Multi-model ensemble forecasting has been shown to offer a systematic improvement in the skill of climate prediction with atmosphere and ocean circulation models. However, little of such work has been done for the land surface component, an important boundary layer for weather and climate forecast models. In this study, several methods of combining individual global soil wetness products from uncoupled land surface model calculations and coupled land-atmosphere model reanalyses to produce an ensemble forecast are examined and evaluated. Forecasts are verified against observations from the Global Soil Moisture Data Bank with forecast skill measured by correlation coefficient and root mean square error. A preliminary transferability study is conducted as well for transferring ensemble forecast parameters between areas with similar climate regime. These methods are then applied and tested for an ensemble estimate of land surface fluxes and state variables derived from the Second Global Soil Wetness Project (GSWP-2) integrations performed by a dozen of different Land Surface Scheme (LSS) in an offline mode. Through validation with in situ measurements available over the 10-year GSWP-2 period, it is shown that diverse renditions of single analysis compiled from an ensemble of different LSSs usually outperform individual members in terms of metrics of skill employed in this study. Such an exercise not only helps us better understand the virtues and limitations of various multi-model ensembling techniques, but also enables progress toward creating an optimum, model-independent analysis. The transferability study also provides useful insights on achieving improved land surface estimates over those areas where observations are not available.