Sensitivity of Climate Simulations to Land-surface Complexity: Beginning AMIP Diagnostic Subproject No. 12

About 40 world AGCMs are participating in Phase II of the Atmospheric Model Intercomparison Project (AMIP II). The Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) is responsible for AMIP Diagnostic Subproject 12 on Land-surface Processes and Parameterizations (DSP12). In the absence of high quality global data sets, DSP12 relies heavily on possibly less reliable model-derived estimates of land-surface variables such as those provided by various reanalyses for model evaluation. Having this caveat in mind, four reanalysis products, NCEP-DOE, NCEP-NCAR, ECMWF and VIC, will be used as validation data sets in DSP 12 to try to account for biases.

The analysis of the AMIP II land-surface simulation are performed globally and regionally. Analyses are reported for GEWEX-CSE (Continental Scale Experiments) regions and for different climate zones, defined by the de Martonne aridity index: I=P / (T+10), where P is mean annual precipitation in mm and T is mean air temperature in degree C. Results from the first phase of AMIP (AMIP I) showed that increased land surface scheme (LSS) complexity alters surface energy partitioning. Given the wider range of LSS employed in AMIP II, DSP12 aims to analyse the surface energy and water budgets as a function of LSS complexity and as a function of process parameterisation.

In this paper surface fluxes and other surface variables derived from NCEP-DOE, NCEP-NCAR and VIC are presented, in order to illustrate the analysis approaches employed in DSP12.The analysis so far has revealed that:

· Large-scale variations of LH are captured by all reanalyses considered. However, there are considerable differences among the four reanalysis products at the regional scale and over different de Martonne climate zones.

· Compared to NCEP-DOE the global mean LH over the land surfaces is underestimated greatly by VIC, mainly due lower LH in wetter climates, and slightly by NCEP-NCAR, due to underestimation in drier climates.

· NCEP-NCAR has a larger and VIC has a smaller inter-climate variability of mean LH than NCEP-DOE. The intra-climate variability of the both cases is smaller than that of NCEP-DOE.

· The spatio-temporal correlation coefficient of estimated LH by NCEP-NCAR and VIC relative to NCEP-DOE is small in drier climates and increases towards the wetter climate zones.