Description and evaluation of the Peking University Land Model (PKULM)

The land surface model is a key component of climate and weather models. It is intractable to accurate modeling land-atmosphere interaction in arid and semi-arid regions. Evolved from the Soil-Plant-Atmosphere Model (SPAM), a new Peking University Land Model (PKULM) has been developed with several augments in physical processes and numerical solution for application in arid and semi-arid environments. The model included five modules: radiation transfer, turbulence, photosynthesis, soil heat diffusion and soil water transport. PKULM has augmented biophysical and hydrological processes over SPAM and introduced a new numerical solution for coupled nonlinear equation set of land surface processes: (1) Radiation transfer process within canopy is introduced into the model. The two-stream approximation is used to simulate this process. (2) We use the photosynthesis-based Ball-Berry stomatal resistance scheme instead of the Jarvis algorithm for modeling transpiration, which controls canopy carbon and water fluxes consistently and simultaneously. (3) The original algorithm used in Community Land Model (CLM) for solving coupled photosynthesis and stomatal resistance model does not converge in a dry environment. We propose a new and efficient bisection method that can ensure convergence with ambient low environmental vapor pressure. (4) We update the numerical solution of Richards equation from a saturation-based form to a head-based form. It is apt for modeling soil water movement either in the vadose zone or saturated zone. (5) A new mechanism runoff model is adopted to replace the parameterized model. (6) We adapt the traditional Thomas algorithm and use an implicit iterative method for solving the coupled energy balance and water balance equations of soil, canopy and land surface. With the data obtained in NWC-ALIEX (Northwest China Atmosphere-Land Interaction Experiment), we evaluated the performance of PKULM over cropland in Pingliang and compared it with the Noah land surface model. PKULM outperforms Noah in modeling reflected solar radiation (SR), emitted longwave radiation (LR), sensible heat flux (SH) and latent heat flux (LH) in terms of lower root mean square error (RMSE) and higher correlation coefficients (CC). The RMSEs of PKULM estimations are 4.1, 6.6, 16.3, and 21.8 W . m(-2) for SR, LR, SH, and LH, respectively. The CCs of PKULM modeled SR, LR, SH, and LH, are 0.999, 0.996, 0.976, and 0.989, respectively. Noah produced top 10 cm soil temperature (ST) with a cold bias of 1.03 K. PKULM performs better than Noah in estimating ST in aspects of reproducing smaller bias (0.289 K), more accurate phase, and amplitude. The RMSE of modeled ST from PKULM in top 5 cm is 0.73 K and CC is 0.978. Both Noah and PKULM reproduced the descending trend of soil moisture in the simulation period, but overestimated the magnitude in the daytime and underestimated it in the nighttime. We developed PKULM on the basis of SPAM. PKULM has augmented physical processes and numerical solution that are adapted for arid and semi-arid regions. The modeled energy and water fluxes between land and atmosphere are evaluated and compared to observations. The results show PKULM could reproduce better estimations of SR, LR, SH, and LH than Noah. However, the estimated soil moisture could be improved in the future by adopting more accurate numerical solutions of Richards equation and utilizing more realistic parameterizations of hydrological processes such as surface runoff, evaporation, and drainage.