Modeling sugarcane yield with a process-based model from site to continental scale: uncertainties arising from model structure and parameter values

摘要

Agro-land surface models (agro-LSM) have been developed from the integrationof specific crop processes into large-scale generic land surface models thatallow calculating the spatial distribution and variability of energy, waterand carbon fluxes within the soil–vegetation–atmosphere continuum. Whendeveloping agro-LSM models, particular attention must be given to the effectsof crop phenology and management on the turbulent fluxes exchanged with theatmosphere, and the underlying water and carbon pools. A part of theuncertainty of agro-LSM models is related to their usually large number ofparameters. In this study, we quantify the parameter-values uncertainty inthe simulation of sugarcane biomass production with the agro-LSMORCHIDEE–STICS, using a multi-regional approach with data from sites inAustralia, La Réunion and Brazil. In ORCHIDEE–STICS, two models arechained: STICS, an agronomy model that calculates phenology and management,and ORCHIDEE, a land surface model that calculates biomass and otherecosystem variables forced by STICS phenology. First, the parameters thatdominate the uncertainty of simulated biomass at harvest date are determinedthrough a screening of 67 different parameters of both STICS and ORCHIDEE ona multi-site basis. Secondly, the uncertainty of harvested biomassattributable to those most sensitive parameters is quantified andspecifically attributed to either STICS (phenology, management) or toORCHIDEE (other ecosystem variables including biomass) through distinct MonteCarlo runs. The uncertainty on parameter values is constrained usingobservations by calibrating the model independently at seven sites. In athird step, a sensitivity analysis is carried out by varying the mostsensitive parameters to investigate their effects at continental scale. AMonte Carlo sampling method associated with the calculation of partial rankedcorrelation coefficients is used to quantify the sensitivity of harvestedbiomass to input parameters on a continental scale across the large regionsof intensive sugarcane cultivation in Australia and Brazil. The tenparameters driving most of the uncertainty in the ORCHIDEE–STICS modeledbiomass at the 7 sites are identified by the screening procedure. We foundthat the 10 most sensitive parameters control phenology (maximum rate ofincrease of LAI) and root uptake of water and nitrogen (root profile and rootgrowth rate, nitrogen stress threshold) in STICS, and photosynthesis (optimaltemperature of photosynthesis, optimal carboxylation rate), radiationinterception (extinction coefficient), and transpiration and respiration(stomatal conductance, growth and maintenance respiration coefficients) inORCHIDEE. We find that the optimal carboxylation rate and photosynthesistemperature parameters contribute most to the uncertainty in harvestedbiomass simulations at site scale. The spatial variation of the rankedcorrelation between input parameters and modeled biomass at harvest is wellexplained by rain and temperature drivers, suggesting different climate-mediated sensitivities of modeled sugarcane yield to the model parameters,for Australia and Brazil. This study reveals the spatial and temporalpatterns of uncertainty variability for a highly parameterized agro-LSM andcalls for more systematic uncertainty analyses of such models.