Improving High Latitude Precipitation Estimation Using A-Train Observations

The Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis has provided three-hourly estimates of precipitation in the range 50°N-50°S throughout the TRMM era (1998-present). In preparation for the follow-on Global Precipitation Mission (GPM), development work is underway to produce truly global estimates. This presentation will examine a methodology for estimating precipitation in middle and higher latitudes.

Global precipitation estimates have been available for years via the Global Precipitation Climatology Project (GPCP). Because current passive microwave retrievals are unable to distinguish between precipitation and cold underlying surface, GPCP retrievals in these areas rely on estimates from the TIROS Operational Vertical Sounder (TOVS) for January 1979-April 2005, and the Atmospheric Infrared Sounder (AIRS) thereafter. However, both TOVS and AIRS estimates have been found to underestimate the true precipitation amount, while overestimating precipitation frequency.

The work presented here takes advantage of the “A-Train”. The Aqua satellite carries both the AIRS and the Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E), followed a minute later by CloudSat, with its Profiling Radar (CPR). This conjunction allows continuous match-ups of precipitation retrievals from these three instruments. A robust database of match-ups has been assembled for the period August 2006-July 2007.

While the temporal match-ups are ubiquitous, spatial match-ups amongst the sensors present a larger challenge. CPR footprints are ~1.4 x 2.5 km along the nadir track, compared to swaths of ~4 x 6 km for AMSR-E, and ~40 x 40 km for AIRS. In constructing the database, CPR estimates were averaged to AIRS footprint size and matched with the corresponding AIRS scene. However, the CPR nadir-only track limits the comparison to just a subset of the AIRS fields-of-view. To overcome this, a similar procedure was used to average AMSR-E scenes up to the AIRS footprint size, first using the nadir subset of both AMSR-E and AIRS, and then repeated using the full two-dimensional range of both. This yields a 1D-to-2D transfer function that is applied back onto the CPR-AIRS match-ups, resulting in estimates of what CPR “would have seen” had it been capable of scanning the entire footprint covered by AIRS.

The resulting calibration is then applied to AIRS estimates, in an attempt to reduce the spatial coverage and increase the amount. The technique has been developed over ocean, by season and latitude band. As a first step, the oceanic results have been applied globally. The presentation will examine global maps and zonal profiles of both precipitation frequency and amount to assess the overall impact of the calibration and to identify areas requiring improvement. Ultimately, this technique should benefit both GPM and GPCP.