Near Real-Time SSTs in Mesoscale Prediction

1
Near Real-Time SSTs in Mesoscale Prediction Dr.
Steven M. Lazarus, Corey Calvert, Mike
Splitt Florida Institute of Technology
2
Regional weather is impacted by air-sea
interaction at varying scales
3

• Advancement of Coastal Science
• upwelling (e.g., event of summer 2003)
• ocean mixing
• land falling hurricanes
• wind and waves (coastal erosion, marine
  forecasts, etc.)
• riptides
• estuarine coastal ocean modeling
• clouds (freezes) and precipitation (coastal
  showers, deep convection)

4
Relevant Science/Operational Questions and Issues

• How do we best blend SSTs from multiple
  platforms?
• 2. Do we apply SSTs as only a boundary condition
  or do we nudge the boundary layer as well (i.e.,
  are the surface fluxes in balance)?
• 3. What are the impacts of high-resolution SSTs
  on short-term model forecasts, how do we best
  evaluate them, and do they match expectations
  (e.g., land vs. ocean)?
• 4. Are short-term forecast improvements (e.g.,
  precipitation) consistent with seasonal
  observations?
• 5. What direction are we heading (e.g., coupled
  ocean/atmosphere)?
• 6. Where can we get the most bang for our ?
• 7. Is the future skin or bulk SSTs?

5
What we are doing now

• Focusing on producing a high-resolution SST
  products that can be used in real-time we are
  trying to capture the best estimate (timely
  accurate) of the bulk SST.
• Evaluating whether or not the high-resolution
  SSTs enhance simulation of observed mesoscale and
  meso-gamma scale processes.
• Where things are headed
• SST product development contingent on model
  evaluations. Retrievals vs. direct use of
  radiances (e.g., Li and Derber, NCEP).
• Movement toward coupling hydrodynamic models with
  atmospheric models (e.g., NWS regional scale ARPS
  runs).
• Regional coupled models may be well-suited to
  applications such as simulating hurricanes and
  short term predictions of mesoscale features.

6

• Data assimilation/analysis from a coupled
  perspective
• Some of the relevant arguments are
• real world coupled thus initial conditions and
  forecasts should be
• performed in coupled mode.
• ocean models need improved surface fluxes.
• atmospheric models need improved SSTs.
• forecast errors are coupled and should be
  minimized in the initial
• conditions.
• The hope is
• we expect more accurate field information
  (data)/analyses.
• we get better initial conditions for
  forecasts/nowcasts.

7

• Does coupling matter at time scales on the order
  of a day, i.e. for short-term forecasts? Is the
  answer situationally dependent?
• Timely SST products will be needed (e.g., see
  World Climate Research Program Observation and
  Assimilation Panel Report) regardless of the
  modeling paradigm (dynamic, static coupling),
  but
• ocean community needs the SST profile in the
  water column.
• Fundamental difficulties persist w.r.t. numerical
  models in simulating processes associated with
  ocean and atmosphere boundary layers

8
EvaluationSSTs WRF
upper air WSR-88D QuikSCAT CloudSAT surface data
RTG-SST (NCEP)
???
MODIS COMPOSITES
WRF (24 h forecasts)
SST ANALYSES
?
BOUY
CLIMO
GOES
smooth
traditional/limited
quality?
How much crunching is needed to assess impact
of SSTs (e.g., seasonal modeling/forecast issues)?
9
Evaluation Efforts
Limitations of over-ocean data are problematic
from both the direct evaluation and
assimilation/modeling perspectives!
sparseness
multi-platform
timeliness
10
Chelton and Wentz BAMS 2005

• SSTs (color)
• wind stress (solid contours)

ECMWF (top) and NCEP (bottom) surface stress w/
Reynolds SSTs
ECMWF (top w/ RTG-SSTs) and NCEP (bottom)
surface stress (w/ Reynolds).
11
MODIS impact on WRF simulations
What is the impact on marine forecasts (i.e.,
Gulf stream convection, coastal showers,
upwelling dynamics/feedback w.r.t SSTs, etc.)?
WRF forecast differences (RTG MODIS) for 1.)
skin temperature (color contours w/ interval0.4
K) and 2.) 10 m wind magnitude (solid contours,
interval .2 ms-1). Also shown are the 10m winds
from WRF (w/ MODIS SSTs, barbs in ms-1) valid at
08 UTC 14 May 2004.
12
MODIS impact on WRF simulations
cooler
cooler
cooler
cooler
warmer
warmer
wind magnitude (ms-1)
RED contours
RH ()
boundary layer impacts (match expectations,
amplitude?)
13
Use of NASA scatterometer data
WRF 10 h forecast valid 1 May 2004. 10 m winds
(black barbs), QuikSCAT (red barbs), surface
observations (ship, buoy, METAR).
14
FIT SST Analysis Cycle
ARPS Forecast Cycle at NWS Melbourne
How best to match up the forecast cycle and SST
analyses?
15
FIT SST Analyses
RTG
MODIS


Analysis
16

• Parting Comments
• Future Efforts
• Near-term
• FIT analyses (glint, diurnal adjustment and
  latency issues)
• Composite Analysis Evaluation (non-modeling)
• Model Evaluation
• Long-term
• transition to operations
• communicating results to users (e.g., what is
  important for a forecaster/user to know about the
  products?)