Linear trends in seasonal vegetation time series and the modifiable temporal unit problem

[摘要] Time series of vegetation indices (VI) derived from satellite imageryprovide a consistent monitoring system for terrestrial plant productivity.They enable detection and quantification of gradual changes within the timeframe covered, which are of crucial importance in global change studies, forexample. However, VI time series typically contain a strong seasonal signalwhich complicates change detection. Commonly, trends are quantified usinglinear regression methods, while the effect of serial autocorrelation isremediated by temporal aggregation over bins having a fixed width.Aggregating the data in this way produces temporal units which aremodifiable. Analogous to the well-known Modifiable Area Unit Problem (MAUP),the way in which these temporal units are defined may influence the fittedmodel parameters and therefore the amount of change detected. This paperillustrates the effect of this Modifiable Temporal Unit Problem (MTUP) on asynthetic data set and a real VI data set. Large variation in detectedchanges was found for aggregation over bins that mismatched full lengths ofvegetative cycles, which demonstrates that aperiodicity in the data mayinfluence model results. Using 26 yr of VI data and aggregation overfull-length periods, deviations in VI gains of less than 1% were foundfor annual periods (with respect to seasonally adjusted data), whiledeviations increased up to 24% for aggregation windows of 5 yr. Thisdemonstrates that temporal aggregation needs to be carried out with care inorder to avoid spurious model results.

[学科分类] 地球化学与岩石