We estimated the snow water equivalent (SWE) of snowpack in central Japan from September 2006 to August 2008 by using a 3.3 km-mesh regional climate model with two land-surface models: Noah land-surface model (Noah LSM), and Noah land-surface model with multiparameterization options (Noah MP). The model validation for temporal variations of SWE at the Tohkamachi station and the comparison of modeled maximum SWE with estimated that from observed maximum snow depth at ten sites showed that Noah MP could simulate spatiotemporal variations of SWE better than Noah LSM which underestimated SWE. Simulated SWE in central Japan peaked in March, but the difference of SWE between the two land-surface models was greatest in April. SWE determined using Noah LSM (Noah MP) in analysis domain reached 18.1% (28.5%) of the total storage capacity of high dams in Japan in March 2007, whereas it reached 32.4% (44.1%) in March 2008. The difference of SWE between the two land-surface models was particularly high under warm conditions, that is, during the snowmelt season, and during a warmer than normal winter. Our results indicate that the choice of land-surface model for estimates of SWE is important under warm climatic conditions.