| description abstract | Three-hourly time series of precipitation from three high-resolution precipitation products [Tropical Rainfall Measuring Mission (TRMM) algorithm 3B-42, the Climate Prediction Center?s morphing method (CMORPH), and the Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN)] and two reanalyses are examined for their frequency characteristics using broad and narrow variance categories. After isolating the diurnally forced peaks (at 24, 12, 8, and 6 h), the power spectra are divided into comprehensive broad bands comprising the annual (?80 days?1 yr), intraseasonal (20 to ?80 days), slow (6?20 days) and fast (36 h?6 days) synoptic, and high-frequency (6?36 h) periods. Global maps accounting for 100% of precipitation?s variance are analyzed to identify unique regional behaviors. Annual variability is strongest over regions affected by the seasonal migration of the intertropical convergence zone, as well as over monsoonal regions. The intraseasonal band displays off-equatorial evidence of the Madden?Julian oscillation (MJO), particularly in the Indian Ocean, but the MJO?s rainfall is partially manifested in the slow synoptic band and at higher frequencies. The fast synoptic band is particularly strong over the oceans, while high-frequency variability is enhanced over land by more extreme surface gradients. Diurnal variance is strongest at low latitudes and is pronounced over regions with well-known diurnal circulations, including mountains and coastlines. Interproduct and intermodel differences also indicate biases of the precipitation product algorithms and convective parameterizations, including a strong bias toward low-frequency variability in the relaxed Arakawa?Schubert parameterization employed by one of the reanalyses, as well as increased white-spectral characteristics over land in the precipitation products. | |
