Remote sensing of oceanic rainrates by passive microwave sensors: A statistical-physical approach.

摘要

An analysis of microwave brightness temperatures (T) from the Nimbus-5 Electrically Scanning Microwave Radiometer (ESMR-5) and coincident, high resolution radar rainrates (R) over the GATE area during Phase I has shown them to be related in a manner consistent with radiative transfer theory and recent model calculations. The observed brightness temperatures can be simulated accurately from the rainrate field by direct application of an appropriate R, T relation.;However, the inference of rainrates from the ESMR-5 temperatures, via the same R, T relation, is hampered by a fundamental remote sensing problem: The R, T relation is non-linear while the sensor field-of-view (FOV) is large enough to encompass substantial rainrate inhomogeneity. These factors combine to produce retrieved rainrates that are, on average, too low, with large random errors. The ESMR-5 rainrate retrieval errors were found to be dominated by this rainrate retrieval problem, generally referred to as the beam filling problem.;ESMR-5 inferred rainrates are half as large as, and highly correlated (0.96) with radar rainrates, when both are averaged over the 400 km diameter radar circle. A simple multiplicative correction factor of 2 brings them into close agreement.;A statistical model of the beam filling problem was developed by envisioning an idealized instrument FOV that encompasses an entire gamma distribution of rainrates. This ensemble FOV model was used to calculate FOV temperatures, FOV averaged rainrates, retrieved rainrates and correction factors. A modeled correction factor of 2.2 was found for rainrate and temperature characteristics consistent with GATE conditions.;An alternative rainrate retrieval procedure, suggested by the statistical model, was tested with the GATE data as follows. For each overflight of the radar array the average ESMR-5 temperature over rainy areas was calculated and used to retrieve a single rainrate for each scene. By averaging the temperature over a broader range of rainrates, the beam filling error becomes larger, but more stable. Retrieved and observed rainrates were highly correlated (0.95), and the correction factor for this procedure was (2.37), only 7.7% larger than predicted by the statistical-physical model.;The statistical model suggests that the correction factor varies from 1.6 to 2.7 for suppressed to enhanced tropical convective regimes and decreases to 1.5 as the freezing level and average rain column height decreases to 2.5 km. These results encourage a re-examination of the ESMR-5 data (1973-1976) with the goal of retrieving and correcting oceanic rainrates using the rationale and techniques outlined in this study.