The Permafrost Water Balance Model (P/WBM) was developed from the Water Balance Model (WBM)
(Vorosmarty et al., 1989) to incorporate a better representation of the seasonal effects from
frozen soil conditions found in the Arctic Regions (figure 1). P/WBM represents the soil zone in
two compartments, the rooting zone and a deep soil zone.  A dynamic active layer (Nelson and
Outcalt, 1987) prevents melt water from entering into the soil zone thus increasing river
discharge during the snow melt season.  Potential evapotranspiration was calculated using Hamon
(1963) which is based only on mean temperature. This PET function accounts for sublimation off
the snow pack and was shown by Vorosmarty et al. (1997) to have the least amount of bias of the
simple PET functions when using continental scale data sets. 

The model is driven by spatially distributed monthly time series of temperature and precipitation
(see Panel 1).  These data sets can be either long term averaged climatologies or multi-year
time series.  The major water balance variables of evapotranspiration, soil water storage, and
runoff, along with maximum active layer depth and permafrost extent, form the primary output
surfaces.  All are calculated for each 0.5 x 0.5 degree grid cell. 

Figure 2 shows observed annual discharge in km^3 per year.  This is an experimental
method which uses only the observed discharge field from the river discharge gauge database 
(Panel 2, figure 1b).  These gauges are used with the digital river network (STN-30) to 
distribute runoff to all grid cells in the interfluvial areas between the gauges.  The runoff 
in each grid cell is routed downstream to give the final observed discharge field.  The width 
of the rivers in figure 2 is a linear function of the local discharge along the river.  
Discharge values below 1.7 km^3 per year were not plotted and values greater than 
300 km^3 per year were assigned a constant width. 

A set of preliminary simulations have been tested for the Pan-Arctic region.  The input data
sets used were a 30 year monthly time-series of temperature and precipitation.  The
precipitation data (Hulme and Jenne, 1993) do not contain any corrections for gauge bias.  Figure
3 shows the results for the Ob River Basin at Salehard.  This is the most downstream basin in the
Ob.  Shown are the basin-wide average values of precipitation, simulated evapotranspiration and
observed and simulated runoff.  The largest interannual variability appears to be in the
Precipitation.  On average, simulated runoff is lower than observed runoff by approximately 14%. 

Current work is focused on finalizing the river discharge data set.  This will allow us to
improve the observed discharge field in figure 2.  The generation of discharge fields for the
simulated results is also required to allow direct visual comparison of the two data sets. 


Hamon, W.R. (1963) Computation of direct runoff amounts from storm rainfall.  International
    Association of Scientific Hydrology Publication, 63:52-62. 

Hulme, M. and R. Jenne (1993) A historical monthly precipitation data set for global land areas:
    applications for climate monitoring and climate model evaluation. Analysis Methods of
    Precipitation on a Global Scale, Report of a GEWEX Workshop, Koblenz, Germany, 14-17 Sept. 
    1992, WRCP-81, WMO/TD-No. 558, pp. A14-A17. 

Nelson, F.E. and S.I. Outcalt (1987) A computational method for prediction and regionalization of
    permafrost.  Arctic and Alpine Research, 19:279-288. 

Vorosmarty, C.J., B. Moore, M.P. Gildea, B. Peterson, J. Melillo, D. Kicklighter, J. Raich, E.
    Rastetter, and P. Steudler (1989) A continental-scale model of water balance and fluvial
    transport: Application to South America. Global Biogeochemical Cycles, 3:241-65. 

Vorosmarty, C.J., C.A. Federer and A. Schloss (1997) Potential evaporation functions compared on
    U.S. Watersheds: implications for global-scale water balance and terrestrial ecosystem 
    modeling, submitted to Ecological Application.