A new localized sampling method to improve grape yield estimation of the current season using yield historical data - Agropolis Accéder directement au contenu
Communication Dans Un Congrès Precision Agriculture Année : 2019

A new localized sampling method to improve grape yield estimation of the current season using yield historical data

Résumé

This paper proposes a methodology to improve grape yield sampling and yield estimation of the current season by using historical yield data. This approach is based on the conjoint use of (i) historical yield data all over the study field and (ii) several yield measurements collected at specific sites within the field during the current season. The proposed methodology (Optimized targeted sampling, OTS) assumes that, in viticulture, within-field yield spatial patterns are temporally stable over time as shown in previous studies (this assumption will also be verified with the data used in this study). The first principal component (PC1) of a principal component analysis (PCA) applied to the historical yield data was used to characterize the temporal behaviour of the within-field yield spatial patterns. Results showed that PC1 was always able to explain a significant percentage of the withinfield yield variability across different years meaning that yield temporal stability could be assumed. PC1 scores were then used to choose the best sites to sample (targeted sampling) to estimate the yield of the current season. Yield measurements at these specific sites were used to calibrate a model relating yield values of the current season to PC1 scores. This latter model was finally used to estimate the yield of the current season at all remaining sites. This sampling method was tested on three vine fields (Vitis vinifera L.) in Chile and France with different cultivars (Chardonnay, Cabernet Sauvignon and Syrah). For each of these fields, yield data of several years (between four and seven years) were available at the within-field level. The optimised targeted sampling method (OTS) was applied to the fields under study once the temporal stability of yield patterns was validated. Results were compared to those arising from (i) a uniform random sampling method (URS) and (ii) a stratified random sampling method (SRS). Errors in yield estimates were reduced by more than 6% and 1% on average with respect to the URS and SRS methods, except when yield stable patterns were affected by specific events, i.e. early frost occurring in the season 2014 on the Chardonnay field. The findings demonstrate that the use of localized historical yield data helps to choose reliable sampling sites to improve grape yield estimation.
Fichier non déposé

Dates et versions

hal-02609246 , version 1 (16-05-2020)

Identifiants

Citer

M. Araya Alman, Corentin Leroux, C. Acevedo Opazo, Serge Guillaume, Hector Valdés Gómez, et al.. A new localized sampling method to improve grape yield estimation of the current season using yield historical data. 11th European Conference on Precision Agriculture (ECPA), Jul 2017, Edinburgh, United Kingdom. pp.445-459, ⟨10.1007/s11119-019-09644-y⟩. ⟨hal-02609246⟩
17 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More