Effects of short spin-up periods on soil moisture simulation and the causes over New Zealand

A spin‐up is needed for a land surface model (LSM) to simulate reliable soil moisture. The best spin‐up is to integrate a LSM for many consecutive years before the start of an experimental period. Alternatively, a one‐year spin‐up period (defined in the text) can be used to spin‐up a LSM if longer periods of forcing data are not available. However, for special cases with only a few months of observations, can one use a few months to spin‐up a LSM? If yes, what are the conditions? To answer these questions, we have run a LSM during 1 August 2008–31 July 2010 at 4 sites in New Zealand with widely different annual rainfall amount and soil textures. Two test groups were conducted at each station. Each group had 6 spin‐up tests differing only in spin‐up periods, ranging from 2 months to 1 year (CTRL). For the 4 stations, a few months had the same effect as CTRL except at Winchmore, where significant differences in simulated soil moisture between a test and CTRL still occurred at or after the equilibrium state. The results at Winchmore repeated at the other 3 stations with reduced rainfall rates, and vice versa. Sensitivity tests with decreased or increased rainfall rates showed that quite large differences in simulated soil moisture at the equilibrium states between CTRL and a test occurred in an annual rainfall window from 350–1000 mm. Further analyses suggested a E/R ratio (the ratio of evaporation accumulation to rainfall accumulation during the spin‐up period at the last spin‐up cycle) to determine whether or not a few months can be used to spin‐up a LSM. A simple rule was suggested for application of these results in New Zealand and other areas with almost similar climates to those of New Zealand. Key Points A few months may or may not be used to spin up a LSM instead of one year We have investigated the conditions when and the reasons why this can be done We suggested the dominant control parameter: E/R ratios