Snow analysis in HIRLAM and HARMONIE

The current version of HIRLAM's snow analysis was implemented by
Cansado (2004) with references to Brasnett (1999) and is running
operationally in many countries. CANARI (Taillefer, 2002) is used for
surface analysis in HARMONIE. The CANARI snow analysis in HARMONIE has
been tested and is under development.

The main features of the snow analysis as introduced in CANARI and
HIRLAM are similar

- analysis method is Optimum Interpolation (OI)

- background error correlation includes a horizontal and a vertical term

- use only synoptic observations which are supposed to have uncorrelated error

The performance of the schemes depends not only on available
observations, but also on the specification of the statistical
parameters. To prepare for the testing of CANARI-F¡s snow analysis, the-A
performance of HIRLAM-F¢s snow analysis were examined carefully and-A
revealed that too much snow were accumulated during the winter and
that the snow disappears too late in the spring. One of the
consequences is that the forecasted 2m temperatures become too
low. Experiments to improve the performance of HIRLAM-F¢s snow analysis-A
were performed. Minor changes of parameters related to quality
control, influence radius and melting rate improved the performance
considerably. Results from HIRLAM experiments are useful as a
reference for the HARMONIE experiments.