Land surface parameterisations

In the last few decades, there have been dramatic developments in models of the atmospheric circulation, driven by the needs for weather forecasting and predicting climate change. Hydrology is important in these models; it controls the latent and sensible heat fluxes that are needed as the lower boundary condition for the atmosphere over land surfaces. Early representations of hydrology in these land surface parameterisations (LSPs) were limited by computational constraints (and still are, even with implementations of global and regional circulation models on current supercomputers). They were therefore of similar form to the very simple soil moisture deficit models outlined in the previous section. The advantages of such an approach were that few parameters were required when applied globally and the low computational requirements. The disadvantages were that such simple models did not properly reflect understanding of the physics. This became particular apparent in higher latitudes where it was necessary to implement multiple layers to allow for heat transfer, storage and freezing in the soil.

Thus the next generation of LSPs were still one dimensional but started to incorporate more information about the soil and vegetation, and more explicit calculation of soil water transfers in multiple soil layers, including root density profiles to allow for root extraction with depth. Evapotranspiration calculations are often based on Penman-Monteith formulations but multiple vegetation layers have also been incorporated in some LSPs with resistance-based calculations of fluxes between layers. A typical example of a LSP is the MOSES scheme used in the UK Met Office Unified Model of the atmosphere (Fig. 10.13). The second-generation MOSES scheme allows for multiple independent tiles of different vegetation/land-use types making up fractions of any grid square of the atmospheric model. The fluxes calculated for each tile are integrated to give the total boundary fluxes for that grid square. The same approach is used in the latest development of MOSES, the Joint UK Land Environment Simulator (JULES4), which allows nine different surface types: broadleaf trees, needleleaf trees, C3 (temperate) grass, C4 (tropical) grass, shrubs, urban, inland water, bare soil and ice.

As well as being used as an LSP for an atmospheric model, MOSES and JULES can also be used as stand-alone representations of the surface hydrology and will replace MORECS as a way of calculating the patterns of soil moisture status across the UK. As an example of the use of this type of land surface model, Smith et al. (1994) have combined MOSES with high resolution analyses of precipitation, cloud and near-surface atmospheric variables (available from the Met Office NIMROD product), and the Probability Distributed (PDM) runoff model (see Chapter 12) to provide hourly updates of soil moisture, snowmelt, runoff and evaporation. Some results of this model in time and space are shown in Figs 10.14 and 10.15.

Fig. 10.13 A representation of the MOSES land surface parameterisation scheme (after Cox et al., 1999, with kind permission of Springer Science and Business Media).
0 0

Post a comment