wiki.met.no

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Flag Tests ======

^Results of testing flag handling for single missing point interpolations^

**Summary**

The following are results obtained by preliminary tests on a development machine. For all runs ftime=1 is set for a 
successful interpolation. The useflag settings are set by the kvalobs Qc1 library. The test cases are based on the first parts of the circulated document (QC2-d2_flagg_08_2010_v20.doc) as described further below (and now in the process of update after first feedback ... SEE DETAILS BELOW).

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000601000000000 |9899900000000000 | | 0000601100000000 | 5894700000000001 |
| 0000601000000000 |9899700000000000 | | 0000601100000000 | 5894700000000001 |
| 0000601000000000 |9897700000000000 | | 0000601100000000 | 5894700000000001 |
| 0000002000000000 |9839900000000000 | | 0000004100000000 | 6090900000000000 |
| 0000602000000000 |9839900000000000 | | 0000604100000000 | 5030700000000001 |
| 0000003000000000 |9899900000000000 | | 0000001100000000 | 6894900000000000 |



For the input control and useinfo it is not clear what all of the flag settings are likely to be. The above is based on one observed case "0000003000000000 9899900000000000" which is then further modified to include different fmis values and other flags ... as defined in the QC2-d2_flagg_08_2010_v20.doc specification. More work to be done here to identify improved test cases for trial on dev-vm101. 

**After feedback** Here is a test of the useinfo generation. The controlinfo is as above, the ftime=1 is set and then the chnage in usefinfo is displayed:
 

^ .. ^ .. ^ .. ^
|For this controlinfo produced by Qc2   |This useinfo changes...  |  ...to this value|
|0000601100000000 | 9894700000000000 | 5894700000000001 |
|0000002100000000 | 7039900000000000 | 6090900000000000 |
|0000002100000000 | 7939900000000000 | 6090900000000000 |
|0000002100000000 | 9039900000000000 | 6090900000000000 |
|0000602100000000 | 7033700000000000 | 5030700000000001 |
|0000602100000000 | 7939900000000000 | 5030700000000001 |
|0000003100000000 | 9899900000000000 | 6899900000000000 |

====== Details ======
<code>
Betingelse
QC2-d2 skal kun brukes for interpolering av manglende verdi og korrigering av forkastet verdi. Par = TA, TAN eller TAX. Par(t) mangler (for tidspunkt t). 
For disse tilfellene er:
a) fmis=1, fnum=6, useinfo(2)=9 (manglende verdi erstattet av modellverdi) eller
b) fmis=2, fnum=0 / 6, useinfo(2)=3 (forkastet verdi ikke erstattet / erstattet av modellverdi)
c) fmis=3, fnum=0, useinfo(2)=9 (manglende original- og corrected-verdi)

Foreløpig prosedyre for igangsetting av QC2-d2
1. Sjekker at Par(t-1) og Par(t+1) begge er kontrollert og funnet OK.
- useinfo(2)=0, uansett hvilke kontroller som er kjørt.

2. Iverksetter beregning av algoritme.
- Par(t) = 0.5* [Par(t-1) + Par(t+1)]
- ftime=1 (det skjer ingen endring av useinfo(2))
- for a) beholdes fmis=1 og useinfo(3), useinfo(4) endres fra 7 til 9,
     for b) settes fmis=4, useinfo(3) settes til 3 og useinfo(4) settes til 9.
     for c) settes fmis=1, useinfo(3) settes til 4 og useinfo(4) settes til 9.

</code>

  * a) fmis=1, fnum=6, useinfo(2)=9 (manglende verdi erstattet av modellverdi)
  * for a) beholdes fmis=1 og useinfo(3), useinfo(4) endres fra 7 til 9,

**Initial Tests:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000601000000000 |9899900000000000 | | 0000601100000000 | 5894700000000001 |
| 0000601000000000 |9899700000000000 | | 0000601100000000 | 5894700000000001 |
| 0000601000000000 |9897700000000000 | | 0000601100000000 | 5894700000000001 |

**After feedback ... this is the case to check:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000601000000000 |9894700000000000 | | ?? | ?? results probably as above ... |


  * b) fmis=2, fnum=0 / 6, useinfo(2)=3 (forkastet verdi ikke erstattet / erstattet av modellverdi)
  * for b) settes fmis=4, useinfo(3) settes til 3 og useinfo(4) settes til 9.

**Initial Tests:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000002000000000 |9839900000000000 | | 0000004100000000 | 6090900000000000 |
| 0000602000000000 |9839900000000000 | | 0000604100000000 | 5030700000000001 |

**After feedback ... this is the case to check:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000002000000000 |7039900000000000 | | ?? | ?? expect 50399 |
| 0000002000000000 |7x3xx00000000000 | | ?? | ?? expect 50399 |
| 0000002000000000 |9039900000000000 | | ?? | ?? expect 60339 |
| 0000602000000000 |7033700000000000 | | ?? | ?? expect 50339 |
| 0000602000000000 |7x3xx00000000000 | | ?? | ?? expect 50339|

  * c) fmis=3, fnum=0, useinfo(2)=9 (manglende original- og corrected-verdi)
  * for c) settes fmis=1, useinfo(3) settes til 4 og useinfo(4) settes til 9.

**Initial Tests:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000003000000000 |9899900000000000 | | 0000001100000000 | 6894900000000000 |

**After feedback ... this is the case to check:**

^ Before^| ... ^After^^
|   controlinfo    |     useinfo      |  |   controlinfo    |     useinfo   |
| 0000003000000000 |9899900000000000 | | 0000001100000000 | 689490000000000**1** |

**!** check for the useinfo(15)=1