A Passive Microwave Algorithm for Precipitation Detection and Convective/Stratiform Classification (PMW_CLASS): Applications for Rain Retrievals and for Developing the Climatology of the Precipitation Types

Remote sensing retrievals of geophysical parameters employ a priori relationships between the set of parameters to be retrieved and the observables associated with them. For physically based retrievals these relationships are obtained from physical descriptions of the atmosphere, and the underlying surface, within a radiative transfer model and take the form of either model functions or retrieval databases that are used as the basis for the development of the inversion: the retrieval algorithm (e.g., Kummerow et al. 2001, 2015; Wentz and Spencer 1998; Panegrossi et al. 1998).

As such, physically-based retrievals of precipitation from passive microwave observations are critically dependent on the assumptions that went into building the retrieval databases - the relationship between the observables (e.g. passive microwave brightness temperatures) and the parameters of interest (e.g. surface precipitation rates).

The most important sources of uncertainty associated with the estimations of precipitation from passive microwave satellite observations are related to three important characteristics of precipitation: horizontal variability, vertical structure and microphysical composition of the hydrometeors within a satellite’s field of view. All these characteristics vary significantly with the type of precipitation.

Precipitating regions can be broadly classified into two major types – convective and stratiform. These two regions differ, in a fundamental way, in the vertical structure of precipitation, the latent heating and the vertical transport of sensible heat, moisture and momentum. Indeed, convective regions are characterized by strong and turbulent vertical updrafts and downdrafts while stratiform regions are associated with much gentler vertical motion. As a result, the particle growth in the two regions is governed by different microphysical processes, leading to the development of distinctly different hydrometeor populations. The spatial and temporal scales and variability of the two regions is much different as well, with the stratiform regions having larger horizontal extent with smaller spatial variability and longer lifecycle as compared to convective regions that are much smaller, isolated, highly inhomogeneous and short-lived. As a result, convective and stratiform regions have different brightness temperatures.

Advancement in satellite retrieval of precipitation cannot be achieved without: i) properly capturing the convective/stratiform variability in the retrieval databases; ii) detecting it during the retrievals (e.g. Panegrossi et al.,1998). Another important source of uncertainty comes from the initial screening that retrieval algorithms adopt to decide whether a particular observation contains rain within the satellite field-of-view.

We developed an algorithm that provides rain detection and classification based on passive microwave observations alone – PMW_CLASS. This algorithm will be used as the initial step in the KGPM precipitation retrieval algorithm to help select the most appropriate retrieval database to be used for a particular observation, selecting from a couple of different databases designed to represent the characteristics of the different precipitation types. Such ability allows improving the precipitation retrievals from passive microwave observations that are a major component of the GPM mission.

The PMW_CLASS algorithm was developed using both model simulations and observations. The precipitation detection and classification is based on the intensity and spatial variability of the Rain Index (Hristova-Veleva et al., 2013) - a multichannel combination of passive microwave observations that was designed to capture the structure of the precipitating systems in a way very similar to how radar reflectivity detects it (Figure 1). Indeed, the resemblance of the storm structure as depicted by the Rain Index to that depicted by the radar reflectivity provides the basis for designing a passive microwave classification in a way similar to the traditional radar-based convective/stratiform separation.

The algorithm was tested and further improved using radar-radiometer collocated observations by the TRMM and GPM-core missions. In this evaluation and improvement, the radar-observed reflectivity profiles and precipitation classification were used a truth.

In this presentation, we will describe the algorithm and will present the results of its validation. Furthermore, we will discuss the proposed use of the algorithm, not only as a critical component of the precipitation retrieval, but also as providing the ability to obtain convective/stratiform classification from passive microwave observations alone.

Currently precipitation type classification over the global oceans is provided only by space-based radars such the GPM’s DPR (Dual Frequency Precipitation Radar) and NASA’s CloudSat. While these radar-based observations are considered the gold standard, today there are only two of these radars flying in space. This, in addition to the very narrow radar swaths, severely limits our ability to obtain a global climatology of the oceanic precipitation types and to understand how their statistics might be changing with the changing climate. PMW_CLASS will allow such precipitation classification based on the multitude of conically-scanning radiometers that are flying today (GMI, AMSR2, and the SSMIS’s – see Fig. 2), thus greatly expanding our ability to develop climatology of these two major types of precipitation and to study their geographical variability and trends on a number of spatial and temporal scales.