Time series analyses are performed at nine land tiles selected over
areas that are expected to be representative of the global variability
of the majority of the MODIS Land products. Shortly after MODIS launch,
well-characterized MODIS data were not available. Therefore, tiles were
selected by examination of the MODIS at-launch land cover product as the
majority of the MODIS Land products are sensitive to or are explicitly
concerned with land cover.
|h09v05: Southwestern U.S.||h11v03: Northwestern Canada||h11v08: Northern Amazon|
|h11v11: Northern Chile||h17v07: Sahel||h20v11: Southern Africa|
|h24v04: Northwestern China||h26v04: Northeastern China||h30v11: Central Australia|
Summary statistics of high-quality, cloud-free science data set pixel values are calculated for each gridded (Level 2G, 3, 4) MODIS Land product in each tile. Mean, standard deviation, minimum and maximum statistics are calculated over 3 x 3 km sites, on a land cover class basis, and on a biome basis. Pixels that are flagged by the product per-pixel QA information as less than high-quality or cloudy are rejected. The remaining pixels are masked with respect to site, land cover and biome maps and the summary statistics derived.