Air Mass Analysis PGE12 SAFNWC     Table of contents 1.- Goal of the AMA product  2.- AMA algorithm summary description  3.- List of inputs for AMA  4.- Coverage and resolution 5.- Description of AMA outputs 6.- Example of AMA visualisation Access to the "Algorithm Theoretical Basis Document for Air Mass Analysis (AMA-PGE12) " for a more detailed description.

 

1.- Goal of the AMA product

The AMA products of the SAFNWC aim to support nowcasting applications by providing several products describing air mass characteristics. The sub-products are:

• A classification of air masses as function of temperature and stability information (derived from satellite measurements) and total precipitable water (derived from NWP).
•  An objective determination of the ridge lines of the equivalent-potential temperature field (indicating warm and moist air in which severe local convection is most likely to occur). This parameter is based on ECMWF model data at 850 hPa.
• An objective determination of regions with high gradients of equivalent-potential temperature (capturing pre-frontal areas where there is high probability for the initiation of convection). Based on ECMWF model data + cloud mask information.
• An objective determination of black stripes in water vapour imagery (dry air in the upper troposphere is an air mass pattern often linked to the initiation of convection).


2.- AMA algorithm summary description
 

• Air mass classification product
 

The implemented scheme is based on finding reference pixels where IR13.4 "almost certainly" indicates the correct air mass type. The motivation to use IR13.4 is that this is the only IR channel with a weighting function not peaking at the earth's surface, thus mitigating land heating effects. Then the WV7.3 brightness temperature at these pixels together with longitude and (often more importantly) latitude information is subject to multiple linear regression, with air mass type (categories "arctic" to "equatorial" numerically translated to values of a "fractional air mass code" from 1.0 to 4.0) as the dependent variable. Under the assumption that equatorial air masses are never adjacent to polar or arctic air and tropical and arctic air masses are always separated by the polar category, the problem becomes one of multifold multiple linear regression applied to the separation sub-problems:

• separation of arctic from polar,
• separation of polar from tropical,
• separation of tropical from equatorial.

After some releases based on the SAFNWC product "Total Precipitable Water" (TPW), v2008 switched over to a NWP-derived TPW. TPW thresholds between "continental" (dry) from "maritime" (moist) air masses are dependent on the numerical values of the functions obtained in the regressions: For “perfectly” arctic, polar, tropical and equatorial air, the thresholds are respectively 0.48, 1.13, 2.66 and 6.25 cm.  

Finally, a stability index imagery may be taken into account to distinguish between stable and unstable air masses:
 

Air mass type	Criterion
Stable	stable according to SAI
Unstable	unstable according to SAI

• WV dark stripe product 

  The detection of dark stripes has the following principal steps:

  • For each pixel, it is checked whether the pixel is darker than an empirical threshold.
  • For each candidate pixel, it is verified that it is darker than most of its neighbours.
  • The pixels passing the test undergo postprocessing: a contouring algorithm is applied to encircle sufficiently extended conglomerates of such pixels.
  
   
• `Ridge line' product
 

To determine the ridge line of the equivalent-potential temperature at 850 hPa, the following steps are carried out:

•  Computation of a "pseudo curvature vorticity" which describes the curvature of the NWP 850 hPa equivalent-potential temperature isolines (standard formulas used for deriving curvature vorticity from the geopotential height field are applied)
• For each point, determination of number of neighbours with higher values of "pseudo curvature vorticity" ("less anticyclonic" neighbours). This number is a measure of the qualification to lie on a ridge line.
• Tracking of lines connecting the most suitable points.

 
• `Gradient zone of equivalent-potential temperature' product
 

The main criteria to indicated a gradient zone are: sufficiently extended zone of gradient above an empirical threshold; either the gradient’s x-component or y-component has to have the same sign all over the indicated zone (otherwise two gradient zones would be joined which for geometrical reasons are unlikely to represent the same prefrontal region). The product is "cleaned" in such a manner that gradient zone markers are withdrawn in regions where the SAFNWC/MSG cloud mask reports clouds (in those areas, the product shows pixel values of an IR channel). Note that the “large-scale” check on the gradient zone comes first, i.e. due to cloud masking, some small patches of indicated gradient may appear in the final product.
 

3.- List of inputs for AMA

• .Air mass classification product
  • MSG radiance values:
    The following SEVIRI channels are required at full MSG resolution:
    • R7.3 micrometer: Mandatory.
    • R13.4 micrometer: Mandatory.
  • ECMWF NWP data:
    • Temperature and humidity information at 850 hPa: Mandatory.
  • SAFNWC/MSG products used as inputs:
    • PGE01(CMa): Mandatory.
    • PGE08(SAI): Optional.
• .WV dark stripe product
  • MSG radiance values:
     The following SEVIRI channels are required at full MSG resolution:
    • R6.2 micrometer: Mandatory.
• .`Ridge line' product
  • ECMWF NWP data:
    • Temperature and humidity information at 850 hPa: Mandatory.
• .`Gradient zone of equivalent-potential temperature' product
  • ECMWF NWP data:
    • Temperature and humidity information at 850 hPa: Mandatory.
  • SAFNWC/MSG products used as inputs:
    • PGE01(CMa): Mandatory.
  • MSG radiance values:
     The following SEVIRI channels are required at full MSG resolution:
    • R13.4 micrometer: Mandatory.

• Air mass classification

The product format is the HDF5 image format. The number of foreseen air mass codes in the current version is 9 or 17 (depending on whether the user turns the SAI inclusion option off or on) +8 placeholders for possible future use if a re-definition of air mass classes is considered beneficial and, for example, adds a definition such as "arctic+cloudy") with the following coding:
 
 

Code	Name
0	unclassified
1	Arctic, continental, stable
2	(Arctic, reserved for future use)
3	Arctic, maritime, stable
4	Polar, continental, stable
5	(Polar, reserved for future use)
6	Polar, maritime, stable
7	Tropical, continental, stable
8	(Tropical, reserved for future use)
9	Tropical, maritime, stable
10	Equatorial, continental, stable
11	(Equatorial, reserved for future use)
12	Equatorial, maritime, stable
13	Arctic, continental, unstable
14	(Arctic, reserved for future use)
15	Arctic, maritime, unstable
16	Polar, continental, unstable
17	(Polar, reserved for future use)
18	Polar, maritime, unstable
19	Tropical, continental, unstable
20	(Tropical, reserved for future use)
21	Tropical, maritime, unstable
22	Equatorial, continental, unstable
23	(Equatorial, reserved for future use)
24	Equatorial, maritime, unstable

There is moreover a quantity AMA_QUALITY, which represents a quality flag implemented in a first version in v2009. The present scheme checks for any pixel being cloudfree in both involved air mass classification products if there is a change in thermal category (i.e. arctic, polar, tropical or equatorial) and/or moisture category (maritime vs. continental) between the current and the previous slot. The definition of previous slot is fixed as: current slot minus 15 minutes. The concept is based on the understanding that despite air masses are of course moving – hence changes in air mass character must be expected to occur in every image -, air mass type is a quite conservative quantity (by definition, actually) and the number of changing pixels should be relatively small in such a short interval. A sudden increase for a certain slot should indicate a quality problem. The parameter AMA_QUALITY has a range of 0-4 with the following meanings:
 

Code	Name
0	no QC possible because pixel cloudy in at least one of the considered AMA_CL fields
1	no change in thermal category (i.e. arctic, polar, tropical or equatorial) and moisture category (maritime vs. continental) between current and previous slot
2	change in moisture category
3	change in thermal category
4	change in both moisture and thermal category

Note that this parameter may be missing in the HDF output file (namely if there is no precursor product, in which case the impossibility of a quality check is also verbally reported (in the log files)).

• `Gradient zone of equivalent-potential temperature' product

From v2008 on, the "gradient zone of the NWP equivalent-potential temperature field" product is an image-like product packed into the same HDF5 file (product code AMA_GZ).

The gradient zone product has 128 different count values with the following meaning:
 

Code	Name
0	cloudfree, no gradient zone identified
1	cloudfree, gradient zone identified
2 - 127	cloudy; information on gradient zone not relevant (since only of significance in prefrontal areas); IR greyscale image shown instead (taken from the 13.4 micron channel)

• WV dark stripe product and `Ridge line' product

These products are output in separate BUFR files. The graphical visualisations are to be done by the user. The detailed template of the BUFR records can be found in the Product User Manual for PGE12, section 2.3.

6.- Example of AMA visualisation

Figure 1 shows a visualisation of the BUFR sub-products: WV stripes, ridge lines of equivalent-potential temperature. Figure 2 gives an example how the air mass classification product may be presented to the user. Finally, Figure 3 holds a visualization of the gradient zone sub-product.

Click on thumbnail for full-size version of this image