Documentation of the changes from version 1 to version 2

Correction of errors and extensions

• The maximum rate of Snow Fall (Snowf) was increased from 0.002 to 0.0085 kg/m²/s. This change is based on observed values in the Rhône basin.
• The maximum depth of frozen soil and thawing is upgraded to 5 meters. The previous value of 2 meters was too small.
• The ALMA standard now allows that only the wind speed is provided. As it is straightforward to implement this flexibility in you code we have provided an example
• The name and definition of the variable DelSoilHeat were misleading. This variable is now labeled DelSurfHeat and provides the change in energy content of the snow free area. This should not only contain the energy change in the soil but also take into account the energy in the canopy if available in the scheme. It is now the true complementary of the DelColdCont. Beware that if you have frozen water in your soil you need to include the energy of the phase change as well in the computation of DelSoilHeat.
• The units of DelSurfHeat and DelColdCont were incompatible with the "accumulated" instruction. The units are now corrected to J/m². Another solution would have been to replace the "accumulated" instruction by "averaged" but this would have needed changes for DelSoilMoist, DelSWE, DelSurfStore and DelIntercept for consistency.
• In order to ensure that the total amount of water present in the soil is kept in only one variable two changes occurred in table O.4 . The variable SoilMoist is now the total mass of water present in the soil in the three phases. To separate the phases two fractions were introduced : The fraction of the moisture mass which is liquid (SMLiqFrac) and the fraction of the mass in the solid phase ( SMFrozFrac). The units of these two variables are "kg/m²/kg/m²" but for simplicity we will consider this variable to be unit-less
• Potential evapotranspiration has been introduced. The aim is to obtain the potential evaporation which is coherent with the way the scheme computes evapotranspiration. That is using the same formulation as for the evapotranspiration but setting all resistances, except the aerodynamic resistance, to zero. In most scheme this will not yield a Penman-Monteith potential evapotranspiration but exactly this discrepancy will be one of the possible diagnostics.
• the description of the averaging process to be used for the surface temperatures was improved.
• Table O.6 (renamed for the occasion) includes now the depth of the water table as an optional variable for the model which simulate it.
• A table is created for variables to be validated with remote sensed data.

Introduction of diagnostics related to the liquid water content in the snow pack

• Evaporation of the liquid water in the snow pack : EvapSnow
• To make sure that nothing is left out, the sublimation of the snow free area has been added. This is usually a small flux and the range for the quality control is still largely overestimated. ( SubSurf )
• When liquid water is included in the snow pack the computation of DelColdCont needs to take into account the phase change.
• A new variable had to be added for the flux of water flowing out of the snow-pack : Qst . If the snow model does not include liquid water this flux will be equal to snow melt flux. Else the difference between Qst and the snow melt will provide an information on the ability of the snow pack to retain liquid water.
• Snow melt (Qsm) has been completed with the re-freezing flux (Qfz). It is the amount of liquid water which re-freezes in the snow pack. It can not be a negative Snow melt as both phase changes can occur at the same time in the snow pack.
• The fraction of liquid water (SliqFrac) in the snow mass has been introduced as a new diagnostic. This does not change the definition of the snow mass (SWE) , which remains the total water in the snow pack, but allows to separate it into the liquid and frozen mass. The units are "kg/m²/kg/m²" but for simplicity we will consider this variable to be unit-less. For a multi-layer scheme this will obviously be a 3D variable.
• As the distribution of rainfall and snow fall on the snow covered and snow free fractions of the grid may change from one model to the other, this information needs to be saved. Two new variables have been introduced to record this : RainfSnowFrac, SnowfSnowFrac .

Documenting the 3D nature of the snow pack

• Now that a 3D description of the snow pack is possible the variable SnowT needs to be redefined . It will only include the snow temperature which interacts with the atmosphere. It can be more complex than just the first layer temperature of a multi-layer snow scheme. The temperature within the snow can now be described with the variable SnowTProf . For a simple one layer snow scheme this new definition of SnowT variable does not change anything. SnowTProf will contain the same values as SnowT.
• SWE becomes a 3D variable. Obviously for simple snow schemes all remains as it was.
• SnowDepth and SliqFrac are now 3D variables. In the case of simple one-layer schemes the vertical dimension becomes a singleton. The sum over the 3rd dimension of SnowDepth will yield the total depth of the snow pack.
• This table now also includes the vertical profile of temperature in the snow pack. In the most basic scheme with only one snow temperature this variable will contain the same values as SnowT.
• Variables DelColdCont and DelSWE remain 2D. They are intended for verification of conservation laws and not for process studies with the data. It is believed that a verification of the energy and water conservation of the integrated snow pack will be sufficient for the years to come.