Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Spatio-temporal estimation of climatic variables for gap filling and record extension using Reanalysis data
Ist Teil von
Theoretical and applied climatology, 2019-07, Vol.137 (1-2), p.1089-1104
Ort / Verlag
Vienna: Springer Vienna
Erscheinungsjahr
2019
Quelle
SpringerLink Journals (MCLS)
Beschreibungen/Notizen
The availability of reliable meteorological records is crucial for the development of a number of environmental studies. Unfortunately, these records are not always complete, usually show errors and/or have an insufficient length. This paper presents a gap filling and data record extension methodology for minimum temperature, maximum temperature, and precipitation. It uses climatic information from the NCEP-NCAR Reanalysis project, identifying pixels (grid cells) within a Reanalysis domain that have the highest Pearson’s correlation coefficient with the variable of interest. Nine stations in the Maipo River basin (Santiago, Chile) were selected for a reconstruction experiment (from 1950 to 1970) and a subsequent gap filling experiment (from 1970 to 2012). A generalized linear mixed model with a bidirectional stepwise fit procedure was used to model temperature, whereas precipitation occurrence was represented using a generalized linear mixed model with binomial distribution, and precipitation amount used an exponential generalized linear model. The performance of the algorithm was compared with inverse distance weighting and spline interpolation methods and further evaluated using the Standardized Precipitation Evapotranspiration Index, contrasting real versus modeled data. Values of the coefficient of determination averaged 0.76 (0.74–0.84) minimum temperature, 0.73 (0.73–0.81) for maximum temperature, and 0.68 (0.51–0.78) for precipitation. Root-mean-squared error was around 1.5 °C and 5 mm for temperature and precipitation, respectively. The model explains local variation of climatic variables and indicators and can be replicated anywhere, as the
Reanalysis
data are easily accessible and have a worldwide coverage.