Regular readers of this blog are probably convinced that the time series of satellite images are perfectly suited to describe the evolution of the Earth's surface. Space agencies are stepping up efforts to make available products to ease the use of these data. If one sets aside the main author of this blog (note : no need to put aside, he is far behind), it is of course NASA that has most contributed to disseminate advanced remote sensing products such as land surface reflectance or indices describing the vegetation health. These products are convenient to use but can also mask the measurement artifacts that are likely to mislead users like me.
Several teams who used the MODIS products...
Several teams who used the MODIS products "leaf area index" (LAI) or "enhanced vegetation index" (EVI) distributed by NASA believed see that the Amazon forest was turning green during the dry season, which is also the sunniest. They deduced that the forest growth was not limited by water availability in soil, but rather by the amount of solar radiation.
Several teams who used the MODIS products "snow albedo" or "reflectance area" distributed by NASA believed to detect a darkening of the Greenland surface, including in areas close to the summit where the snow melts very rarely. They concluded that the impurity content in the snow surface layer had increased due to the deposition of inorganic or organic dust from the atmosphere.
These two studies have important implications for understanding and anticipating climate change. What will Amazon forest become if the weather dries up? How fast will the Greenland ice sheet melt if the impurities keep on accumulating?
Wait a minute
The results of these studies were contradicted and can be explained by sensor effects. In the Amazon forest, it was a directional effect caused by the angle of the MODIS sensor relative to the sun: at the end of the dry season, the sensor is aligned with the sun which increases the reflectance in the bands NIR and red and therefore causes an apparent increase of the LAI or EVI. The increase disappears if this geometric effect is corrected (Morton et al., 2014). In the case of Sentinel-2, which observes virtually vertically, these effects will be reduced but not completely eliminated and will also have to be corrected.
For the Greenland snow, Polashenski et al. (2015) recently showed that the aging of the MODIS sensor aboard the Terra satellite could be enough to explain the darkening trend. The deterioration of the sensor sensitivity is a normal phenomenon that should be corrected in the new version of MODIS products (Collection 6). The correction of this effect could even help solve other issues that agitate the scientific community as the problem of a mysterious missing carbon sink (Lyapustin et al., 2014).
Of course, these examples are rare and should not discredit remote sensing products. The only lesson is that we must continue to demand our agencies quality products along with their uncertainties, take a look at the doc, and double check our results... The release of even imperfect data also helps speed up the implementation product update and correction, and thus to initiate a virtuous circle between spatial data users and producers!
Figure: Evolution of the reflectance from band 3 from L1B product on a stable desert site in Libya (black: Collection 5, red: Collection 6). The deterioration of the MODIS sensor does not concern the Aqua satellite that is more recent (taken from Lyapustin et al., 2014).
Lyapustin et al. (2014). Scientific impact of MODIS C5 calibration degradation and C6+ improvements, Atmos. Meas. Tech., 7, 4353-4365, doi:10.5194/amt-7-4353-2014.
Morton et al. (2014). Amazon forests maintain consistent canopy structure and greenness during the dry season. Nature, 506 (7487), 221-224, doi:10.1038/nature13006
Polashenski et al. (2015). Neither dust nor black carbon causing apparent albedo decline in Greenland's dry snow zone; implications for MODIS C5 surface reflectance. Geophysical Research Letters. 42, doi:10.1002/2015GL065912.