Presentation is loading. Please wait.

Presentation is loading. Please wait.

DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 1 Statistical Extrapolation of Vertically.

Similar presentations


Presentation on theme: "DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 1 Statistical Extrapolation of Vertically."— Presentation transcript:

1 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 1 Statistical Extrapolation of Vertically Resolved Cloud Information from CloudSat/CALIPSO Observations to Regional Swaths John Forsythe, Steven D. Miller, Phil Partain and Tom Vonder Haar Cooperative Institute for Research in the Atmosphere (CIRA) Colorado State University Fort Collins, CO with Rich Bankert and Jeff Hawkins Naval Research Laboratory Monterey, CA CIRA

2 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 2 Research Questions To what extent do limited (but detailed) observations of cloud vertical structure (water content) and geometric boundaries (top/base) relate to surrounding clouds? To what extent do limited (but detailed) observations of cloud vertical structure (water content) and geometric boundaries (top/base) relate to surrounding clouds? Can this concept be applied to ‘vertical-slice’ observations from active sensors (radar/lidar) to augment the information provided from passive sensor 2-D imagery? Can we use sparse cloud observations to create local 3-D cloud scenes? Can this concept be applied to ‘vertical-slice’ observations from active sensors (radar/lidar) to augment the information provided from passive sensor 2-D imagery? Can we use sparse cloud observations to create local 3-D cloud scenes? CIRA Hypothesis Certain kinds of clouds form under characteristic environmental conditions of temperature, moisture, stability, etc., that occur over regional/synoptic scales. Certain kinds of clouds form under characteristic environmental conditions of temperature, moisture, stability, etc., that occur over regional/synoptic scales. Local observations of clouds occurring within that regional/synoptic-scale environment may provide useful information about the surrounding cloud field. Local observations of clouds occurring within that regional/synoptic-scale environment may provide useful information about the surrounding cloud field. CloudSat (cloud radar) and CALIPSO (cloud lidar) have been providing vertical cloud profiles at ~ 1.1 km horizontal and 240 m vertical resolution since 2006.

3 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 3 Relevance Cloud base estimation in data void regions UAV operations (visibility / icing) Model evaluation (e.g. scene contributions to the NCAR Model Evaluation Tool) Representativeness of data near taken CloudSat/CALIPSO track (like in a field experiment) CIRA

4 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 4 Opaque Ice cloud Water cloud Target is under a Cirrus cloud, use these observations 30000’ 3000’ 10000’ Current Dilemma: Given sparse ceiling observations, how to determine the ceiling over a data-void region? x x x 1. Average the observations? 2. Be cautious and use the lowest value? 3. Use the nearest observation? 30000’ x 100 km (notional)

5 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 5 GSIP (GOES Surface and Insolation Product) Cloud Classification (applied to MODIS channels used here):   A 5-channel (0.65, 3.9, 6.7, 11, 12-13 µm) physically- based pixel-level cloud classification with heritage from AVHRR and GOES (Pavolonis and Heidinger, 2005). These are the baseline channels for most contemporary geostationary meteorological satellites.   Additional published spectral tests using MODIS 1.38 and1.61 µm channels added. Visible / Infrared Cloud Typing to Provide Spatial Context CIRA

6 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 6 CIRA CloudSat Track 5 channel MODIS GSIP (“GOES Imager-like”) 5 channel MODIS GSIP + 1.38 µm and synthesized 1.61 µm MODIS-derived cloud type provides spatial context. The GSIP classifier (Pavolonis and Heidinger, 2005)) shown for Typhoon Choi-wan case, September 15, 2009.

7 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 7 Typhoon Choi-wan MODIS Visible (view from east) with CloudSat / CALIPSO curtain overlaid in colors of MODIS cloud type. 10 km 0 km September 15, 2009 Question: What is the cloud vertical structure away from the curtain?

8 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 8 Brief Review of Method to Gather Cloud Type Spatial Statistics

9 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 9 1) Form Geometric “Traces” Cirrus Top Base Cumulus Top Base Used Cloud Scenario Classification to compute departures in base/top height for contiguous cloud layers of a given cloud type, traced from a reference point. Orbital sub-segment example CIRA

10 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 10 2) Composite these Traces > 3 E+08 cloud layer samples CIRA

11 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 11 3) Compute Statistics on the Composites of Traces Cloud Top Cloud Base Distance (km) CIRA Statistics also collected by region, season and surface type

12 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 12 The Application Concept CIRRUS STRATOCUMULUS CUMULUS Cloud Type & Top Height: Retrieved Liquid Water Path (LWP): Retrieved 1) Estimate the cloud base height: 2) Distribute LWP between cloud Top and Base according to cloud type: CIRA

13 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 13 Taking the Next Steps Towards 3-D Cloud Fields – Behavior of the GSIP Classifier

14 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 14 Good news, supposed to be multilayer Deep clouds From MODIS cloud mask Where Modis Cloud Mask says cloudy, a cloud layer is almost always present (i.e. Number of Layers > 0) # of cloud layers (GEOPROF_LIDAR) by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50°

15 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 15 5 channel classification # of cloud layers (GEOPROF_LIDAR) by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50° 5 channel + 1.38 and 1.61 µm tests More Overlap Generated

16 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 16 Aggregated cloud thickness (m) by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50° Missed clouds are thin Almost always > 5 km thick “true” cirrus Lower cloud masked Very similar to Glaciated class Perhaps the most variable class clear Partly Cloudy LiquidSupercooled Opaque Ice CirrusOverlap 5-channel classification

17 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 17 Summary A technique for providing vertically-resolved cloud information for the top-most layer(s) of passive imager swath data, based on cloud type dependent statistics from CloudSat/CALIPSO, has shown some skill. Early results show skill in prediction of cloud ceilings when applying the correlative approach and constraining cloud type. As the active datasets continue to grow, statistics for the stratified datasets will become increasingly robust. CIRA

18 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 18 Backup Slides

19 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 19 In-class vs Out-of-class Data Denial Experiment Design Distance (km) along CloudSat track (1.1 km spacing between samples) 050 100 150200 Truth point: Class = Opaque Ice Opaque Ice Statistics for different class (dotted line) Statistics for same class (solid line)

20 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 20 Predictions shown for lowest cloud base Percentage of lowest cloud bases within 1 km for members vs nonmembers of 5 GSIP cloud classes as a function of distance from predicted location. January 20, 2009 (14 CloudSat granules). Different Class: WaterSupercooledOpaque Ice Overlap Cirrus Distance between solid and dotted lines is justification for our hypothesis Same Class as Truth: KEY: Predictions Using

21 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 21 Standard Deviation Fits Deep Convection Tops CIRA

22 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 22 How Far to Extend?  Computed cumulative density functions for cloud top/base ‘climatologies’ from CloudSat as F(zone,season,sfc_type). Approximate Gaussian distribution to obtain standard deviations. CIRA At some point we can’t beat climatology…so join it.

23 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 23 Climatology-Based Limits Some variability with season & land/ocean found, but main variance tracks with latitude (  depth of troposphere) CIRA

24 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 24 Example 1: Application Along the CloudSat Groundtrack

25 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 25 Jan. 20, 2009 2130 UTC East Pacific Case Study * * * * * * * * * * * * ** CIRA

26 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 26 Altitude (AMSL, km) 0 5 10 15 100200 300 400500600700 700600 500 400300200100 REF Altitude (AMSL, km) 0 5 10 15 0 1000 2000 3000 4000 5000 Distance Along Ground Track (km) Cloud Top Cloud Base Predicted Base/Top Predicted Base/Top Predicted by Withholding Observations Within 1km of Reference Point Observed base/top colored by cloud class

27 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 27 Altitude (AMSL, km) 0 5 10 15 Altitude (AMSL, km) 0 5 10 15 0 1000 2000 3000 4000 5000 Distance Along Ground Track (km) Cloud Top Cloud Base Predicted Base/Top Predicted Base/Top Predicted by Withholding Observations Within 100 km of Reference Point Observed base/top colored by cloud class 100200 300 400500600700 700600 500 400300200100 REF

28 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 28 Altitude (AMSL, km) 0 5 10 15 Altitude (AMSL, km) 0 5 10 15 0 1000 2000 3000 4000 5000 Distance Along Ground Track (km) Cloud Top Cloud Base Predicted Base/Top Predicted Base/Top Predicted by Withholding Observations Within 200 km of Reference Point Observed base/top colored by cloud class 100200 300 400500600700 700600 500 400300200100 REF

29 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 29 Example 2: Toward “Off-CloudSat Groundtrack” Applications

30 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 30 CloudSat 2B GEOPROF-LIDAR profile overlaid on MODIS 11 µm image of atmospheric river. CloudSat track shown as purple line. GSIP class for each profile shown by color. Jan 20, 2009, 2150 UTC. 10 km 0 km

31 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 31 Clear Cirrus Altostratus Altocumulus Stratocumulus Cumulus Nimbostratus Deep Convection Overlap Cirrus Opaque ice Supercooled/Mixed Liquid water CloudSat 2B-CLDCLASS Type at Top of Cloud GSIP MODIS Class January 20, 2009. All daylight granules equatorward of 50° latitude. Only cloudy cases from MODIS Cloud Mask shown. % (of clouds in each GSIP class)

32 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 32 Cloud vertical occurrence by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50° Only cases where MCM says clear, but has cloud layer Expected to be bimodal Missed cirrus

33 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 33 Aggregated cloud thickness (m) by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50° Missed clouds are thin Almost always > 5 km thick “true” cirrus Lower cloud masked Very similar to Glaciated class Perhaps the most variable class

34 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 34 Good news, supposed to be multilayer Deep clouds From MODIS cloud mask Where Modis Cloud Mask says cloudy, a cloud layer is almost always present (i.e. Number of Layers > 0) # of cloud layers (GEOPROF_LIDAR) by GSIP MODIS class. January 20, 2009. Daylight granules equatorward of 50°

35 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 35 Predictions shown for base of topmost layer Percentage of highest cloud bases within 1 km for members vs non-members of 5 GSIP cloud classes as a function of distance from predicted location. January 20, 2009 (14 CloudSat granules). Same class: Any class: Different class: WaterSupercooled Opaque Ice OverlapCirrus Distance between solid and dotted lines is justification for using GSIP classes Less skill for cirrus and overlap classes, requires more work Solid: “How well we potentially could do” Dashed: “How well a forecaster can do now” KEY: Predictions Using

36 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 36 Next Steps Encode the range-dependent limits of application (based on climatology analyses). Introduce refined statistics based on zones, season, etc. Refine the GSIP software to incorporate more information from MODIS (e.g., 1.38 micrometer band) to improve cloud classification. Conduct reanalysis when the combined CloudSat + CALIPSO cloud classifier becomes available. CIRA

37 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 37 Conditions @ TARGET Obs. class: opaque ice. Predicted ceiling: 19750’ Predicted ice profile also supplied. Obs. class = opaque ice Obs. ceiling = 20000’ 21000’ ceiling over carrier; class = opaque ice Obs. class = opaque ice Obs. ceiling = 19500’ ! ! Forecast models have difficulties predicting cloud cover (horizontal and vertical extent) and water content. CONCEPT: Take cloud observations in friendly areas, extend them into data-denied areas. AFGHANISTAN x x x

38 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 38 CIRA Water Content Vertical Structure Image Courtesy of UCAR * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Using CloudSat Level 2 Cloud Water Content (2B-CWC; R03) and Cloud Scenario Classification (2B-CLDCLASS) products, we computed liquid/ice water content (g/m 3 ) profiles for each cloud type. 0 1 1 0 2.3 e+06 samples Water Content (Normalized) Height (Normalized) Top Base

39 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 39 Observed base/top colored by cloud class Cloud Tops 1 km Exclusion Observed (km) 100 km Exclusion 200 km Exclusion Observed (km)

40 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 40 Observed base/top colored by cloud class 1 km Exclusion 100 km Exclusion 200 km Exclusion Observed (km) Lowest Cloud Base

41 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 41 Landscape of CG/AR Extended Cloud Statistics Work Create truth datasets for cloud forecast model verification. Aligns with AWFA ACAPS effort via scene contributions to the NCAR Model Evaluation Tool (MET). Three year NASA project underway to use A-Train datasets for model evaluation. Current CG/AR work: Create methodology to nowcast (0-3 hr) cloud base in data-denied areas from sparse observations. Many applications such as WRE-N and UAV routing tool. Improve science understanding of cloud vertical occurrence and remote sensing techniques. Connects with NRL Monterey work on cloud classification (Miller, Bankert, Mitrescu et al.) CIRA

42 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 42 Vertical Structure Statistics Vertical structure consistent with expected LWC profiles of convective/stratiform types - - Cirrus: growth of IWC in fall streaks prior to sublimation - - Cumulus: growth of droplets in ascending air CIRA

43 DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 43 4) Stratify Composites CirrusCumulus Distance (km) Tops Bases 90  75  45  15  -15  -45  -75  -90  NHEM1 NHEM2 NHEM3 SHEM1 SHEM2 SHEM3 TROP CIRA


Download ppt "DoD Center for Geosciences/Atmospheric Research at Colorado State University CoRP Symposium August 10-11, 2010 1 Statistical Extrapolation of Vertically."

Similar presentations


Ads by Google