RIO model


RIO model is an advanced interpolation-regression model. The idea is that the concentration of a particular pollutant at a particular location can be divided to two components: regional and local. The local contribution to the concentration is determined by different geographically dependent factors - so called spatial drivers, such as, e.g., elevation, traffic intensities, ventilation index, gridded residential heating emissions, etc). This local contribution can be removed from the measured station concentration values before spatial interpolation to a uniform grid is performed. After the interpolation, the local trend is re-introduced to each grid cell based on the spatial driver "configuration" in that particular grid cell. 

As a first step of the procedure, spatial correlations between the pollutant concentrations, measured at monitoring stations and each particular relevant driver is computed separately, in order to establish their relevances to the pollutant. Next, a so-called  β-parameter is computed and optimized as a combination of selected spatial drivers which represent a best correlation with the measured data. The differences between the concentration values measured at the monitoring stations and those computed using the  β-parameter are then interpolated using ordinary kriging method. The final concentration for each grid cell is computed as a sum of the concentration computed using the β-parameter and the interpolated difference value. 

The best combination of spacial drivers is specific for each pollutant. Even the results of regional dispersion models, such as CMAQ, or satellite data can be used as spatial drivers, in order to achieve better spatial resolution of the concentration maps. 

More details can be found here: 

NOTE: The color scale used in the concentration maps differs from that of the measured concentration labels, with the latter based on the color legend explained at the bottom of this page: