摘要：Proper assessment of air quality is of paramount importance. Accordingly, authorities in large cities have established air pollution monitoring networks that register levels of the most dangerous pollutants in a number of city locations on an hourly basis. Thus, the dataset including such measurements can be considered as a panel dataset where data are spatially fixed. Geostatistics deals with such panel or longitudinal data differently to econometrics: geostatistics takes advantage of the spatio-temporal dependencies to make kriging or cokriging predictions at non-observed sites. However spatio-temporal kriging implies a prohibitive computational burden and, as a consequence, functional kriging has emerged as an alternative strategy for dealing with this type of data since it deals with functional data representing the observations recorded at each location observed. Functional kriging could be said to reproduce the history of the phenomenon under study at non-observed sites. This novel approach has been applied in the city of Madrid (Spain) to particulate matter registers. We predict functional data at some of the most polluted sites of the city.
其他摘要：Proper assessment of air quality is of paramount importance. Accordingly, authorities in large cities have established air pollution monitoring networks that register levels of the most dangerous pollutants in a number of city locations on an hourly basis. Thus, the dataset including such measurements can be considered as a panel dataset where data are spatially fixed. Geostatistics deals with such panel or longitudinal data differently to econometrics: geostatistics takes advantage of the spatio-temporal dependencies to make kriging or cokriging predictions at non-observed sites. However spatio-temporal kriging implies a prohibitive computational burden and, as a consequence, functional kriging has emerged as an alternative strategy for dealing with this type of data since it deals with functional data representing the observations recorded at each location observed. Functional kriging could be said to reproduce the history of the phenomenon under study at non-observed sites. This novel approach has been applied in the city of Madrid (Spain) to particulate matter registers. We predict functional data at some of the most polluted sites of the city.
