Skip all navigation and jump to content Jump to site navigation
About MODIS News Data Tools /images2 Science Team Science Team Science Team

   + Home
MODIS Publications Link
MODIS Presentations Link
MODIS Biographies Link
MODIS Science Team Meetings Link



Porcu, F, Capacci, D (2008). Seasonal sensitivity of a VIS-NIR-IR rain-no rain classifier. METEOROLOGY AND ATMOSPHERIC PHYSICS, 101(4-Mar), 147-157.

Mid-latitude precipitation characteristics are influenced by the seasonal cycle: general circulation patterns, moisture distribution and cloud type occurrence vary throughout the year over a wide range of different structures. Since radiation in the visible-infrared part of the spectrum is sensitive to the cloud upper layers, the seasonal variability of the cloud structure is expected to affect the capabilities of satellite measurements to infer the precipitation at the ground. This work aims to assess and quantify the seasonal sensitivity of a statistical rain-no-rain classifier applied to data from the moderate resolution imaging spectroradiometer (MODIS) collected for summer and winter seasons over the UK region. In the first part, the satellite radiance measurement distributions for the two seasons were compared and discussed. Then, the comparison between satellite and true rain-no rain classification was carried out in term of statistical parameters (such as the Equitable Threat Score: ETS), showing their dependence on the dry to wet ratio of the statistical ensemble considered. Finally, by considering summer and winter datasets, the seasonal variability of MODIS rain-no rain classifier performance has been established and discussed. The sensitivity of the algorithm to the number and wavelengths of the channels used has been addressed, showing the high impact of the 1.6 mu m channel if combined with one visible channel. The best performance was reached with six channels (0.85, 1.6, 3.9, 7.3, 8.5, and 12 mu m), plus the solar zenith angle as additional input, for which the computed ETS is about 45% for summer and 37% for winter, keeping a fixed dry to wet ratio of 6. The use of an annual algorithm, trained with ensemble of summer and winter pixels, and applied on independent summer and winter ensembles, led to similar values for both summer and winter.



NASA Home Page Goddard Space Flight Center Home Page