As part of the Project for the Intercomparison of Land-surface Parameterisation Schemes (PILPS), Phase 4b investigates the impact of several land-surface parameterisation schemes when coupled to the same atmospheric host mesoscale numerical weather prediction (NWP) model. The aim of Phase 4b is to investigate the importance of land-surface-atmospheric feedbacks on the short-term forecast (up to 48-h) and to detail the extent to which the complexity of a land- surface parameterisation scheme may affect the accuracy of the forecast.
This study concentrates on a case study (6- 7th March 1996) over the Australian region. During this 48-h period significant rainfall occurred over northern Australia associated with the Australian monsoon. Elsewhere dry conditions prevailed with the exception of the passage of a frontal system which moved through southeastern Australia. The numerical model used for this study is the Australia Bureau of Meteorology Limited Area Prediction System (LAPS) to which 4 land-surface parameterisation schemes are explicitly coupled; (a) the Manabe (1969) bucket scheme, (b) ISBA (Noilhan and Planton, 1989), (c) BATS (Dickinson et al., 1993) and (d) BASE (Desborough, 1997). These 4 schemes vary in complexity from the simple bucket which has no explicit canopy representation and single level of soil moisture to BASE which explicitly represents the effects of canopy-deep soil-surface-atmospheric interactions using a multi-layer mosaic approach.
Six 48-h simulations are performed for each land-surface scheme using the mesoscale model, LAPS (Puri et al., 1998). For the first simulation each land-surface scheme is allowed to set their own internal parameters based on the assumption that the vegetation over Australia was uniformly grass, and a realistic initial soil moisture field was prescribed. In the next 5 simulations, 1st order variables (albedo, roughness length, minimal stomatal resistance, fractional vegetation and maximum leaf area index) were held fixed for all schemes and the initial soil wetness index was prescribed as (a) realistic (as in the 1st simulation), (b) saturated (100% of the bucket capacity, (c) field capacity (75% of the bucket capacity), (d) wilting point (1mmof water) and (e) zero moisture (0mm of water).
This study focusses on two main questions. Firstly, does the complexity of the land surface parameterisation scheme have an impact on the accuracy of the forecast, in particular the accurate simulation of the areal rainfall pattern and intensity ? Secondly, what are the important land-surface-atmospheric interactions (if any exist) which affect the timing, strength and position of the frontal system?