A statistical model for spatial inventory data: a case study of N2O emissions in municipalities of southern Norway

In this paper we apply a linear regression with spatial random effect to model geographically distributed emission inventory data. The study presented is on N₂O emission assessments for municipalities of southern Norway and on activities related to emissions (proxy data). Taking advantage of the spatial dimension of the emission process, the method proposed is intended to improve inventory extension beyond its earlier coverage. For this, the proxy data are used. The conditional autoregressive model is used to account for spatial correlation between municipalities. Parameter estimation is based on the maximum likelihood method and the optimal predictor is developed. The results indicate that inclusion of a spatial dependence component lead to improvement in both representation of the observed data set and prediction.