|Reference : Wheat yield and PAI decreasing shape curve|
|Scientific congresses and symposiums : Unpublished conference|
|Life sciences : Agriculture & agronomy|
|Wheat yield and PAI decreasing shape curve|
|Kouadio, Amani Louis [Université de Liège - ULg > > > Doct. sc. (sc. & gest. env. - Bologne)]|
|Tychon, Bernard [Université de Liège - ULg > Département des sciences et gestion de l'environnement > Département des sciences et gestion de l'environnement >]|
|ENVITAM-UNITER PhD day workshop|
|12 Janvier 2010|
|Ecoles doctorales ENVITAM et UNITER|
|[en] Plant Area Index (PAI), Yield forecasting, Senescence, Winter wheat|
|[en] Estimation of cereal-crop production is considered as a priority in most crop research programmes due to the relevance of food grain to world agricultural production. Today, a large number of agrometeorological models for crop yield assessment are available with various levels of complexity and empiricism.
A preliminary study was performed with simulated data of wheat yield and LAI derived from the WOFOST/CGMS agrometeorological model. The main hypothesis underlying this study is that it’s possible to improve wheat yield estimates from metrics stretched from LAI decreasing curves. This preliminary study showed that wheat yield can be estimated by metrics stretched from simulated LAI curve-fitting done by a modified Gompertz function
[G =A*exp (-exp(-k(t-m)))] and a logistic function [G = A / 1+exp(-k(t-m)); where G is the green LAI (gLAI), A the initial percentage of LAI, m the position of the inflexion point in the decreasing part of the LAI curve and k the relative senescence rate.
In 2009 a large field campaign in the Grand-Duchy of Luxembourg and France was done to check the validity of such approach with field data. Hemispheric images were taken on 18 winter wheat fields during the crop cycle, preferentially from inflorescence emergence to maturity. The variable of interest, green PAI (Plant Area Index), was retrieved after analyses of images by the CAN-EYE software (v. 6.1).
Data used as input to establish the model of wheat yield estimate are the value of observed PAImax, and metrics k and m, stretched from observed PAI curves fitted by Gompertz and logistic functions. The model obtained by multilinear regression with these variables reveals that wheat yield can be estimated, at the scale of the plot, with a r² ≈ 0.70 and a RMSE = 0.87 t/ha (RRMSE = 9%).
The validation of such approach at the scale of an agricultural zone or region will be performed in the next step of our study, by using remote sensing data (air temperature, PAI or LAI) and phenology data as input. Such simple models may be considered as a first yield estimates that may be completed, if justified, by other agrometeorological models in order to provide a better integrated yield assessment.
|File(s) associated to this reference|
All documents in ORBi are protected by a user license.