Application of Regression Kriging to Air Pollutant Concentrations in Japan with High Spatial Resolution

Abstract

The application of regression kriging to air pollutants in Japan was examined for the purpose of providing a practical method to obtain a spatial distribution with sufficient accuracy and a high spatial resolution of 1 × 1 km. We used regulatory air monitoring data from the years 2009 and 2010. Predictor variables at 1 × 1 km resolution were prepared from various datasets to perform regression kriging. The prediction performance was assessed by indicators, including root mean squared error (RMSE) and R 2 , calculated from the leave-one-out cross validation results, and was compared to the results obtained from a linear regression method, often referred to as land use regression (LUR). Regression kriging wellexplained the spatial variability of NO 2 , with R 2 values of 0.77 and 0.78. Ozone (O 3 ) was moderately explained, with R 2 values of 0.52 and 0.66. The reason for this difference in performance between NO 2 and O 3 might be the characteristics of these pollutants -primary or secondary. Regression kriging outperformed the linear regression method with regard to RMSE and R 2 . The performance of regression kriging in this work was comparable to that found in previous studies. The results indicate that regression kriging is a practical procedure that can be applied for the prediction of the spatial distribution of air pollutants in Japan, with sufficient accuracy and a high spatial resolution.

Appraising the effects of atmospheric aerosols and ground particulates concentrations on GPS-derived PWV estimates

“…The downloaded MODIS precipitable water vapour variable (IR retrieval) datasets were accessed using Panoply software (Vollmer, 2010). The subset MODISAOD and MODISPWV estimates were then collocated over the ABUZ station using the Kriging interpolation mechanism (Araki et al, 2015;Li et al, 2016).…”

Ozone troposphérique en France : distribution spatiale et tendances sur la période 1999-2012

“…Sur la base des coefficients de Pearson, les variables les plus significatives (p < 0,05) sont prises en compte dans la régression multiple pour définir le modèle de régression locale. Un krigeage des résidus obtenus à l'issue de la première étape est réalisé pour améliorer les valeurs interpolées (Araki et al, 2015). Le modèle de régression locale a été appliqué à tous les pixels selon une résolution spatiale de 250 m en France.…”

“…Des méthodes similaires de krigeage ont été appliquées avec succès pour cartographier la répartition spatiale des concentrations de polluants atmosphériques dans des études environnementales et épidémiolo-giques (ex. Ross et al, 2007 ;Pearce et al, 2009 ;Araki et al, 2015).…”

“…Cependant, la performance de l'interpolation est fortement affectée par le nombre et la répartition spatiale des stations de surveillance disponibles, et par la prise en compte ou non du relief (Bogaert et al, 2009). Pour contrer cette difficulté, une approche hybride d'interpolation locale a été testée (Joly et al, 2011 ;Araki et al, 2015). Monks et al (2015) a souligné qu'à ce jour une bonne compréhension des tendances de l'O 3 troposphérique et une bonne quantification de ses impacts restent encore difficiles.…”

Effect of spatial outliers on the regression modelling of air pollutant concentrations: A case study in Japan

“…The concentrations at the locations with no observations are predicted by the obtained regression model. In some studies, residuals of a regression model are interpolated by the kriging method and summed up to the predictions by the regression model (e.g., Beelen et al, 2009;Pearce et al, 2009;Sampson et al, 2013;Araki et al, 2015).…”