richard t. mcnider atmospheric sciences department university of alabama in huntsville

80
Richard T. McNider Atmospheric Sciences Department University of Alabama in Huntsville [email protected] Assimilation of Satellite Cloud Products to Improve Radiative Characterization

Upload: channer

Post on 24-Feb-2016

50 views

Category:

Documents


0 download

DESCRIPTION

Assimilation of Satellite Cloud Products to Improve Radiative Characterization. Richard T. McNider Atmospheric Sciences Department University of Alabama in Huntsville [email protected]. Use of Satellite Data to Improve the Physical Atmosphere in Air Quality Decision Models - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Richard T. McNider

Atmospheric Sciences Department

University of Alabama in Huntsville

[email protected]

Assimilation of Satellite Cloud Products to Improve Radiative Characterization

Page 2: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Use of Satellite Data to Improve the Physical Atmosphere in Air Quality

Decision Models

NASA Air Quality Applied Science Team Project

Physical Atmosphere Panel Meeting

April 25-26, 2012

Atlanta, GA

Page 3: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Physical Atmosphere Can Significantly Impact Atmospheric Chemistry and Resulting Air Quality

Most Importantly the Physical Atmosphere Can Impact Control Strategy Efficacy and Response

Temperature Mixing HeightsClouds Winds

Page 4: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Clouds

Insolation

Temperature

Mixing Heights

Emissions

Photolysis J (NO2)

Deep Vertical Mixing

Boundary Layer Venting

Aqueous Chemistry

Aerosol Formation and Aging

Page 5: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Light Intensity

Light intensity (μmoles m-2 s-1)

Isop

rene

flux

(nm

oles

m-2 s

-1)

60

40

20

0

Light intensity (μmoles m-2 s-1)

Isop

rene

flux

(nm

oles

m-2 s

-

1 )

* a) Lerdau et al. (1997), b) Fuentes et al. (2000)

a) b)

White oak leaves

Saturation point

Page 6: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

0.65um VIS surface, cloud features

One of the things geostationary satellites do best is observe the reflectance of clouds

Page 7: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

10.7um IR sfc/cloud top temperature

Also can measure cloud top tempertaure/height

Page 8: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

10.7um IR sfc/cloud top temperature0.65um VIS surface, cloud features

Use of Daytime Cloud Albedo/Cloud Top Temperature for Model Evaluation

Model Cloud Albedo Model Cloud Top T

Page 9: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Underprediction

Overprediction

Areas of disagreement between model and satellite observation

Areas of Underprediction/Overprediction can be identified for Correction

A contingency table can be constructed to explain agreement/disagreement with observation

Page 10: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

So – What can we do to improve clouds in air quality studies?

One path is to simply replace model radiative properties with satellite observed properties.

Insolation (McNider et al 1995 Int. J. Remote Sensing)

Photolysis (Biazar et al. 2007 J. Geo. Res.)

Page 11: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

SUN

g

c

h

g

)(.1 cldcldcld absalbtr

Cloud albedo, surface albedo, and insolation are retrieved based on Gautier et al. (1980), Diak and Gautier (1983).

Surface

Insolation – Satellites can do an excellent job in providing insolation if care is taken to develop a consistent surface albedo.

Page 12: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Satellite Derived Insolation

Page 13: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Cloud Base

Determined from LCL

transmittance

Satellite MethodBaseline

WRF/CMAQ Method

Photolysis Rates

Transmittance =

1- reflectance - absorption

Observed by satelliteF(reflectance)

Cloud top

Determined from satellite IR temperature

Cloud top

Transmittance

Determined from LW= f(RH) and

assumed droplet size

Determined from model LW = f(RH)

Determined from model LW = f(RH)

Cloud Base

Page 14: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Photolysis Rates – Errors in photolysis rates can change response time of ozone production and change significantly levels at a given monitor

Page 15: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Observed O3 vs Model Predictions(South MISS., lon=-89.57, lat=30.23)

-40

-20

0

20

40

60

80

100

8/30/00 0:00 8/30/00 6:00 8/30/00 12:00 8/30/00 18:00 8/31/00 0:00 8/31/00 6:00 8/31/00 12:00 8/31/00 18:00

Date/Time (GMT)

Ozo

ne C

once

ntra

tion

(ppb

)

Observed O3

Model (cntrl)

Model (satcld)

(CNTRL-SATCLD)

70

-70

0

35

-35

Maximum Difference in Ozone due to Satellite Photolysis Fields

With Satellite

Without SatelliteObservations

Page 16: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Largest Surface O3 Differences Due to Cloud Errors - August 2006 (SatCld-Cntrl)

Page 17: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Dallas

Page 18: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 19: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 20: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Dallas Dallas

Photolysis rates at first model layer for August 21, 2006, at 1200LST

Model JNO2 Satellite JNO2

Page 21: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Photolysis rates at first model layer for August 21, 2006, at 1400LST

Model JNO2 Satellite JNO2

Page 22: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 23: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

While insertion of radiative properties (insolation and photolysis) can improve model performance the simple replacement of radiative properties means that we have an inconsistency in that radiative clouds are displaced from cloud water and cloud mixing in the model.

PBL Model Without Clouds

PBL Model WithClouds

CO profiles from P3 upwind, over, and downwind of Nashville (symbols)

Courtesy Wayne Angevine NOAA

Page 24: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

One solution -

Insert satellite derived cloud water into model (Lipton and Modica 1999 MWR)

In general clouds need positive vertical velocity to exist. Thus, insertion of liquid water into the model where it does not support clouds (low relative humidities and negative velocity) means that the water quickly evaporates.

In fact it can hurt model performance. If a model does not have clouds at a location that the satellite has clouds then the model likely has negative vertical velocity. Inserting the water and then having it evaporate will mean even greater downward motion which is the opposite of the satellite reality.

Insertion of satellite derived liquid water

Model Low RH and W<0 Evaporation

Sinking

Page 25: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

This looks like a perfect job for 4DVAR - simultaneously adjust moisture, winds, and cloud liquid water to dynamically and thermodynamically support the cloud using linear forward models.

However, a survey of the state of science reveals a lack of success in developing 4DVAR for clouds except for simple representations in very coarse grid models. This is in large part because cloud processes and cloud initiation are highly non-linear.

This is further exacerbated by the fact that clouds in models are highly parameterized and the coded relationships have many conditional and on-off switches which make developing the required inter-parameter relationships difficult if not impossible (Mu and Wang, 2003).

Page 26: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

W<0

W>0

Our first view comes from Meteorology 101 clouds have to have positive vertical motion to exist.

Page 27: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Cloud albedo

Graphical depiction of model relationship for cloudy condition between cloud albedo and vertical velocity

W > 0

The percentage of cloudy condition where vertical motion is positive is ~ 70%

Page 28: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Cloud albedo

W < 0

Graphical depiction of model relationship for clear condition between cloud albedo and vertical velocity

The percentage of time clear skies have negative vertical motion is greater than 60%

Page 29: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Statistical VAR Look for statistical relationships in model to specify target W

A scatter plot of cloud albedo versus cloud watersmall positive cloud water are associated with a wide range of cloud albedo

a scatter plot of cloud water versus W max even slight vertical velocities are related to a wide range of liquid water. 

By stratifying and scale selection were able to build multi-regression model to estimate a target w.

Page 30: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

SYMBOL DESCRIPTION UNITS

ZWMAX,MOD

Height of maximum vertical velocity as based on model data.

m

ZWMAX,EST

Height of maximum vertical velocity as based on a multiple linear regression equation.

m

DU,MOD

Depth of layer with upward motion as based on model data.

m

DU,EST

Depth of layer with upward motion as based on a multiple linear regression equation.

m

PS,MOD

One-hour stable precipitation amounts as based on model data.

mm

PS,EST

One-hour stable precipitation amounts as based on a multiple linear regression equation.

mm

PC,MOD

One-hour convective precipitation amounts as based on model data.

mm

PCEST

One-hour convective precipitation amounts as based on a multiple linear regression equation.

mm

Page 31: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Statistical VAR

1. Develop statistical relationships between model cloud albedo and model vertical velocity and RH to define a target vertical velocity.

2. Apply target vertical velocity where satellite has clouds and model does not.

3. Apply negative vertical velocity where satellite is clear and model has clouds.

4. Use variational analysis to minimally adjust model horizontal components (u,v) to meet target vertical velocity at cloud top (from satellite)

5. Solve Poisson type equation by SOR to ensure mass is conserved globally.

6. Nudge new horizontal wind field into the model.

Page 32: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

A

B C

Fig. 2. Downward shortwave radiation in W m-2 from different sources. Color shading is in increments of 50 W m-2 for values from 0 to 1100 W m-2 and is displayed in the color bar at the bottom of the figures. Panels (A) and (C) are at the time 1445 UTC 2 July 1999. Panel (B) is at the time 1500 UTC 2 July 1999. (A) Derived from GOES–8 satellite. (B) Control run with no assimilation. (C) Run with assimilation of satellite cloud information.

Satellite Insolation

Model Control Insolation

Model insolation after cloud assimilation

Page 33: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Contingency TableWRF

NC C

GOES

NC Clear agreed Over-prediction

# of GOES clear pixels

C Under-prediction

Cloudy agreed

# of GOES cloudy pixels

# of WRF clear

# of WRF cloudy

# of total grid cells

NC: No CloudsC: Clouds

We need metric for determining improvements in model clouds

Agreement Index (AI) is the sum across the diagonal

Page 34: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

AI for WRF_cntrl AI for WRF_assim

Agreement Index = (# of cloudy/clear grids in agreement) / (Total # of grids)

Over-prediction

Under-prediction

Created clouds

Removed clouds

Needs refinement

Page 35: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

GOES Calbedo WRF (cntrl) Calbedo

WRF (wind nudging) Calbedo

SNAPSHOTDate: August 13th , 2006 at 19 UTC

Page 36: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Daily Agreement Index

0.6

0.65

0.7

0.75

0.8

0.85

8/3/2006

8/4/2006

8/5/2006

8/6/2006

8/7/2006

8/8/2006

8/9/2006

8/10/20068/11/20068/12/20068/13/20068/14/20068/15/20068/16/20068/17/20068/18/20068/19/20068/20/20068/21/20068/22/20068/23/20068/24/2006

Date

Agr

eem

ent i

ndex

(fra

ctio

n)

AI_cntrl

AI_assim

Agreement index increased by 7-10%

Assimilation

Control

Agreement Index = (# of cloudy/clear grids in agreement) / (Total # of grids)

Page 37: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Analytical Approach: Under-prediction

Cloud top is known from GOES Search the column (from top) for

the air parcel that can be lifted to saturation.

Given a fixed time period (30 min), estimate the target vertical velocity.

Use 1-D variational technique to estimate horizontal wind components.

Nudge the model winds.

dtdzw

Page 38: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

dtdzw

Analytical Approach: Over-prediction

Model cloud properties are known. Estimate the height needed to reduce RH

below 100%. Given a fixed time period (30 min), estimate

the target vertical velocity (subsidence). Use 1-D variational technique to estimate

horizontal wind components. Nudge the model winds.

Page 39: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Cloud albedo

Cloud albedo W < 0

RH > 95

RH < 95

Graphical depiction of model relationship between cloud albedo, vertical velocity, and relative humidity

W > 0

There is no functional relationship of clouds with meteorological variables, but the sign of vertical velocity is relevant to cloud fields.

Page 40: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Typical Schemes used to infer moisture availability

Page 41: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Models have attempted to improve performance by developing improved land use classes (LUC) using in situ and satellite data

Page 42: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Unfortunately models don’t use land surface classes directly. Physical parameters such as heat capacity, canopy resistance, surface moisture have to be defined for the Land Use Class

Page 43: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

We have taken a different approach and have embraced simple models but highly constrained by observations.

Moisture, Heat Capacity, Solar Radiation from Satellite Observations

Page 44: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

EGHRdtdTC N

Gb

Surface Energy BudgetThree Uncertain Parameters

Bulk Heat Capacity Evaporative Heat Flux

MM5 Landuse Heat Capacity MM5 Landuse Moisture Availability

Net Short-wave radiation obtained from Satellite

Page 45: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Taken from Carlson (1986) to demonstrate the sensitivity of the surface energy budget model. Each panel represents the sensitivity of the simulated LST to uncertainty in a given parameter

Sensitivity of Surface Energy Budget to Various Parameters

Moisture Availability

Thermal Inertia

Page 46: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed between 1300-1400 UTC

12Z 00Z00Z 12Z

13Z 14Z

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

* Solar insolationLST

Rising in Morning

Model too moist

Model too dry

Page 47: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mm

G

S

GbS E

dtdT

dtdTCE

Can retrieve and analytically calculate the surface moisture that makes model temperature rate of change equal to satellite observed temperature change.

AIRGSAT

Ho

SS qTquzz

EM

)(

ln

Page 48: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

SatelliteObservation

Assimilation Control

Page 49: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mmNm

Gb EGHR

dtdT

Cm

ssNs

Gb EGHR

dtdT

Cs

s

G

m

Gbmbs dt

dTdtdT

CC

/

Determining Bulk Heat Capacity

Model Energy Budget

Satellite Energy Budget

Derived Heat Capacity

Page 50: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed in early evening

12Z 00Z00Z 12Z

17LST 19LST

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

LST Dropping in evening

Model heat capacity too small

Model heat capacity too large

Page 51: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Temperature TendencyMay 19,2002

22 UTC to 02 UTC(4-hours)

Heat Capacity

Model Default GOES-Inferred

Model GOES

Satellite-Inferred Heat CapacityMcNider et al. 2005 JCAM

Page 52: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 53: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Typical Schemes used to infer moisture availability

Page 54: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Models have attempted to improve performance by developing improved land use classes (LUC) using in situ and satellite data

Page 55: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Unfortunately models don’t use land surface classes directly. Physical parameters such as heat capacity, canopy resistance, surface moisture have to be defined for the Land Use Class

Page 56: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

We have taken a different approach and have embraced simple models but highly constrained by observations.

Moisture, Heat Capacity, Solar Radiation from Satellite Observations

Page 57: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

EGHRdtdTC N

Gb

Surface Energy BudgetThree Uncertain Parameters

Bulk Heat Capacity Evaporative Heat Flux

MM5 Landuse Heat Capacity MM5 Landuse Moisture Availability

Net Short-wave radiation obtained from Satellite

Page 58: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Taken from Carlson (1986) to demonstrate the sensitivity of the surface energy budget model. Each panel represents the sensitivity of the simulated LST to uncertainty in a given parameter

Sensitivity of Surface Energy Budget to Various Parameters

Moisture Availability

Thermal Inertia

Page 59: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed between 1300-1400 UTC

12Z 00Z00Z 12Z

13Z 14Z

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

* Solar insolationLST

Rising in Morning

Model too moist

Model too dry

Page 60: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mm

G

S

GbS E

dtdT

dtdTCE

Can retrieve and analytically calculate the surface moisture that makes model temperature rate of change equal to satellite observed temperature change.

AIRGSAT

Ho

SS qTquzz

EM

)(

ln

Page 61: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

SatelliteObservation

Assimilation Control

Page 62: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mmNm

Gb EGHR

dtdT

Cm

ssNs

Gb EGHR

dtdT

Cs

s

G

m

Gbmbs dt

dTdtdT

CC

/

Determining Bulk Heat Capacity

Model Energy Budget

Satellite Energy Budget

Derived Heat Capacity

Page 63: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed in early evening

12Z 00Z00Z 12Z

17LST 19LST

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

LST Dropping in evening

Model heat capacity too small

Model heat capacity too large

Page 64: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Temperature TendencyMay 19,2002

22 UTC to 02 UTC(4-hours)

Heat Capacity

Model Default GOES-Inferred

Model GOES

Satellite-Inferred Heat CapacityMcNider et al. 2005 MWR

Page 65: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 66: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Typical Schemes used to infer moisture availability

Page 67: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Models have attempted to improve performance by developing improved land use classes (LUC) using in situ and satellite data

Page 68: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Unfortunately models don’t use land surface classes directly. Physical parameters such as heat capacity, canopy resistance, surface moisture have to be defined for the Land Use Class

Page 69: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

We have taken a different approach and have embraced simple models but highly constrained by observations.

Moisture, Heat Capacity, Solar Radiation from Satellite Observations

Page 70: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

EGHRdtdTC N

Gb

Surface Energy BudgetThree Uncertain Parameters

Bulk Heat Capacity Evaporative Heat Flux

MM5 Landuse Heat Capacity MM5 Landuse Moisture Availability

Net Short-wave radiation obtained from Satellite

Page 71: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Taken from Carlson (1986) to demonstrate the sensitivity of the surface energy budget model. Each panel represents the sensitivity of the simulated LST to uncertainty in a given parameter

Sensitivity of Surface Energy Budget to Various Parameters

Moisture Availability

Thermal Inertia

Page 72: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed between 1300-1400 UTC

12Z 00Z00Z 12Z

13Z 14Z

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

* Solar insolationLST

Rising in Morning

Model too moist

Model too dry

Page 73: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mm

G

S

GbS E

dtdT

dtdTCE

Can retrieve and analytically calculate the surface moisture that makes model temperature rate of change equal to satellite observed temperature change.

AIRGSAT

Ho

SS qTquzz

EM

)(

ln

Page 74: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

SatelliteObservation

Assimilation Control

Page 75: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

mmNm

Gb EGHR

dtdT

Cm

ssNs

Gb EGHR

dtdT

Cs

s

G

m

Gbmbs dt

dTdtdT

CC

/

Determining Bulk Heat Capacity

Model Energy Budget

Satellite Energy Budget

Derived Heat Capacity

Page 76: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

* Assimilation performed in early evening

12Z 00Z00Z 12Z

17LST 19LST

Free Forecast

12h fcst 24h fcst 36h fcst

AssimilationPeriod

Upwelling IR at E20 (Meeker)7/7/95

350

360

370

380

390

400

410

420

0:01

:00

1:01

:00

2:01

:00

3:01

:00

4:01

:00

5:01

:00

6:02

:30

7:02

:30

8:02

:30

9:02

:30

10:0

2:30

11:0

2:30

12:0

2:30

13:0

2:30

14:0

2:30

15:0

2:30

16:0

2:30

17:0

2:30

18:0

2:30

19:0

2:30

20:0

2:30

21:0

2:30

22:0

2:30

23:0

2:30

UTC

W/m

*m

Upwelling Radiation

Assimilate:* Land Surface Temperature

Tendencies computed from hourly images.

LST Dropping in evening

Model heat capacity too small

Model heat capacity too large

Page 77: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Temperature TendencyMay 19,2002

22 UTC to 02 UTC(4-hours)

Heat Capacity

Model Default GOES-Inferred

Model GOES

Satellite-Inferred Heat CapacityMcNider et al. 2005 MWR

Page 78: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville
Page 79: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Temperature

In most areas maximum temperature is most correlated with ozone.

Thermal decomposition of nitrogen species – (Sillman and Samson 1995)

Emissions – Biogenic and anthropogenic evaporative VOCs

Emissions – Soil NO and electric demand

Page 80: Richard  T.  McNider Atmospheric Sciences  Department University of Alabama in Huntsville

Mixing Heights – Underestimate of mixing heights can cause an over-estimate of the sensitivity of controls. Emission reductions confined to a smaller volume cause a larger reduction in pollutants. A 30% error in mixing heights can produce 30% error in emission change impacts