HIRLAM (High Resolution Limited Area Model) is an operational NWP
model, whose data assimilation and parametrizations related to snow
and ice are being modified and tested in the conditions of winter
2009-2010. The new methods are shown to solve some old problems
related to excess cooling of ice-covered surfaces and to too cold
spring days in the model. If the model knows where on the sea and
lakes there is ice, it is able to produce realistic
forecasts. However, it is not easy to describe the ice extent when
only conventional (SYNOP) observations and climatological information
is available. Some problems and possible solutions are presented and
discussed in the presentation.