Sparse and hybrid modelling of relative humidity: the Krško basin case study. Issue 1 (30th January 2020)

• Record Type: Journal Article
• Authors: Kocijan, Juš; Perne, Matija; Grašic, Boštjan; Božnar, Marija Zlata; Mlakar, Primož
• Is Part Of: CAAI transactions on intelligence technology. Volume 5:Issue 1(2020)
• Page Start: 42
• Page End: 48
• Publication Date: 2020-01-30
• Subjects: environmental science computing -- air pollution -- learning (artificial intelligence) -- regression analysis -- Gaussian processes -- humidity -- geophysics computing; physics‐based atmospherical model -- Gaussian‐process regression model -- GP model -- sparse GP modelling -- empirical model training -- physics‐based model -- hybrid modelling -- relative humidity -- Krško basin case study -- atmospheric variable -- data‐driven model -- air‐pollution‐dispersion model; A0250 Probability theory, stochastic processes, and statistics -- A9260T Air quality and air pollution -- A9385 Instrumentation and techniques for geophysical, hydrospheric and lower atmosphere research -- C1140Z Other topics in statistics -- C6170K Knowledge engineering techniques -- C7340 Geophysics computing -- C7360 Environmental science computing; Artificial intelligence -- Periodicals; Computer science -- Periodicals; Artificial intelligence; Computer science; Electronic journals; Periodicals; 006.305
• DOI: 10.1049/trit.2019.0054
• Languages: English
• ISSNs: 2468-6557

Abstract

Abstract : This study describes an application of hybrid modelling for an atmospheric variable in the Krško basin. The hybrid model is a combination of a physics‐based and data‐driven model and has some properties of both modelling approaches. In the authors' case, it is used for the modelling of an atmospheric variable, namely relative humidity in a particular location for the purpose of using the predictions of the model as an input to the air‐pollution‐dispersion model for radiation exposure. The presented hybrid model is a combination of a physics‐based atmospherical model and a Gaussian‐process (GP) regression model. The GP model is a probabilistic kernel method that also enables evaluation of prediction confidence. The problem of poor scalability of GP modelling was solved using sparse GP modelling; in particular, the fully independent training conditional method was used. Two different approaches to dataset selection for empirical model training were used and multiple‐step‐ahead predictions for different horizons were assessed. It is shown in this study that the accuracy of the predicted relative humidity in the Krško basin improved when using hybrid models over using the physics‐based model alone and that predictions for a considerable length of horizon can be used.