Evaluation of Landsat Archive for Global and Time Series Validation of AVHRR-MODIS Vegetation Index Data Records

Spectral vegetation indices (VIs) derived from satellite images are useful in land surface phenology studies. These time series portray seasonal and annual changes to vegetation globally. Advanced Very High Resolution Radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer (MODIS) are two satellite sensors which provide daily global VIs. AVHRR and MODIS time series could be used together to assess longer-term changes in global phenological and climate change studies. Inherent differences of sensor characteristics potentially lead to inconsistencies between these two sensor time series, which needs to be evaluated. Landsat provides the longest data record; i.e., Landsat-5 Thematic Mapper had been acquiring data since 1984. The Landsat dataset could provide an opportunity to evaluate multi-sensor continuity of these global datasets. The objective of this study was to evaluate the availability of low cloud coverage or cloud-free Landsat scenes at the United States Geological Survey (USGS) for validating data consistency and continuity of AVHRR-MODIS global vegetation index datasets over an 18-year time period, from 1984 to 2002. The USGS Landsat Global Archive file was obtained from the USGS Landsat website, which provided a graphic view of the geographic and temporal acquisition of Landsat data from 1972 to 2011. International Geosphere-Biosphere Programme (IGBP) land cover data were extracted from the 2002 MODIS land cover product (MYC12C1) to stratify available scenes for land cover types. Archived scene metadata were used to find Landsat scenes with less than 25% cloud cover. The most comprehensive data were available over the conterminous United States, the southern part of Canada, and the northern part of Mexico, where full 18-year time series of Landsat-5 TM image stack (10 - 20 image per year) could be generated at many PATH/ROW locations. Outside of North America, no single region had sufficient data archived for all years of the 1984 to 2002 time period to allow for a continuous time-series to be generated. However, occasional regional coverage of Landsat data were observed for South America, Europe, Southeast Asia, Austraia, and Africa, which could be used to assess global consistency of AVHRR and MODIS datasets, with the exception of a few years with insufficient data collection. We conclude that the USGS EROS Landsat-5 TM archive can provide a useful temporal and spatial coverage to validate data consistency and continuity of AVHRR-MODIS global vegetation index time series.