1. Characterizing the response of simulated atmospheric boundary layers to stochastic cloud radiative forcing
Robert Tardif, Josh Hacker
(NCAR Research Applications Lab.)
Howard Bondell, Justin Shows
Montserrat Fuentes and Brian Reich
(SAMSI)

2. Outline Part 1 (NCAR): Part 2 (SAMSI):
Scientific problem – link between cloud & boundary layer variability
General approach – ensemble simulations of idealized boundary layers
Toolbox
Numerical model (single column model)
Observations (cloud properties)
Statistical models
Part 2 (SAMSI):
Evolving cloud properties … a statistical perspective
Stochastic model design
From the simplified to the comprehensive

3. Scientific problem - concepts
Atmospheric boundary layer
Definition: “Part of the atmosphere directly influenced by the presence of the surface over “short” time scales”
turbulent fluxes
to/from
atmosphere
turbulence
vertical transport
of mean quantities
e.g. momentum
roughness
net energy
water
Diurnal cycle
(day)
(night)
Height
free atmosphere
residual layer
mixed layer
entrainment zone capping inversion
stable boundary layer

4. Scientific problem - concepts
Clouds & boundary layer variability
Clouds interact with radiation
Reflection/absorption of sunlight (shortwave radiation) - daytime
Emission of longwave radiation – nighttime
Height
free atmosphere
residual layer
mixed
layer
stable boundary layer
variability in boundary layer
diurnal evolution
→ modulation of surface radiation (net energy)
→ modification of boundary layer turbulent energy transfer

5. Scientific problem - concepts
Clouds show considerable variability
Present or not
Height
Optical properties (dependent on macro- and microphysical cloud structure)
Large uncertainties on cloud parameters in numerical weather forecasts
Clouds greatly influenced by small(subgrid)-scale processes
Not completely accessible from observations
Not fully resolved in models
stochastic cloud variability boundary layer variability

6. Scientific problem - goals
Theme: Investigate the factors that limit the skill of current boundary layer predictions
Questions:
How strong is the link between cloud variability & variability in simulated boundary layer structure (focus on wind)?
Any enhanced sensitivity to timing of cloudy periods? (w.r.t. boundary layer transitions)
Any dependence on time scales characterizing cloud variability?
Better understanding of variability in low-level jet characteristics
Step toward characterization of model errors
Reproduced from Hoxit (1975)
Wind speed

7. Scientific problem - strategy
How to characterize link between cloud & boundary layer variability?
Complex system (non-linear interactions)
Going beyond simple sensitivity experiments
Numerical modeling / ensemble simulations
Single-column model (SCM) w/ WRF parameterizations
Surface-atmosphere interactions
Boundary layer turbulence
Radiative transfer
radiation
turbulence
surface-atmosphere
interactions

8. Scientific problem - strategy
Numerical modeling / ensemble simulations
Ensembles
Members defined by realizations of stochastic model
Wide spectrum of cloud variability
Statistical characterization of output (sensitivity)
Work within the Data Assimilation Research Testbed
Ensemble simulation capabilities
Analysis tools
Poised for data assimilation experiments

9. Scientific problem - strategy
Cloud variability defined with:
Cloud presence
Cloud height
Liquid water path (LWP) z ql zt zb
cloud
water
Obs. clouds
cloud base
height
LWP
Stochastic model
Single column boundary layer model
Radiation
Wind
solar
thermal

10. Dataset (stochastic model development)
Observations from DOE ARM/SGP/CART in Oklahoma
Integrated ground-based instrumentation for measuring cloud properties
State-of-the-art instrumentation
Ceilometer (cloud detection, cloud base height)
Microwave radiometer or MWR (liquid water path)
Complemented by co-located:
Upper-air soundings
Cloud radar (MMCR)
Surface obs. (T, Td, precipitation)
soundings
radiometer
ceilometer
cloud radar
sfc. obs.

11. Dataset Most “up-to-date” MWR dataset (Turner et al.,2007)
From April 2001 to December 2005
Temporal resolution: 20 sec.
Focus on “fair-weather” days
absence of precip. & deep convective clouds
Divide dataset into:
“coupled” (to surface) clouds (within boundary layer)
“de-coupled” (from surface) (free atmospheric) clouds
cloud base
height
LWP
“coupled” Zi
LCL
night
“de-coupled”

12. Experiments Start simple(“er”)… “de-coupled clouds” (no feedback)
How is diurnal cycle of BL structure (temperature, wind) affected by:
Cloud times w.r.t. “key times” in BL evolution
→ (morning transition, afternoon, evening transition …)
Characteristic of cloud field (duration “ℓ” of cloudy “periods”)
Cloud height variability - within cloudy period (nighttime)
Cloud thickness variability - within cloudy period (daytime)
Various combinations thereof…
Large “parameter space”!
Need large ensembles
Feasible with SCM approach ℓ
sunrise
sunset
height

13. Experiments LWP ~ 40 kg m-2 solar radiative flux @ surface day night
spin-up
LWP ~ 40 kg m-2

14. Experiments
Sensitivity to cloud presence!

15. Now to Howard …

16. What’s next … Implement ARMA stochastic models into SCM
Validate implementation w/ comparison of time series generated w/ SAMSI and NCAR codes
Run ensembles of SCM simulations w/ stochastic uncoupled clouds
Analyze results
Statistical properties of ensemble output
Characterize relationship to cloud statistics
Link results to boundary layer predictability
Go to more comprehensive stochastic models
Impact?
Coupled clouds…