Land surface spinup for episodic modeling
- 1Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, and NOAA Earth System Research Laboratory, Boulder, Colorado, USA
- 2CNRM/GMAP/PRO, Météo-France, Toulouse, France
- 3CNRM-GAME, Météo-France/CNRS, Toulouse, France
- 4Department of Applied Physics, Universitat Politècnica de Catalunya-BarcelonaTech & Institute for Space Studies of Catalonia (IEEC-UPC), Barcelona, Spain
Abstract. Soil moisture strongly controls the surface fluxes in mesoscale numerical models, and thereby influences the boundary layer structure. Proper initialization of soil moisture is therefore critical for faithful simulations. In many applications, such as air quality or process studies, the model is run for short, discrete periods (a day to a month). This paper describes one method for soil initialization in these cases – self-spinup. In self-spinup, the model is initialized with a coarse-resolution operational model or reanalysis output, and run for a month, cycling its own soil variables. This allows the soil variables to develop appropriate spatial variability, and may improve the actual values. The month (or other period) can be run more than once if needed.
The case shown is for the Boundary Layer Late Afternoon and Sunset Turbulence experiment, conducted in France in 2011. Self-spinup adds spatial variability, which improves the representation of soil moisture patterns around the experiment location, which is quite near the Pyrenees Mountains. The self-spinup also corrects a wet bias in the large-scale analysis. The overall result is a much-improved simulation of boundary layer structure, evaluated by comparison with soundings from the field site.
Self-spinup is not recommended as a substitute for multi-year spinup with an offline land data assimilation system in circumstances where the data sets required for such spinup are available at the required resolution. Self-spinup may fail if the modeled precipitation is poorly simulated. It is an expedient for cases when resources are not available to allow a better method to be used.