model Extension to LICEFs November 10 Pacific Descending (red) vs Ascending (blue) Passes Fixed-Cal Reprocessing solutions (includes NIR AB)

  20. IMPACT OF NEW LOSS MODEL In June, when the direct sun correction has less impact than in November, drift with latitude is not a large problem in any solution. Actually, it seems the direct sun correction may introduce an offset that is largely independent of latitude (Needs to be confirmed using L1PP). In any case, the new loss model does improve agreement between ascending and descending passes. Original loss model New loss model Extension to LICEFs June 13 and 14 Pacific Descending (red) vs Ascending (blue) Passes Fixed-Cal Reprocessing solutions (includes NIR AB)

  21. BIAS TRENDS OVER TIME With the old loss model, differences between descending and ascending pass SMOS-model biases are significant and drift over time, as we have observed for a while: L1PP Level 1B JRECON

  22. BIAS TRENDS OVER TIME With the fixed calibration used to obtain the reprocessing solutions, ascending/descending differences remain significant but the drift over time of these differences seems to be reduced: L1PP Level 1B JRECON

  23. BIAS TRENDS OVER TIME With the new antenna loss model (NAM, here applied only to the NIRs), differences between descending and ascending pass solutions are much reduced, but the overall drift over time reaches about 0.8 K: L1PP Level 1B JRECON

  24. BIAS TRENDS OVER TIME Extension of the loss model to the LICEFS makes only a small impact in the bias trends over time: L1PP Level 1B JRECON

  25. CONCLUSIONS Relative to our best available ocean scene model, significant biases exist in all SMOS-derived brightness temperatures over the alias-free field of view. The biases seem to have an impact on retrieved SSS that is roughly independent of longitude (see the SSS bias maps). These biases drift with latitude (short time scale) and slowly over time (weekly to monthly time scales). The latitudinal bias drift appears to involve both bias in the calibrated visibilities and bias introduced by the direct sun correction algorithm (see also Jerome Gourrion’s presentation). The new antenna loss model reduces discrepancies between descending and ascending pass biases, but also increases the bias trend with time.

  26. EXTRA SLIDES

  27. LAND CONTAMINATION AND BIAS Maps of bias between SSS derived using DPGS L1B data and SSS derived using L1A and the JRECON breadboard are shown below. The strong land contamination halos in the DPGS solutions are not as evident in the JRECON solutions. For the descending passes, the linear bias gradient with latitude in the DPGS solutions (lower left) is not seen in the JRECON solutions (lower right). For all maps the sets of fixed OTTs shown earlier were used for both DPGS and JRECON solutions. DPGS ASC JRECON ASC DPGS DESC JRECON DESC

  28. THIRD AND FOURTH STOKES

  29. THIRD AND FOURTH STOKES

  30. AF-FOV MEAN TEC

  31. AF-FOV MEAN CELESTIAL SKY NOISE

  32. Combining three descending passes over the last six months… April 2 June 25 Sep 20

  33. The difference between the two preceding maps gives an indication of how the model is in error. Here we have taken just the SMOS sky map produced from data corresponding to ECMWF 10-m wind between 3 and 6 m/s and we subtracted the 3 m/s model solution. The difference map shows how the model underestimates the data in the vicinity of the galactic equator by over 1 K. Next we will compare the data and model in more detail along the cross section shown by the magenta line segment. SMOS-MODEL

  34. This is a cross section through the dwell lines near the galactic equator along the magenta line segment in the previous slide. The solid black curve is the ideal flat surface solution, the dashed black curve is the 3 m/s solution, and the colored curves are the actual SMOS data in various wind speed ranges with all geophysical sources except galactic noise removed. These curves are averages over celestial dwell lines and thus do not correspond to a specific viewing geometry. It is obvious from this plot that the actual SMOS data have peaks that are between the flat surface solution and the 3 m/s solution. However, away from the galactic plane the flat and rough surface solutions are very similar so that the actual SMOS residual is NOT A SIMPLE LINEAR COMBINATION of the flat and 3 m/s model solutions. Colored SMOS residualsky noise lies between the flat and rough surface model solutions (black curves) here. Colored SMOS residualsky noise curvesdo not lie between the flat and rough surface model solutions here.

  35. TP7 VARIATIONS MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  36. TP7 VARIATIONS: SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  37. TP7 VARIATIONS: SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  38. TP7 VARIATIONS: SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

  39. TP7 VARIATIONS: SUN GEOMETRY MAR APR MAY JUN JUL AUG SEP OCT NOV DEC MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

More Related