A new paper for the assessment of the performance of satellite rainfall products on Hydrology and Earth System Sciences Discussions Journal by Massari et al. (10.5194/hess-2017-163) has been published.
The assessment of the performance of satellite rainfall estimates is commonly carried out by comparing these estimates with ground based observations like radars and rain gauges. However, such instruments are not always present everywhere and in some regions of the world like Africa and Asia they are very scarce. It turns out that understanding how satellite rainfall estimates work in these areas is practically impossible.
Born from a collaboration with the United States Department of Agriculture, the paper investigates a new technique known as Triple Collocation to estimate global correlations of different state-of-the-art satellite rainfall products. Triple collocation has already been applied to precipitation data but, given the strong dependence among satellite based products, it required the use of at least one ground based dataset. The main innovation of the paper stands for the use of the recently developed SM2RAIN method which has provided a totally independent estimate of global rainfall that has been used in place of ground based observations.
The results show that satellite based products work reasonably well in many parts of the planet, however they are outscored by reanalysis dataset in the Northern Hemisphere and by SM2RAIN in the Southern Hemisphere. However, the reanalysis product and SM2RAIN have problems in rainfall estimation in dense forests where satellite based observations work relatively well.
This new study will have immediate effects for hydrology, atmosphere and agricultural studies.