Hard-coded parameters have the largest impact on fluxes of the land surface model Noah-MP
Abstract:Land surface models incorporate a large number of processes, described by physical and empirical equations. The agility
of the models to react to different meteorological conditions is artificially constrained by having hard-coded
parameters in their equations.
The land surface model Noah with multiple process options (Noah-MP) is one of the standard land surface schemes in WRF
and gives the flexibility to experiment with several model parameterizations of biophysical and hydrological
processes. The model has around 80 parameters per plant functional type or soil class, which are given in tabulated
form and which can be adjusted. Here we looked into the model code in considerable detail and found another 140
hard-coded values in all parameterizations, called hidden parameters here, of which around 50-60 are active in
specific combinations of the process options.
We quantify global parametric sensitivities (SI) for the traditional and the hidden parameters for five model outputs
in 12 MOPEX catchments of very different local hydro-meteorologies. Outputs are photosynthesis, transpiration,
latent heat, surface and underground runoff.
Photosynthesis is mostly sensitive to parameters describing plant physiology. Its second largest SI is for a hidden
parameter that partitions incoming into direct and diffuse radiation. Transpiration shows very similar SI as
photosynthesis. The SI of latent heat are, however, very different to transpiration. Its largest SI is observed for a
hidden parameter in the formulation of soil surface resistance, due to low transpiration in Noah-MP. Surface runoff is
mostly sensitive to soil and infiltration parameters. But it is also sensitive to almost all hidden snow parameters,
which are about 40% of all hidden parameters. The largest SI of surface runoff is to the albedo of fresh snow and the
second largest to the thermal conductivity of snow. Sensitive parameters for underground runoff, finally, are a mixture
of those of latent heat and surface runoff.
In summary, five of the six largest sensitivities in Noah-MP belong to hidden parameters. All other hidden parameters
have little effect on output variability. There are about 40 sensitive, traditional parameters in comparison. We
therefore recommend to include these five hidden parameters in any model assessment or calibration study.