GPM / DPR Level-2 Algorithm Theoretical Basis Document (ATBD)
This document describes the basic idea of DPR data processing. It was originally written for the
algorithm used in the at-launch version (V03). The algorithm has been modied and improved
since then. Although the basic idea of data processing remains the same, the actual ow of
processing, in particular that in the solver module, has changed substantially. As a result, some
part of description in section 3.1 may not be relevant any more.
Changes in the DPR algorithm from V05 to V06
- A new SRT code has been modied to include calculations of Hitschfeld-Bordan PIA as well as a hybrid PIA that combines HB and SRT results. As a result of applying the new PIA in SLV module, erroneously large estimates of high precipitation over ocean (near coast) are mitigated and both DPR(MS) and Ku rain estimates in V06A agree better with Gound Validation data over USA.
- A new classication algorithm is introduced by the University of Washington (Stacy Brodzik and Robert Houze) to reclassify the stratiform rain type. The new algorithm improved an angle-bin dependence of rain classication and SLH proles.
Minor changes in DPR L2 algorithm:
- KaHS's sidelobe clutter is mitigated and it is used to re-calculated the data base for KuPR's sidelobe clutter.
- New variables are introduced. They are flagScanPatten in PRE module, PIAhb, PIAhy-brid, reliabFactorHY, reliabFlagHY, stddevEff, stddevHY and zeta in SRT module.
- The meanings of these variables should be referred to the user's manual.
- The surface snow index is modied for the winter temperature inversions.
- The latest SRT data base is applied.
- A bug in agEcho is xed in dual frequency data processing.
- There are cases of instantaneous retrieved precipitation rate classied as stratiform rain that exceed 40 mm/hr. These cases are rare, and the frequency is latitude dependent with more cases near the equator than in the extra-tropics.