1. A collection of cloudsat/LAPS comparisons from late 2006 and 2007 Brief overview of the cases 21 Dec 07, 0745 UTC cloudsat passage – end of a huge snowstorm in eastern Colorado. Our best winter case, and also the only one where we are able to use the special processing of the cloudsat data by the University of Utah, as this is unavailable after January 2007. One problem is the 0800 UTC LAPS is missing, so comparison is made to the 0700 and 0900 UTC LAPS cross-sections. Decent agreement between cloudsat and LAPS. 22 July 07, 2033-2040 UTC cloudsat passage – monsoon convection intersected across the western portion of the LAPS domain. Convection is not very active at this time (20-21 UTC), but it is apparent that some of the relatively weak convection is totally missed by LAPS, likely in areas where ground radar coverage is poor. 24 July 07, 2021-2027 UTC cloudsat passage – monsoon convection again, but this time the cloudsat pass is more across the middle of the LAPS domain and the convection is more active. A better comparison; LAPS generally agrees with the location of the storms, but has higher cloud tops (by ~1-4 km) than indicated by cloudsat (tops ~10 km). Also, again some convection missed by LAPS in areas of poor ground radar coverage. 28 July 07, 0900 UTC cloudsat passage – monsoon moisture but this time cloudsat crosses an MCS with a strong line of convection and good area of trailing stratiform rain in far ne CO into nw KS. Extensive anvil with this system and the anvil is resolved better by cloudsat than LAPS. 7 Aug 07, 1800 UTC cloudsat passage - Typhoon Pabuk case in Taiwan. Overall good agreement in location and height between cloudsat and LAPS, with the cross-section just off the east coast of Taiwan and the typhoon over southern Taiwan, so all within good ground radar coverage.

2. A collection of cloudsat/LAPS comparisons from late 2006 and 2007 Brief overview of the cases (continued) 9 Aug 07, 0530 UTC cloudsat passage – 2 days later a much weaker Typhoon Watup crosses southern Taiwan. Cross-section is off the west coast of Taiwan but the north and south ends are well beyond Taiwan ground radar range. It appears that the GSD LAPS analysis does not have any radar data ingested into it. Cloud top in LAPS is near 10 km in agreement with cloudsat, but there is an overshooting top in LAPS that is not in cloudsat. Echoes in cloudsat in China are not found in the LAPS analysis to the same strength. 17-18 Aug 07, 0522 UTC - Super Typhoon Sepat hits Taiwan on 17-18 Aug 2007, with the cloudsat segment ~nnw-sse across the eastern portion of Taiwan on the 18 th after the storm had emerged off the west coast of Taiwan. Again good agreement between the general LAPS cloud top and that in cloudsat, but LAPS has some taller cells not found in cloudsat. Good resolution of a rain shadow caused by terrain in central Taiwan in both LAPS and cloudsat. Nice case of how cloudsat is able to fill in where it is obvious LAPS is lacking radar in the analysis, and this occurs in the southern part of the cross-section where LAPS agrees with cloudsat on the general cloud top (using GOES satellite information), but does not have much echo below this, whereas cloudsat does show echo in this area. 18 Sep 07, 1745 UTC - Super Typhoon Wipha grazes northern Taiwan, with the center passing just to the north of the island. Cloudsat passage is when the storm is still some distance to the east, but gets a good portion of the western side of the storm with a ssw-nne segment off the east coast of Taiwan. Probably owing to this segment being in good ground radar coverage, this case has quite good agreement between LAPS and cloudsat.

3. Case 1: Weather map with radar at 1200 UTC 21 Dec 2006. Weather overview (from Unisys web site): 00z 21 Dec - Height of a huge winter storm in Colorado.

4. Case 1: Satellite image at 0000 UTC 21 Dec 2006.

5. Case 1: Weather map with radar at 1200 UTC 21 Dec 2006. Weather overview (from Unisys web site): Tail end of a huge winter storm in Colorado.

6. Case 1: Satellite image at 1200 UTC 21 Dec 2006.

7. Case 1: Location of cloudsat/LAPS cross-sections Line from B to B' shows the LAPS 0700 UTC and 0900 UTC cross- section. A-A' represents the cloudsat segment plotted in the next 2 figures.

8. Case 1: Satellite image at 1200 UTC 21 Dec 2006.

9. Case 1: Comparison of vertical cross sections of CPR radar (top, at 0745 UTC on 21 Dec 2006) from cloudsat and LAPS cloud cover (bottom, at 0700 UTC on 21 Dec). Horizontal height lines are shown at 5 and 10 km. Vertical lines locate similar topographic features in both images. The echo tops are at very low reflectivity levels so should be very close to the cloud top. Best agreement with LAPS on cloud top is with the reflectivity mass on the left side. LAPS does not do as well with the structure on the right (north end of the cross- section).

10. Case 1: Comparison of vertical cross sections of CPR radar (top, at 0745 UTC on 21 Dec 2006) and LAPS cloud cover (bottom, at 0900 UTC on 21 Dec). Horizontal height lines are shown at 5 and 10 km. Vertical lines locate similar topographic features in both images. LAPS cloud top for the main cloud is in decent agreement with cloudsat, but a higher level is also found in LAPS near 10 km and this is not found in cloudsat (but is it possible that cloudsat does not pick up thin cirrus?).

11. Case 2: 22 July 2007. Cloudsat pass that coincides with the western third of the LAPS coverage and intersects some monsoon thunderstorms/showers at about 1930 UTC. The LAPS cross-section runs along the line shown from point E to point G. Parts of two cloudsat segments are combined. Cloudsat segment 20 extends from southern Canada to Point H, while cloudsat segment 19 covers the southern part of the line. Comparison (shown in the next few slides): The 1900 UTC LAPS analysis is used for the comparison (2000 UTC was not available). For this case LAPS analyzed one main storm which corresponds in location fairly well with the cloudsat echo, however it is analyzed to be at least a couple of km higher than cloudsat indicates. Additionally, a couple of echoes that cloudsat shows farther south are not in the LAPS analysis. Radar coverage in the intermountain West can be spotty, and perhaps this explains the missing echoes in the LAPS analysis.

12. Case 2: Cloudsat segment 20. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

13. Case 2: Cloudsat segment 19. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

14. LAPS cloud cover and radar cross-section compared with cloudsat cross-section.

15. Case 3: More monsoon storms intersected by cloudsat on 24 Jul 07 -The 2 cloudsat segments run along the line shown from point E to Point J. Segment 19 goes from J to F, and segment 20 from F to E. -As in the 22 Jul case, LAPS overlays the southern portion of one cloudsat segment, and the northern portion of another, along the segment from I to A. A closeup of this segment overlaid with a visible satellite image (at 2015 UTC) and 1-h lightning plot is on the right.

16. Case 3: More monsoon storms intersected by cloudsat on 24 Jul 07 -The 2 cloudsat segments run along the line shown from point E to Point J. -As in the 22 Jul case, LAPS overlays the southern portion of one and the northern portion of another, along the segment from I to A. A closeup of this segment overlaid with a visible satellite image (at 2045 UTC) and 1-h lightning plot is on the right.

17. Case 3: Cloudsat segment 19. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

18. Case 3: Cloudsat segment 20. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

19. LAPS cloud cover and radar cross-section at 2000 UTC compared with cloudsat cross-section from ~2025 UTC.

20. LAPS cloud cover and radar cross-section at 2100 UTC compared with cloudsat cross-section from ~2025 UTC.

21. Summary of the 24 July case The cloudsat passes of interest occur between the LAPS hourly analyses (at 2024 and 2027 UTC), so we compare to both the 2000 and 2100 UTC LAPS analyses. Some observations Pretty good agreement in location between the cloudsat resolved convection over Colorado and the LAPS analyses. But LAPS consistently analyzes the echo top as well over 10 km, whereas cloudsat indicates echo tops around 10 km. The convection at the northern end of the LAPS cross sections (left side in the previous figures) in northern Wyoming into very southern Montana is not resolved as well by LAPS. For both times the LAPS analyses are too weak with this convection, although the trend is in the right direction. Radar coverage in this area is likely quite spotty owing to terrain blockage of the KBLX radar to the north, and distance to the KRIW radar well to the south.

22. Case 4: Cloudsat pass ~0900 UTC on 28 July 07. The cloudsat pass (line segment D- D') skirts the eastern portion of the LAPS domain (shown by the area covered by wind barbs).

23. Case 4: This pass goes across an active line of storms with a huge trailing area of more stratiform rain. Shown is radar reflectivity at 0900 UTC on 28 Jul 07.

24. Case 4: Location of cloudsat and LAPS cross-sections. -Both cross-sections go from south to north. -The cloudsat segment is a little bigger on both ends then the LAPS domain. -Main problem is that the cloudsat segment is from ~0855 UTC, and the closest LAPS is from 1000 UTC. Conclusions (from slides following this): -This is a big complex, with both a squall line and a large trailing region of rain, so there should be an extensive anvil, as shown in cloudsat. -There are some echoes in cloudsat not found in LAPS, but this could be partly caused by the time difference. -Good agreement between the general cloud top in LAPS and the cloudsat echo top. -But the extent of the anvil is much greater in cloudsat as opposed to the somewhat displaced anvil in LAPS.

25. Case 4: Cloudsat segment 28, 0853-0856 UTC on 28 July 2007, going from ~south to north. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

26. LAPS cloud cover cross-section at 1000 UTC, compared with cloudsat cross-section from ~0855 UTC, on 28 Jul 07

27. Case 5: Cloudsat pass in Typhoon Pabuk moving across southern Taiwan on 7-8 August 07. The cloudsat pass (shown later) goes ~north to south just off the east coast of Taiwan, intersecting the eastern portion of the typhoon along with a major rainband to the north. Convection in this case is more tropical of course than sampled in Colorado, and as such extends to quite high altitudes.

28. Case 5: Typhoon Pabuk forecast from the Central Weather Bureau of Taiwan.

29. Case 5: Similar graphic from the Joint Typhoon Warning Center

30. Case 5: Cloudsat pass is just off the east side of Taiwan. Cloudsat segment #30 goes north to the southern cloudsat #29 point, while cloudsat segment #30 extends north of the northern extent of LAPS. LAPS cross-section runs from LAPS to LAPS'.

31. Case 5: Cloudsat segment 30, 1751-1754 UTC on 7 August 2007, the southern segment, going from ~south (left side) to north. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

32. Case 5: Cloudsat segment 29, 1748-1751 UTC on 7 August 2007, the northern segment, going from ~south to north. The vertical scale of the cloudsat radar image goes from 0 to 20 km.

33. LAPS CWB cloud cover (image) and reflectivity (contours) cross-section at 0800 UTC, compared with cloudsat cross-section from ~1748-1754 UTC, on 7 Aug 07. South is at the left side of the cross-section. rainbandtyphoon core

34. Summary of Case 5 Cloudsat compared to LAPS cloud and reflectivity analysis Good agreement between the location of the convection and general intensity With the exception of cells at the far southern end of the LAPS domain that LAPS has, but are too weak, compared to cloudsat Height of the rainband anvil as analyzed by LAPS agrees well with cloudsat The cloud top for the typhoon core is analyzed a little higher than indicated by cloudsat for one convective cell (in LAPS), but overall is in decent agreement, with both cloudsat and LAPS showing a higher general echo top in the typhoon core region.

35. Case 6: Two days later another storm hits Taiwan, a weaker tropical storm called Wutip on 9 Aug 2007. This storm was quite a bit weaker than the earlier typhoon but dropped a lot of rain. There is a great cloudsat pass that nearly coincides with the LAPS domain. CWB radar image is at 1200 local time on 9 Aug, or 0400 UTC.

36. Track of Tropical Storm Wutip from 7-9 Aug 07, from the Unisys hurricane page. Color code for intensity is given.

37. LAPS cloud cross-section location (from “Home” to D') and cloudsat scan (H-H'), with 0600 UTC IR image.

38. Cloudsat cross-section segment 18, extending ~north to south, from ~0530 UTC, on 9 Aug 07

39. LAPS cloud cover and reflectivity cross-section from the CWB analysis at 0500 UTC, compared with cloudsat cross-section from ~0530 UTC, on 9 Aug 07.

40. LAPS cloud cover cross-section from the GSD analysis at 0500 UTC, compared with cloudsat cross-section from ~0530 UTC, on 9 Aug 07.

41. LAPS cloud cover and reflectivity cross-section from the CWB analysis at 0600 UTC, compared with cloudsat cross-section from ~0530 UTC, on 9 Aug 07.

42. LAPS cloud cover cross-section from the GSD analysis at 0600 UTC, compared with cloudsat cross-section from ~0530 UTC, on 9 Aug 07.

43. Summary for 9 August comparison The GSD LAPS analysis does not have any radar data ingested into it. Because of this the lower structure of the clouds is way too weak, but the cloud top is generally in good agreement with cloudsat. Cloud top in LAPS is near 10 km in agreement with cloudsat, but there is an overshooting top in the LAPS CWB analysis that is not in cloudsat. Echoes in cloudsat in China are not found in the LAPS analysis to the same strength. The exception is the 0600 UTC CWB analysis, that has a single echo on the southern China coast that is in fairly good agreement with cloudsat, except for perhaps extending a little bit too high.

44. Case 7: Super Typhoon Sepat hit Taiwan on 17-18 Aug 2007. Cloudsat passage available after the storm had crossed Taiwan.

45. Comparison of LAPS and cloudsat cross-section locations for the 18 Aug 07 case.

46. Cloudsat cross-section going from north to south (segment 18) from ~0525 UTC, on 18 Aug 07

47. LAPS cloud cover and reflectivity CWB cross-section at 0500 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section North South

48. LAPS reflectivity CWB cross-section at 0500 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section North South

49. LAPS cloud cover and reflectivity GSD cross-section at 0500 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section NorthSouth

50. LAPS cloud cover and reflectivity CWB cross-section at 0500 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section North South

51. LAPS cloud cover and reflectivity CWB cross-section at 0600 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section

52. LAPS cloud cover and reflectivity GSD cross-section at 0600 UTC/18 Aug, compared with 0530 UTC cloudsat cross-section

53. Some observations of the 18 August 2007 comparison 0500 UTC LAPS compared to cloudsat CWB LAPS Cloud cover – general cloud top height is comparable to cloudsat, but some LAPS cells are taller. CWB LAPS reflectivity – echo height and location look good compared to cloudsat, but note how LAPS is missing the cells farther south of Taiwan. This area is likely out of radar range so without that input the analysis is not nearly as good. The “rain shadow” just south of the Taiwan highest terrain is nicely shown in both LAPS and cloudsat. GSD LAPS analyses are very close to the CWB ones (and do use radar data for this case). 0600 UTC LAPS compared to cloudsat As with 0500 UTC, the CWB and GSD LAPS cloud cover and reflectivity analyses are very close The main echo height in both LAPS analyses is around 10 km, similar to cloudsat As before, the echoes that cloudsat sees farther to the south are not found in LAPS, although, as before, the cloud height is still about the same, suggesting that without any radar data (or surface stations) LAPS used the GOES imagery to correctly analyze the anvil top, but does not have enough information to put any cells beneath this.

54. Cross-section locations for Super Typhoon Wipha case of 18 September 2007. The LAPS cross-section across the typhoon is covered by two cloudsat segments (#29 and #30).

55. CWB radar reflectivity views of Wipha The cloudsat pass and LAPS analyses are ~90 min before this image, and sample the western portion of the storm. CWB radar image later in the day as it passes to the north of Taiwan. Image is at 1800 LT = 1000 UTC on 18 Sep 07.

56. Track of Super Typhoon Wipha, from Unisys hurricane graphics.

57. Southern cloudsat segment (#30), running from south to north, and radar cross-section at ~1745 UTC on 17 Sep 2007

58. Northern cloudsat segment (#29) and radar cross-section at ~1742 UTC on 17 Sep 2007

59. LAPS cloud cover image and reflectivity CWB cross-section at 1800 UTC/17 Sep, compared with ~1745 UTC cloudsat cross-section

60. LAPS reflectivity image and cloud cover fraction (contours) CWB cross-section at 1800 UTC/17 Sep, compared with ~1745 UTC cloudsat reflectivity cross-section

61. LAPS cloud cover image and reflectivity GSD cross-section at 1800 UTC/17 Sep, compared with ~1745 UTC cloudsat cross-section

62. LAPS cloud liquid image and cloud ice GSD cross-section at 1800 UTC/17 Sep, compared with ~1745 UTC cloudsat cross-section

63. LAPS cloud cover and reflectivity image GSD cross-section at 1800 UTC/17 Sep, compared with ~1745 UTC cloudsat cross-section

64. Summary of the 17 September 2007 case Pass across the eastern side of a strong typhoon that grazes the northern portion of Taiwan GSD and CWB LAPS analyses are similar (for reflectivity and cloud fraction). The top of the anvil as analyzed by LAPS is in pretty good agreement (just above 10 km) with cloudsat Cloudsat reflectivity top is a little higher though than the LAPS reflectivity tops. This could be because of very low reflectivities sensed by cloudsat, though LAPS minimum contour is at 0 dBZ. Overall, there is probably better agreement with the location of echoes for this case than the other August typhoon case, but this is likely because the typhoon is close enough that CWB radars are resolving the echoes for the LAPS analysis.

65. Overall Summary The main goal of this preliminary study was to use CloudSat to validate the LAPS cloud analysis, and thereby determine how well LAPS, using GOES data, is able to resolve various cloud structures under a variety of weather scenarios. Our cases include two LAPS analysis areas: the ROC domain centered on Colorado, and the Taiwan area. Weather varies from a snowstorm case in Colorado to monsoon season convective situations in Colorado, as well as several typhoons in the Taiwan region. Below are some conclusions from the CloudSat comparisons to LAPS: For organized cloud systems such as typhoons or appreciably sized areas of convection (MCS, for example), GOES data as used by LAPS does a good job of determining cloud top height, as shown by the comparisons with the CloudSat data. This is also true for the winter storm case, with the cloud top for the main storm area well resolved by LAPS. For more scattered convection, such as the 22 and 24 July Colorado cases, LAPS tended to overestimate the height of the clouds. This could be a result of the scale of the clouds not being well-resolved by GOES IR imagery, so that LAPS uses radar more. Under these situations a higher cloud top could be analyzed by LAPS if the convection is some distance from the radar. Cloud heights can also be underestimated, however, when the convection is more scattered and located beyond radar coverage. If ground radar coverage is lacking and surface data is not available, then GOES data alone allows LAPS to still do a good job with the cloud top height, but often a poor job in analyzing the structure of the corresponding echo below the cloud. The best examples of this were found from the Taiwan cases for convection located some distance from Taiwan itself (or for the LAPS GSD analysis on 9 Aug, where LAPS did not ingest any radar data). The location of clouds as analyzed by LAPS usually agreed well with that indicated by CloudSat.