Table 1. The validation sites characteristics
and the corresponding seasons
Air Pressure (hPa)
Col de Porte (France)
Goose Bay (Canada)
Sleepers River (USA)
Hourly short- and long-wave solar radiation, air temperature,
precipitation, precipitation phase, wind speed, and relative humidity were
provided for each site for selected seasons (see Table). Each
participant should produce simulation results including snow water equivalent,
snow depth, and other variables (e.g. surface temperature, albedo, liquid
water content, etc.) depending on a model complexity. Observed snow
water equivalent and snow depth data were not released before simulation
outputs were submitted. It did not allow any calibration of model
parameters. Some additional information on SnowMIP project can be found
More than 20 snow models from 10 countries including 7
from USA, 4 from France, 3 from Japan, etc. were submitted to the project,
see participants list in Appendix I.
Participating models represented a wide range of snow physics complexities
including 1) simple temperature index-based models, 2) one layer
energy-based models, and 3) most sophisticated multilayer energy-based
HL submitted results from two models of different complexity:
1) SNOW-17, which is a temperature index-based model with a simplified
treatment of an energy balance transfer; 2) NOAH snow-frozen ground parameterization
that is a one layer energy-based model. SNOW-17 requires only air
temperature, precipitation, and precipitation phase (if available) while
NOAH uses solar radiation, wind speed, humidity, air pressure in addition.
Because there was no calibration, most SNOW-17 model parameters were assigned
values recommended by the NWSRFS User's manual. A parameter UADJ
(rain-on-snow melt events) was estimated as a ratio of 6hr wind travel
during snowmelt for each site. Calibrated early parameter MFMAX for the
Sleeper River was slightly adjusted based on the snowmelt period for each
site (it was increased/decreased by ratio of increase/decrease of daylight
time for different site).
The project provides an opportunity to compare non-calibrated
index-based parameterization to more sophisticated energy-based models,
and to define most common causes of index-based model deficiencies.
At this stage, observed data are still not available to make a comprehensive
analysis of simulation results. However, during the IAMAS Conference
in Innsbruck, Austria in July 10-13, 2001, the organize committee presented
preliminary results from all participated models. A range of simulated
snow depth and water equivalent was unexpectedly very wide from close to
zeros for most of cold season to unrealistically high values. Figures
1 through 5 display observed and simulated from all participated models
snow water equivalent, and Figures 6 through
10 display snow depth simulation results. Most important finding
was that there was no dependency of simulation accuracy on the model complexity.
Some simple index-based models reproduced snow cover dynamics better than
some of the most complex models. Results from HL submitted index-based
and energy-based models agreed with this general conclusion. Both
models reproduced snow water equivalent and depth reasonably close.
However, these results suggest that under some weather conditions the energy-based
model works better if radiation data are high quality.
Our plan is to perform more comprehensive analysis of
SNOW-17 and NOAH simulations after data will be released. Effects
of weather conditions on SNOW-17 simulation results will be investigated.
Link. When you click on any of the following links you will be leaving
the HL Web site. You may wish to review the privacy notice on those
sites since their information may differ from that of OHD and NOAA.
In addition, our linking to this site does not constitute an endorsement
of any products, services or the information found on the that site.