Comment on "Hierarchical Bayesian space-time interpolation versus spatio-temporal BME approach" by Hussain et al. (2010)
Year: 2010
Type: letter
Abstract: This is a clarifying comment regarding the article “Hi-erarchical Bayesian space-time interpolation versus spatio-temporal BME approach ” (Hussain et al., 2010). In the ar-ticle, the authors present a very interesting methodological study to compare the results obtained by the Bayesian Max-imum Entropy (BME) and hierarchical Bayesian interpola-tion (HBI) techniques in a spatiotemporal context. The study concludes in declaring HBI a more accurate contestant that also produces lower prediction error, compared to BME. BME began its development as a novel approach and an improvement in space-time analysis in the early 1990’s (Christakos, 1990, 1992, 2000). BME has a distinctive edge compared to the classical linear model predictive approaches because its foundations are free from restrictions and as-sumptions typically present in competing methods (e.g., data distribution assumptions, consideration of soft data as hard values, etc.), and it can incorporate a broader spectrum of in-formation such as uncertain observations, probabilistic data, physical laws, etc. Naturally, like any consistent theory, BME includes pre-existing theories as its special cases. The article authors are clear, accurate and explicit in stating that they apply BME kriging in their study, that is, they apply the limited case of BME when it is mathematically reduced to the kriging classical technique. The above remark is at the heart of my comment: The article describes inaccurately the comparison as HBI versus BME, because the conclusions place the critique on the BME methodology in its entirety without having used any of the exclusive BME features (e.g., use of soft data, physical laws, non-linear prediction). To the informed reader, the article es-sentially illustrates a comparison between HBI and kriging applied through BME software. I think this clarification is necessary to prevent readers that are less familiarized with
Source: Advances in geosciences
Author: A. Kolovos
Institution: SAS Institute (United States)
Language: English
Subfield: Soil Science
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $301
OpenAlex ID: W2118914904