COSMIC GPS Radio Occultation Temperature Profiles in Clouds

L. LIN AND X. ZOUThe Florida State University, Tallahassee, Florida

R. ANTHESUniversity Corporation for Atmospheric Research, Boulder, Colorado

Y.-H. KUONational Center for Atmospheric Research, Boulder, Colorado

1. Introduction

2. A brief description of observations and analyses

3. Comparisons between COSMIC GPS RO data and large-scale analyses

a. Comparison of cloud-top temperatures

b. Model and RO refractivity differences in clear and cloudy air

4. A new GPS cloudy retrieval algorithm

a. Cloudy retrieval assuming saturation

b. An empirical GPS RO cloudy retrieval algorithm without assuming saturation

c. Numerical results

5. Summary

1. Introductionsitu dropsonde measurements from aircraft and weather balloons (radiosondes), have limited Earth coverage and are not available under severe weatherremote sensing from weather satellitescan only identify horizontal distributions of clouds

GPS RO retrievals global coverage, vertical high resolution

Earlier validations of RO data have been carried out in terms of refractivity as well as temperature retrieval and without separating cloudy- and clear-air soundings.

The standard 1DVar temperature retrieval did not include any information on clouds.

In this study:1. compare COSMIC RO data (T,N) with NCEP & ECMWF analysis from soundings in clouds and clear air separately2. examine whether the standard 1DVar T retrieval results are improved by addingancillary information on the vertical extent of clouds and their liquid water–ice content from collocated satellite observations.

2. A brief description of observations and analyses

CloudSat:

• 705-km near-circular sun-synchronous polar orbit

• launched on 28 April 2006

• orbits Earth approximately once every 1.5 h

• 94-GHz, nadirpointing cloud profiling radar (CPR)

• along-track temporal sample interval equals 0.16 s

• > 30 000 vertical profiles of radar reflectivity, liquid water content, and ice water content for each granule

• 1.1 km spatial resolution along-track

• FOV 1.4 km×2.5 km

• also provides cloud layers, cloud type, altitudes of cloud tops and cloud bases

2. A brief description of observations and analyses

• Cloudy COSMIC RO soundings must collocate with a CloudSat cloudy profile, where collocation is defined by a time difference of no more than a half hour and a spatial separation of less than 30 km.

• The cloud top must be above 2 km in order to minimize the impact of the uncertainty of RO wet retrievals in the lower troposphere.

• Only soundings with a single layer cloud are chosen.

FIG. 1. (a) CloudSat orbital track (red curve) at 1702:24 UTC 5 Jun 2007 and a GPS RO sounding (yellow marker) located at (43.798°N, 72.988°W) whose distance from the CloudSat orbital track is less than 30 km.

2. A brief description of observations and analyses

FIG. 1. (b) Reflectivity (unit: dBZ) observed by CloudSat along a segment of the orbit track indicated by the green line segment in (a).

2. A brief description of observations and analyses

FIG. 1. (c) Vertical profile of reflectivity observed by CloudSat at a point on its track (43.828°N, 72.888°W), that is closest to the GPS RO location. Dotted lines in (c) indicate the cloud top and cloud base. The yellow line through the dot in (a) indicates a segment of the tangent line representing the averaged ray direction, the middle point of the line (yellow dot) indicates the average tangent point position, and the length of the yellow line represents the vertical shift of all the tangent points of the occultation event.

2. A brief description of observations and analyses

FIG. 2. (top) Cloud types and (bottom) vertical profiles of N GPSwet － N ECMWF (red) and NGPSwet

－ N NCEP (blue) for three selected COSMIC cloudy soundings. The cloud top and cloud base are indicated by solid black lines. The dashed horizontal lines indicate the tropopause height calculated from GPS wet retrieval (black), ECMWF (red), and NCEP–NCAR (blue) analyses.

0922 UTC 2 Sep 2006 0331 UTC 20 Aug 2007 2018 UTC 20 Sep 2006

Ci: cirrus

Sc: stratocumulus

As: altostratus

Ns: nimbostratus

2. A brief description of observations and analyses

2. A brief description of observations and analyses

LWCwetdry

wobs

NNN

WT

P

T

PN

,4.11073.36.772

5

FIG. 3. (a) T NCEP (blue), T ECMWF (red), T GPSwet (black), and T GPSdry (green) at cloud top for the 11 single-layer cloudy soundings in June 2007. (b) Cloud-top height (open circle), cloud types, and cloud thickness (symbols).

3. Comparisons between COSMIC GPS RO data and large-scale analysesa. Comparison of cloud-top temperatures

FIG. 4. Temperature differences at the cloud top: (a) T ECMWF － T GPSwet, (b) T NCEP － -T GPSwet, and (c) T GPSdry － T GPSwet. (d) Mean (solid) and RMS differences (dashed) of T ECMWF － T GPSwet (red), T NCEP － T GPSwet (blue), and T GPSdry － T GPSwet (green). The y axis indicates cloud-top height.

T E

CM

WF－

T G

PS

wet

T N

CE

P－T

GP

Sw

et

T G

PS

dry－

T G

PS

wet

3. Comparisons between COSMIC GPS RO data and large-scale analysesa. Comparison of cloud-top temperatures

FIG. 5. Mean (solid) and RMS (dashed) of T ECMWF － T GPSwet (red) and T NCEP － T GPSwet (blue) within 2 km above and below the cloud top for soundings with the cloud top between (a) 2 and 5, (b) 5 and 8, and (c) 8 and 12 km. (d) The total numbers of cloudy soundings included in (a) (solid), (b) (dashed), and (c) (dotted).

2km~

5km

5km~

8km

8km~

12km

3. Comparisons between COSMIC GPS RO data and large-scale analysesa. Comparison of cloud-top temperatures

FIG. 6. (left) Mean and (right) RMS around the mean of COSMIC GPS and ECMWF fractional refractivity difference (a),(b) (N GPS － NECMWF)/N ECMWF and (c),(d) (N GPS － N NCEP)/N NCEP for cloudy soundings (solid) and clear soundings (dashed).

(N GPS － N ECMWF)/N ECMWF

(N GPS － N NCEP)/N NCEP

3. Comparisons between COSMIC GPS RO data and large-scale analysesb. Model and RO refractivity differences in clear and cloudy air

4. A new GPS cloudy retrieval algorithma. Cloudy retrieval assuming saturation

)61.01( sd qTRP

)(378.0

)(622.0

378.0

622.0

TeP

Te

PP

Pq

s

s

w

ws

65.29

)15.273(67.17exp112.6)(

T

TTes

)T,(

)(378.0

)(622.061.01

P

TeP

TeTR

P

s

sd

)( 11 mmmmmm zzgPP

topz

basez

0T 0P

mz

1mz

4. A new GPS cloudy retrieval algorithma. Cloudy retrieval assuming saturation

1mP

topz

basez

0T 0P

mz

1mz obs1mN GPSsat

1mT

LWCwetdry

wobs

NNN

WT

P

T

PN

,4.11073.36.772

5

4. A new GPS cloudy retrieval algorithma. Cloudy retrieval assuming saturation

topz

basez

0T 0P

mz

1mz

GPSsat1mT

CTCT mm 5,5

2obs1)()( mNTNTJ

N are calculated at different T in this temperature range at 0.1°C interval

GPSsat1mT

FIG. 7. The squared difference of refractivity (black curve) between GPS RO observations (Nobs) and model-calculated refractivity at different temperatures, assuming saturation within cloud (N sat) at six selected vertical levels (altitudes are indicated on the right). The solution of GPS saturation retrieval T GPSsat where N obs － N sat ≈ 0 is represented by the purple line. The RO sounding shown in this figure is located at 67.228°S, 144.18°W at 1437 UTC 21 Jul 2006.

4. A new GPS cloudy retrieval algorithma. Cloudy retrieval assuming saturation

FIG. 8. Dependence of the differences between GPS saturation retrieval and wet retrieval (T GPSsat － T GPSwet) on relative humidity calculated from ECMWF analysis using data from the middle of the cloud.

4. A new GPS cloudy retrieval algorithma. Cloudy retrieval assuming saturation

4. A new GPS cloudy retrieval algorithmb. An empirical GPS RO cloudy retrieval algorithm without assuming

saturation

cloud1

clear1

obss1 )1( mmm NNN

wet1

dry1

clear1 mmm NNN

sat1

dry1

cloud1 mmm NNN

25sat 1073.3

T

eN s

FIG. 9. Vertical variations of (a) liquid water content and (b) ice water content observed by CloudSat within the cloud. Vertical mean of (c) liquid water content and (d) ice water content. The y axis in (c),(d) is the vertical mean of relative humidity within cloud calculated from ECMWF analysis. The dashed line in (d) represents a linear regression model between ice water content and relative humidity with cloud. The open circles in (d) indicates outlier removed from linear fitting whose distance to the biweighting mean is more than 3 times the biweighting standard deviation.

4. A new GPS cloudy retrieval algorithmb. An empirical GPS RO cloudy retrieval algorithm without assuming

saturation

4. A new GPS cloudy retrieval algorithmb. An empirical GPS RO cloudy retrieval algorithm without assuming

saturation

8.0

,0.05975gmIWC,1

gm05975.0IWC,6849.0IWC273.53-

3

Ice Water Cloud

Liquid Water Cloud

IWC is the vertically averaged ice water content

FIG. 10. COSMIC GPS and ECMWF fractional refractivity difference (N GPS － N ECMWF)/N ECMWF with (stars) and without (dots) including the cloud liquid water term at the height of the maximum liquid water content (color bar) within 19 water clouds for which CloudSat data are available.

4. A new GPS cloudy retrieval algorithmc. Numerical results

FIG. 11. Vertical profiles of the mean and the 99% confidence interval (dashed line) of temperature differences between the cloudy retrievals and theROwet retrieval [i.e., T GPScloud －T GPSwet (green) and T GPScloud85 － T GPSwet (purple)] within clouds aligned at (a),(b) the middle of the cloud; (c),(d) the cloud top; and (e),(f) the cloud base.

middle of the cloud cloud top cloud base

4. A new GPS cloudy retrieval algorithmc. Numerical results

FIG. 12. (top) Biweight mean and (bottom) standard deviation of the lapse rate within the cloud calculated from T GPSwet (black), T ECMWF (red), T NCEP (blue), T GPScloud (green), and T GPScloud85 (purple) aligned by (a),(b) the middle height of cloud; (c),(d) the cloud top; and (e),(f) the cloud base. The numbers on the y axis indicate the vertical distance above (positive) and below (negative) the middle height, the cloud-top height, and the cloud-base height.

middle of the cloud cloud top cloud base

4. A new GPS cloudy retrieval algorithmc. Numerical results

TABLE 1. Biweight mean and standard deviation (in parentheses) of the moist-adiabatic lapse rate calculated from GPS wet retrieval, ECMWF analysis, NCEP–NCAR analysis, and GPS RO cloud retrieval data at the middle height of the cloud, cloud top, and cloud base. Unit: K km-1.

4. A new GPS cloudy retrieval algorithmc. Numerical results

5. Summaryexamined the vertical structure of T and lapse rates within clouds as retrieved from COSMIC GPS RO refractivity profiles and compared to T profiles from ECMWF and NCEP–NCAR analyses.

Observed RO N are greater than the N of the ECMWF and NCEP–NCAR large-scale analyses within clouds

Observed RO N are less than the N in the ECMWF and NCEP–NCAR analyses in clear areas .

With a new cloudy retrieval algorithm proposed in this study, the in-cloud T profile is retrieved from a weighted sum of clear-sky and cloudy N values.

Cloudy T retrieval is consistently warmer within clouds by ~2 K and slightly colder near the cloud top and cloud base,

The average lapse rate within clouds is nearly constant and close to the moist lapse rate in the lower half of the clouds. In the absence of ice water content measurements, an empirical value of 85% seems to be a good approximation to α.