Comparison of time series of the vegetation index derived from NOAA AVHRR and SPOT VEGETATION satellite observations

The paper aimed at comparison of remotely sensed data for forest areas in Poland. The main objective was to study how time-series of satellite data derived from NOAA AVHRR and SPOT VEGETATION (SPOT VGT) are compatible and can be useful for forest cover monitoring. NOAA AVHRR fifteen-year observations collected by the receiving station mounted at the Institute of Geodesy and Cartography were processed with radiometric, geometric and atmospheric corrections to obtain Normalized Difference Vegetation Index (NDVI) images. Next, daily NDVI images were clustered into ten-day composites and smoothed with Savitzky-Golay filter. The long-term NDVI data derived from SPOT VGT were post-processed in order to prepare reference set to cross-comparison analysis. The study area encompasses forest complexes representative for various climatic conditions: Borecka, Augustow, Knyszyn and Bialowieta Forests as well as forests of the Karkonosze and Beskid Zywiecki mountains. The satellite data were analyzed within vegetation period (April-September). The comparison of frequency of cloud-free observations and mean NDVI values over study sites was performed to follow the deviations in two datasets. The Wilcoxon signed-rank test proved that there are statistically significant NDVI differences between NOAA and SPOT data for Augustow, Knyszyn, Bialowieta and Karkonosze Forests. On the other hand, NDVI observations for Borecka Forest and Beskid Zywiecki Mts. revealed quite high compatibility. The Spearman rank correlation results proved that NDVI time-series derived from NOAA AVHRR and SPOT VGT are incoherent and depended on test site or year of the observation. The cross-comparison analysis revealed that there is a need for additional investigation into achieving full compatibility of satellite datasets for successful forest cover monitoring.