The present study aims to assess the spin-up time (i.e., the time required by a regional model to reach a dynamical equilibrium between its own physical dynamics and the lateral boundary information) for soil moisture, depending on soil initial conditions, over the Iberian Peninsula. To address this task, the Weather Research and Forecasting (WRF) model driven with ERA-Interim reanalysis data has been used to carry out dynamical downscaling simulations over this area. The simulations have been performed by considering two nested domains, the first one covering the EUR-44 EURO-CORDEX region with resolution of 0.44º (~50 km) and another one, finer, spanning the Iberian Peninsula with resolution of 10 km, approximately. Depending on the prescribed SM values used to initialize the WRF model, soil initial conditions can be divided into three different types: wet, dry and very dry conditions. The analysis has been focused on the 1990-2000 period and two different initial dates have been taken into account: 1990-01-01, 00:00 UTC and 1990-07-01, 00:00 UTC. Hence, a total of 6 simulations (3 soil initial conditions x 2 initial dates) have been examined. Additionally, a simulation initialized with SM data from ERA-Interim has been taken as reference.
Results obtained in this study show that drier soil initial conditions lead to longer spin-up times, as remarked by previous studies. Likewise, the deepest soil layers need more time to reach a dynamical equilibrium than the shallowest ones. Particularly, the Guadalquivir and Duero basins are among the regions which need the longest spin-up times. These results will be very helpful to evidence the added value provided by the model initialization with realistic SM data, especially for those applications in which accurate initial conditions are expected to increase the predictability of climate variability, such as decadal climate predictions.
Keywords: soil moisture, model initial conditions, spin-up, Weather Research and Forecasting model, Iberian Peninsula
ACKNOWLEDGEMENTS: This work has been financed by the project CGL2017-89836-R (MINECO-Spain, FEDER). Juan José Rosa-Cánovas was supported by the Spanish FPI Fellowship Programme (grant code: PRE2018-083921). The WRF simulations were performed in the Picasso Supercomputer at the University of Málaga, member of the Spanish Supercomputing Network.