Publication | Closed Access
Long sequence time series evaluation using standardized principal components
334
Citations
0
References
1993
Year
Earth ObservationEngineeringLong Time SeriesEarth ScienceGeophysicsStandardized Principal ComponentsData ScienceData MiningNdvi AnomaliesMeteorological MeasurementPublic HealthPrincipal Component AnalysisNonlinear Time SeriesGeodesyClimate VariabilityMeteorologyGeographyKnowledge DiscoveryTemporal Pattern RecognitionForecastingEarth Observation DataFunctional Data AnalysisClimatologyRemote Sensing
The potential of using Standardized Principal Components for the analysis of long time series of spatial environmental data is assessed using a series of 36 monthly AVHRR-derived NDVI images for Africa for the years 1986-88 as an illustration. The first component is found to represent the characteristic NDVI regardless of the season. The second, third, and fourth components relate to seasonal changes in NDVI. The fifth and sixth components uncover a sensor-related drift in the NDVI values due to successively later equatorial crossings of the NOAA-9 satellite. The seventh and eighth components illustrate NDVI anomalies related to significant El Nino/Southern Oscillation (ENSO) events, primarily in southern Africa. The technique is shown to be a comprehensive indicator of change events in time series data that is sensitive to periodic and aperiodic events alike.