Forcing a wheat crop model with LAI data to access agronomic variables: Evaluation of the impact of model and LAI uncertainties and comparison with an empirical approach

The objective of this study is to evaluate the performances of estimating agronomic variables, such as total above ground biomass at key stages, or yield, from LAI data that could potentially be obtained from remote sensing observations. Approaches based either on empirical relationships or on forcing LAI within the STICS model (Brisson et al., 2009) are considered, with emphasis on the effect of the accuracy and frequency of LAI data used. Both actual and simulated case studies on wheat for Northern France conditions were investigated under several levels of knowledge of the model input parameters and initial conditions.The results highlight the interest of using model based approaches for the estimation of agronomic variables. Forcing LAI data into the crop model allows compensating for the lack of detailed knowledge on management practices or soil characteristics. However, error and frequency of LAI observations may have an important impact on the estimation of agronomic variables, particularly for the early growth stages. In these conditions, an empirical approach, based on the calibration of a relationship between LAI at a given stage and the agronomic variable, provides an efficient alternative, though the validity of empirical relationships depends greatly on the database on which they have been obtained.

► Forcing LAI into the crop model STICS allows to obtain information on agronomic variables.
► LAI forcing has an equivalent effect as improved knowledge of model inputs.
► Errors in forced LAI affect biomass estimation accuracy at early growth stages.
► Interactions appear among LAI data frequency, error level and interpolation procedure.