Representation of dynamic grass density in land surface model ORCHIDEE trunk v4.

In semi-arid regions, grasslands naturally display a self-organized pattern that optimizes resource utilization and productivity. Representing this type of vegetation in land surface model constitutes a difficult challenge. To simulate these grasses, the ORCHIDEE land surface model treats grass density as the ratio of the area occupied by individuals to the Plant Functional Type (PFT) area, assuming a fixed grass density of 1 for maximal occupancy. However, the fixed maximal grass density lacks the response of grassland to environmental perturbations. In addition, the low biomass contained in certain pixels results in frequent mortality, indicative of resource limitations at the plant individual level. To address this considerable limitation, we introduced dynamic reduction of grass density based on mortality indicators, hence enhancing individual biomass and alleviating mortality occurrences. The adaptive approach significantly decreased mortality events across most pixels while enhancing leaf area index (LAI) for the majority of them. Our findings suggest that optimizing resource through grass density reduction in response to environmental condition, could not only improve individual biomass to alleviate mortality but also enhance overall grassland production.