A Synopsis and Comparison of
Selected Snowmelt Algorithms
RAE A. MELLOH
Models requiring only minimal meteorological
INTRODUCTION
data input rely on air temperature data to define
Estimating watershed runoff in areas with sea-
the surface energy balance, with precipitation
sonal snow cover requires a snowmelt algorithm
amounts sometimes used to further define cloudy
or clear weather. More-detailed surface energy
be part of the modeling system. In recent years,
balance models require air temperature, relative
advances in computing speed have made possible
humidity, wind speed, and either cloud cover or
the implementation of more detailed one-dimen-
radiation data. Internal snowpack processes may
sional snowpack models and distributed model
be simply ignored and the surface energy balance
techniques. A review of old and new algorithms
equated to snowmelt, or processes within a melt-
is needed to determine which model elements
ing snowpack may be considered in detail. Which
should be chosen for use in current modeling sys-
type of model is chosen for a particular basin or
tems, and what additional work is needed to
application will vary with data availability and
adapt these new tools to water resource uses.
the detail or type of model results needed.
Previous reviews of operational snowmelt run-
Historically, operational hydrological models
off models include a comparison of models by the
have been of limited detail both in the surface
World Meteorological Organization (WMO)
energy balance and in the internal snowpack pro-
(1986), the U.S. Army Corps of Engineers (1998),
cesses. All of the models reviewed were devel-
and a book on computer models of watershed
oped initially from snowmelt studies at a point in
hydrology (Singh 1995). These studies compared
snow laboratories or research watersheds. The
the structure and approach of snowmelt models,
point models were then extended to lumped basin
but did not attempt to judge whether one model
scales, or more recently to hydrologic response
performed better than another. The WMO com-
units (Leavesley 1983) by using coefficients var-
parison did not separate snowmelt algorithms
ied to capture the impact of winds or solar expo-
from the hydrograph transformation and routing
sure, as dictated by landscape position or vegeta-
aspects of the models, but looked at the net result
tive cover. With availability of fast computers, and
rather than the performance of individual parts
digital elevation and land cover data, there is now
of their modeling systems. In this review we look
ability to describe the surface energy balance in
specifically at the snowmelt algorithms used in
some detail across a landscape. This ability has led
current operational model systems in the United
to the development of distributed snowmelt mod-
States: their developmental setting, equations,
els. Today's computational efficiency also allows
strengths, and limitations.
simulation of internal snowpack processes that
All snowmelt algorithms consist of a surface
energy balance and most also include some account-
have been mostly ignored in hydrologic models.
ing of internal snowpack processes. These two
A preliminary hydrologic model that accounts for
parts of snowmelt algorithms follow simple to
both a varied surface energy balance across a
detailed approaches in the models reviewed here.
landscape and details within a layered snowpack
Back to contents
Previous Page
