CHAPTER 1 Incremental
formulation of 3D/4D variational assimilation-an overview
CHAPTER 2 3D variational assimilation
CHAPTER 3 4D variational assimilation
CHAPTER 4 Background term
CHAPTER 5 Conventional observational
constraints
CHAPTER 6 Satellite observational
constraints
CHAPTER 7 Background, analysis
and forecast errors
CHAPTER 8 Gravity-wave control
CHAPTER 9 Data partitioning (OBSORT)
CHAPTER 10 Observation screening
CHAPTER 11 Analysis of snow
CHAPTER 12 Land surface analysis
CHAPTER 13 SST and sea-ice analysis
CHAPTER 14 Reduced-rank Kalman filter
REFERENCES
|
|
Next
Section
Previous Section
The snow analysis software is implemented as a branch of the more comprehensive
surface and screen-level analysis (SSA) package. The other branches currently
include two-metre temperature and relative humidity analysis, and also sea
surface temperature and sea-ice fraction analyses. The program organization
when performing snow analysis is roughly as follows:
The main program SSA calls CONTROL_SSA where most of the setup and namelist
handling are done. Routine INISNW performs initialization of the actual
snow analysis by sensing the size of the observation array file (CMA-file)
in SCAN_DDR and generating latitudinal coordinates that stem from the model
resolution in concern and zeros of the Bessel function.
After this, all input fields are read into memory in GETFIELDS. They consist
of the snow water equivalent and snow density from the first-guess (6-hour
forecast), 2 m temperature first guess, snow-depth climate (varies monthly
with a linear temporal interpolation), land/sea mask and finally the orography
in a form of the geopotential.
In SCAN_CMA observations are read into memory and a quick validity check
of the non-applicable observations for this analysis is performed. Furthermore,
the land/sea mask is calculated in LAND_OBS for the retained snow depth
observation points.
Additional screening is done in INITIAL_REJECTION and in REDUNDANT_OBS.
The former one sets up an internal table where all the observations which
survived from the quick screening are placed with a minimum context information.
This routine rejects some of the observations entered into the table due
to inconsistencies.
The routine REDUNDANT_OBS removes time duplicates and retains the observations
of the station in concern with the closest (and the most recent) to the
analysis time. Since only synoptic observations are considered, slowly moving
platform handling present in the REDUNDANT_OBS is not applicable to the
snow analysis.
The actual snow analysis is performed under SNOW_ANALYSIS. The analysis
technique is Cressman's successive correction method (routine SUCSNW). The
structure functions are set to be separable in horizontal and vertical directions.
A special mountain region handling is performed, depending whether the datum
or grid point is in the valley or at high altitudes, as explained before.
The snow-depth background (i.e. first guess) field is constructed from the
model first-guess snow water equivalent and snow density. Once the snow-depth
first guess field is present, it is used to calculate the first guess departure
at snow-depth observation points. This increment is finally added to the
snow depth fields at grid points producing the final snow depth output field,
which is output in routine FDB_OUTPUT.
The accuracy of the analysis is estimated in PRINT_SUMMARY where some important
statistics are summarized. The internal observation table can be printed
if requested from PLOTDATA and an updated observation file for feedback
purposes can be created in routine FEEDBACK.
The main logicals of the namelist NAMSSA are :
| |
• L_SNOW_ANALYSIS : When set to TRUE, the
snow analysis is performed. |
| |
• L_SNOW_DEPTH_ANA : When set to TRUE,
the snow analysis is performed in snow depth (in opposition to snow
water equivalent assuming a constant value of 250 kg m-2
for observed snow density). |
| |
• L_USE_SNOW_CLIMATE : When set to TRUE,
a relaxation of the snow analysis towards a monthly climatology is
performed with a time scale of 12.5 days (this constant is hard coded
in SNOW_FG). |
| |
• L_USE_FG_FIELD : When set to TRUE the
snow analysis is set to the first-guess value (no use of observations)
and there is no relaxation to climatology. |
Next Section
Previous Section
|
