Introducing Bayesian priors to semi-variogram parameter estimation using fewer observations - Archive ouverte HAL
Communication Dans Un Congrès Année : 2023

Introducing Bayesian priors to semi-variogram parameter estimation using fewer observations

Résumé

Correct estimation of variogram parameters relies on having a sufficiently large dataset. However, operational agri-datasets are often not large enough for variogram fitting. This article presents a new approach to estimating semi-variogram parameters from a small dataset by using a Bayesian approach. The three variogram parameters of the Spherical-Plus-Nugget model were fitted to the semi-variances of a vineyard water stress indicator. Two sources of prior information (i.e. using ancillary data, and using some simplistic assumptions), and six reduced datasets were tested. The results showed that using prior information introduced less variability in estimation results than with the classical approach. The priors extracted from the Sentinel-2 data significantly improved the estimation of the nugget effect, which allowed better preservation of the spatial pattern of kriging predictions.
Fichier principal
Vignette du fichier
Article ECPA reviewed YL Zhang.pdf (512.98 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04158719 , version 1 (11-07-2024)

Identifiants

Citer

Y. Zhang, Leo Pichon, J.A. Taylor, B. Oger, B. Tisseyre. Introducing Bayesian priors to semi-variogram parameter estimation using fewer observations. 14th European Conference on Precision Agriculture, Jul 2023, Bologna, Italy. pp.651-658, ⟨10.3920/978-90-8686-947-3_82⟩. ⟨hal-04158719⟩
36 Consultations
9 Téléchargements

Altmetric

Partager

More