1 / 22

3D Cloud Structure and Retrievals of Droplet Sizes

3D Cloud Structure and Retrievals of Droplet Sizes. Alexander Marshak (GSFC) Steven Platnick (GSFC) Támas Várnai (UMBC) Guoyong Wen (UMBC) Robert Cahalan (GSFC). ASTER image and MODIS cloud product. Brazil, Aug. 9, 2001 centered at 17 o S, 42 o W. r e. .

cbernard
Download Presentation

3D Cloud Structure and Retrievals of Droplet Sizes

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. 3D Cloud Structure and Retrievals of Droplet Sizes Alexander Marshak (GSFC) Steven Platnick (GSFC) Támas Várnai (UMBC) Guoyong Wen (UMBC) Robert Cahalan (GSFC) Alexander Marshak

  2. ASTER image and MODIS cloud product Brazil, Aug. 9, 2001 centered at 17oS, 42oW re  Alexander Marshak

  3. ASTER image and MODIS cloud product Brazil, Jan. 25, 2003 centered at 0o, 53.78oW re  Alexander Marshak

  4. Are these large values of re real? Alexander Marshak

  5. 1km MODIS pixel What are the 3D Radiative Effects? Unresolved or sub-pixel variability Resolved or effects of neighboring pixels (e.g., shadowing) 1km MODIS pixel Alexander Marshak

  6. A little bit of theory • Under some general assumptions, it can be shown • that ignoring sub-pixel (unresolved) variability produces a negative bias in the retrieved re • while ignoring cloud inhomogeneity at scales larger than a pixel scale (resolved variability) leads on average to overestimation of re Alexander Marshak

  7. qv=60o jv=0o qo=60ojo=0o qv=60o jv=90o qv=0ojv=0o qv=60o jv=180o Simulation I(direct problem) Cu cloud field from the Stevens’s LES model and calculated 4 bidirectional reflect. at 2.13 mm assuming re=10 mm. Alexander Marshak

  8. Simulation II (retrievals) Cu cloud top height field and retrieved re Alexander Marshak

  9. Simulation III (averaging) re retrieved at resolution 67 m and 335 m Alexander Marshak

  10. Simulation IV (re vs. ) Ignoring shadowing in high-resolution 1D retrievals causes substantial overestimation of rethat usually corresponds to underestimation of t Correlation between retrieved  and re for illuminated and shadowed areas. Alexander Marshak

  11. 60 km 300 km 8 km Observations from Terra: MODIS and ASTER Alexander Marshak

  12. ASTER image and MODIS cloud product Brazil, Aug. 9, 2001 centered at 17oS, 42oW re  Alexander Marshak

  13. Optical thickness for large re re  mean=16.1 m std=6.4 m mean=7.1 std=6.7 Pdfs of the retrieved re for all cloudy pixelsand retrieved for pixels with re > 25 m. Alexander Marshak

  14. MODIS (diffuse-incidence) surface albedo vs. nadir reflectance for pixels with retrieved re > 25 m at 0.67, 1.6 and 2.13 m. Sub-pixel clear sky contamination For the 2.1 m (and 1.6 m) band more than half of all pixels have a surface albedo larger than the nadir reflectance, which would act to decrease the retrieved re rather than increase it when the surface has sufficient direct illumination. Alexander Marshak

  15. 8 by 8 km sub-scene I MODIS re 1 km resolution ASTER B3N (~0.82 m) 15 m resolution ASTER B14 (~11.35 m) 90 m resolution Alexander Marshak

  16. 8 by 8 km sub-scene II The cloud tops of all pixels with large reare lower than the cloud tops of their neighboring (towards the sun) pixels. Therefore, here 2.13 mm reflectance is small NOT because of large highly absorbing droplets BUT RATHER because little direct solar radiation can reach those pixels. Alexander Marshak

  17. Brazil, Jan. 25, 2003 centered at 0o, 53.78oW re  Does the 3D effect explain all (operationally) retrieved large re for broken Cu clouds? Alexander Marshak

  18. 9 by 10 km sub-scene (MODIS) Cloud top height  re Alexander Marshak

  19. re 9 by 10 km sub-scene (ASTER) ASTER B14 (90 m) ASTER B3N (15 m) Alexander Marshak

  20. Possible reasons for large re • How to explain the combination of low 2.13 m and relatively high 0.67 m reflectance (needed to get a reasonable optical thickness of 15 and higher)? • 3D effects • small cloud optical thickness • sub-pixel clear sky contamination • wrong thermodynamic phase • multilayered clouds • inhomogeneous surface albedo • absorbing aerosol • instrumental issues In this case we cannot think of anything but large droplets to reduce the 2.13 m nadir reflectance, and thus the retrieval of large re appears justified Alexander Marshak

  21. Conclusions • Averaging (degrading to larger scales) decreasesre. • With respect to the pp approximation, shadowing increasesre more than illumination decreases it; this results in an overall bias towards largerre. • Ignoring shadowing in 1D retrievals results in substantial overestimation of re that often goes in pair with underestimation of . • Sub-pixel clear sky contamination and low surface albedo at 2.13 m contribute to small nadir reflectance and thus largere. • Not all retrieved re > 20-25 m can be explained by unaccounted 3D (or 1D) radiative features; in these cases, the retrieved large values of re for Cu clouds could be real. Alexander Marshak

  22. 3D Cloud Structure and Retrievals of Droplet Sizes Alexander Marshak (GSFC) Steven Platnick (GSFC) Támas Várnai (UMBC) Guoyong Wen (UMBC) Robert Cahalan (GSFC) Alexander Marshak

More Related