6. Conclusions

The Global Soil Wetness Project (GSWP) is an ongoing modeling activity of the International Satellite Land-Surface Climatology Project (ISLSCP), a contributing project of the Global Energy and Water Cycle Experiment (GEWEX). The GSWP is charged with producing a 2-year global data set of soil moisture, temperature, runoff, and surface fluxes by integrating one-way uncoupled land surface process models (LSPs) using externally specified surface forcings and standardized soil and vegetation distributions, namely, the ISLSCP Initiative I CD-ROM data (Sellers et al. 1996a). Approximately one dozen participating LSP groups in five nations have taken the common ISLSCP forcing data to execute their state-of-the-art models over the 1987-1988 period to generate global data sets. See IGPO (1995) or Dirmeyer et al. (1999) for further details.

The motivation for GSWP stems from the paradox that soil wetness is an important component of the global energy and water balance, but it is unknown over most of the globe. Soil wetness is the reservoir for the land surface hydrologic cycle, it is a boundary condition for atmosphere, it controls the partitioning of land surface heat fluxes, affects the status of overlying vegetation, and modulates the thermal properties of the soil. Knowledge of the state of soil moisture is essential for climate predictability on seasonal-annual time scales. However, soil moisture is difficult to measure in situ, remote sensing techniques are only partially effective, and few long-term climatologies of any kind exist.

The goals of GSWP are fourfold. The project will produce state-of-the-art global data sets of soil moisture, surface fluxes, and related hydrologic quantities. It is a means of testing and developing large-scale validation techniques over land. It serves as a large-scale validation and quality check of the ISLSCP Initiative I data sets. GSWP is also a global comparison of a number of LSPs, and includes a series of sensitivity studies of specific parameterizations which should aid future model development.

The GSWP consists of three components: the Production Group, the Validation Group, and the Inter-Comparison Center. The Production Group consists of land surface modelers who conduct offline integrations of land surface models over a global 1 grid for 1987-1988 using prescribed atmospheric forcing based on observations, remote sensing and analyses. Each member of the production group produces global time-mean and instantaneous fields of surface energy and water balance terms three times per month using his/her model. These data are produced in a standard format and sent to the Inter-Comparison Center. In addition, each model is used to perform specific sensitivity studies. The sensitivity experiments are intended to evaluate the impact of uncertainties in model parameters and forcing fields on simulation of the surface water and energy balances.

The inter-comparison effort has shown that there is a large spread among the participating LSPs in terms of their partitioning of surface energy between latent and sensible heat flux, and of water between runoff and evapotranspiration. Most of the LSPs underestimated basin-scale runoff, possibly due to the GSWP specification of the treatment of convective precipitation. Nonetheless, validation of the consensus runoff against streamflow data show that the LSPs as a group perform quite well where sufficient gauge-based precipitation forcing data were available, and performed poorly where gauges are sparse.

A number of different sensitivity studies were conducted by members of the Production Group. Perhaps the most significant general conclusion that can be drawn from the studies is that sub-grid scale variability in infiltration, whether due to heterogeneity in soil properties or the distribution of rainfall within a grid box, has a significant impact on the simulation of runoff. Variations in vegetation properties, the vertical structure of the soil, and radiation seem to have less of an impact on simulations. These results suggest that some sort of accounting for sub-grid heterogeneity, whether through an explicit modeling of small tiles, or a statistical approach, is necessary to properly partition surface water between runoff and evapotranspiration.

Meteorological and land surface data are becoming available at an ever increasing resolution and of ever increasing quality. New high resolution data sets are being released almost continuously. The ISLSCP Initiative II products will have a higher resolution and accuracy than those used in the GSWP. They will furthermore cover a longer period, making possible better studies at the interannual timescale. The validation of the input data has shown that a number of critical parameters, such as rooting depth, are not adequately represented in the CD-ROM. Use of aggregation techniques allow a better representation of soil hydraulic parameters at the climate model grid scale. In the long term, as the grid length of these models becomes smaller, the need for aggregation may become obsolete as the length scale of landscape and soil type variation, becomes comparable to the model grid lengths.

The comparison of the LSP simulations against point data showed the same worryingly poor and inconsistent performance as observed in the PILPS project. This raises the question of whether the model logic applied in these models is adequate, or whether a given model formulation can be appropriate at both point and grid scales. There is no immediate reason to suspect that all the estimates of the GSWP would agree with observations at a single location, but it would give credibility to the estimates if general patterns were satisfactory, with a small residual error. This unfortunately is not the case, and more research into the quality of the forcing data would appear to be needed before any form conclusions can be drawn.

The improvement in our ability to compare runoff estimates with observations through routing schemes cannot be underestimated. Apart from the model results of GSWP, this is a major step forward in our methodology to test large scale land surface hydrology models. It is worrying that the data on runoff is hard to access, and that the number of observation points is steadily declining since the late eighties. This is an area where the international programs, such as the GEWEX Hydrometeorology Panel (GHP) and IGBP together with GEWEX should take action immediately.

Results of this pilot phase suggest that the GSWP framework is very useful and valuable for assessing and developing land surface models on a global scale with relatively little computational expense, and to investigate questions of land surface hydrology and land-atmosphere interaction.