|
|
Dumont, B., Basso, B., Leemans, V., Bodson, B., Destain, J. - P., & Destain, M. - F. (2015). A comparison of within-season yield prediction algorithms based on crop model behaviour analysis. Agricultural and Forest Meteorology, 204, 10–21.
Abstract: The development of methodologies for predicting crop yield, in real-time and in response to different agro-climatic conditions, could help to improve the farm management decision process by providing an analysis of expected yields in relation to the costs of investment in particular practices. Based on the use of crop models, this paper compares the ability of two methodologies to predict wheat yield (Triticum aestivum L.), one based on stochastically generated climatic data and the other on mean climate data. It was shown that the numerical experimental yield distribution could be considered as a log-normal distribution. This function is representative of the overall model behaviour. The lack of statistical differences between the numerical realisations and the logistic curve showed in turn that the Generalised Central Limit Theorem (GCLT) was applicable to our case study. In addition, the predictions obtained using both climatic inputs were found to be similar at the inter and intra-annual time-steps, with the root mean square and normalised deviation values below an acceptable level of 10% in 90% of the climatic situations. The predictive observed lead-times were also similar for both approaches. Given (i) the mathematical formulation of crop models, (ii) the applicability of the CLT and GLTC to the climatic inputs and model outputs, respectively, and (iii) the equivalence of the predictive abilities, it could be concluded that the two methodologies were equally valid in terms of yield prediction. These observations indicated that the Convergence in Law Theorem was applicable in this case study. For purely predictive purposes, the findings favoured an algorithm based on a mean climate approach, which needed far less time (by 300-fold) to run and converge on same predictive lead time than the stochastic approach. (C) 2015 Elsevier B.V. All rights reserved.
|
|
|
|
Dumont, B., Leemans, V., Ferrandis, S., Bodson, B., Destain, J. - P., & Destain, M. - F. (2014). Assessing the potential of an algorithm based on mean climatic data to predict wheat yield. Precision Agric., 15(3), 255–272.
Abstract: The real-time non-invasive determination of crop biomass and yield prediction is one of the major challenges in agriculture. An interesting approach lies in using process-based crop yield models in combination with real-time monitoring of the input climatic data of these models, but unknown future weather remains the main obstacle to reliable yield prediction. Since accurate weather forecasts can be made only a short time in advance, much information can be derived from analyzing past weather data. This paper presents a methodology that addresses the problem of unknown future weather by using a daily mean climatic database, based exclusively on available past measurements. It involves building climate matrix ensembles, combining different time ranges of projected mean climate data and real measured weather data originating from the historical database or from real-time measurements performed in the field. Used as an input for the STICS crop model, the datasets thus computed were used to perform statistical within-season biomass and yield prediction. This work demonstrated that a reliable predictive delay of 3-4 weeks could be obtained. In combination with a local micrometeorological station that monitors climate data in real-time, the approach also enabled us to (i) predict potential yield at the local level, (ii) detect stress occurrence and (iii) quantify yield loss (or gain) drawing on real monitored climatic conditions of the previous few days.
|
|
|
|
Dumont, B., Leemans, V., Mansouri, M., Bodson, B., Destain, J. - P., & Destain, M. - F. (2014). Parameter identification of the STICS crop model, using an accelerated formal MCMC approach. Env. Model. Softw., 52, 121–135.
Abstract: This study presents a Bayesian approach for the parameters’ identification of the STICS crop model based on the recently developed Differential Evolution Adaptive Metropolis (DREAM) algorithm. The posterior distributions of nine specific crop parameters of the STICS model were sampled with the aim to improve the growth simulations of a winter wheat (Triticum aestivum L) culture. The results obtained with the DREAM algorithm were initially compared to those obtained with a Nelder-Mead Simplex algorithm embedded within the OptimiSTICS package. Then, three types of likelihood functions implemented within the DREAM algorithm were compared, namely the standard least square, the weighted least square, and a transformed likelihood function that makes explicit use of the coefficient of variation (CV). The results showed that the proposed CV likelihood function allowed taking into account both noise on measurements and heteroscedasticity which are regularly encountered in crop modelling. (C) 2013 Elsevier Ltd. All rights reserved.
|
|
|
|
Ferrise, R., Toscano, P., Pasqui, M., Moriondo, M., Primicerio, J., Semenov, M. A., et al. (2015). Monthly-to-seasonal predictions of durum wheat yield over the Mediterranean Basin. Clim. Res., 65, 7–21.
Abstract: Uncertainty in weather conditions for the forthcoming growing season influences farmers’ decisions, based on their experience of the past climate, regarding the reduction of agricultural risk. Early within-season predictions of grain yield can represent a great opportunity for farmers to improve their management decisions and potentially increase yield and reduce potential risk. This study assessed 3 methods of within-season predictions of durum wheat yield at 10 sites across the Mediterranean Basin. To assess the value of within-season predictions, the model SiriusQuality2 was used to calculate wheat yields over a 9 yr period. Initially, the model was run with observed daily weather to obtain the reference yields. Then, yield predictions were calculated at a monthly time step, starting from 6 mo before harvest, by feeding the model with observed weather from the beginning of the growing season until a specific date and then with synthetic weather constructed using the 3 methods, historical, analogue or empirical, until the end of the growing season. The results showed that it is possible to predict durum wheat yield over the Mediterranean Basin with an accuracy of normalized root means squared error of <20%, from 5 to 6 mo earlier for the historical and empirical methods and 3 mo earlier for the analogue method. Overall, the historical method performed better than the others. Nonetheless, the analogue and empirical methods provided better estimations for low-yielding and high-yielding years, thus indicating great potential to provide more accurate predictions for years that deviate from average conditions.
|
|
|
|
Mansouri, M., & Destain, M. - F. (2015). Predicting biomass and grain protein content using Bayesian methods. Stoch. Environ. Res. Risk Assess., 29(4), 1167–1177.
Abstract: This paper deals with the problem of predicting biomass and grain protein content using improved particle filtering (IPF) based on minimizing the Kullback-Leibler divergence. The performances of IPF are compared with those of the conventional particle filtering (PF) in two comparative studies. In the first one, we apply IPF and PF at a simple dynamic crop model with the aim to predict a single state variable, namely the winter wheat biomass, and to estimate several model parameters. In the second study, the proposed IPF and the PF are applied to a complex crop model (AZODYN) to predict a winter-wheat quality criterion, namely the grain protein content. The results of both comparative studies reveal that the IPF method provides a better estimation accuracy than the PF method. The benefit of the IPF method lies in its ability to provide accuracy related advantages over the PF method since, unlike the PF which depends on the choice of the sampling distribution used to estimate the posterior distribution, the IPF yields an optimum choice of this sampling distribution, which also utilizes the observed data. The performance of the proposed method is evaluated in terms of estimation accuracy, root mean square error, mean absolute error and execution times.
|
|
