UAE Radar Data Enhancment Efforts (by NCAR/RAP)

"Radar Data Quality Enhancements" is an effort to improve the quality of the radar base data fields for the EEC C-band radars of the UAE. The purpose of the Radar Data Quality Enhancements work is to improve the radar-derived rainfall estimates by enhancing the quality of the EEC radar base data through the identification and removal of certain contaminants, specifically ground clutter caused by anomalous propagation (AP) and normal propagation (NP) conditions as well as contamination from sea surface return (also called "sea clutter"). Removal of these contaminants is needed because they cause erroneous radar-derived rainfall estimates as well as errors in interpretation.

Background

Using existing software developed for use on the USA Weather Surveillance Radar-Doppler 1988 (WSR-88D) radars, modifications and improvements were made to optimize the software for use on the UAE radars. This software, called the radar echo classifier (REC) was developed with funding from the USA Radar Operations Center (ROC). During September 2002, the REC was installed within the WSR-88D system in the Open Radar Product Generator (ORPG -Saffle 1997; Reed and Cate 2001) as part of the Anomalous Propagation (AP) Clutter Mitigation Scheme (Keeler et al. 1999; Kessinger et al. 2001 and 2002; Ellis et al. 2003). Currently, the operational version of the REC on the WSR-88D detects only AP ground clutter; however, the prototype software in use at NCAR can detect additional echo types. Through approval by the ROC, with limitations that the REC only be used for civilian purposes, it has been possible to install the REC on the UAE radars. The REC deployment on the UAE radars was accomplished during October 2002.

Ground Clutter Contamination

Ground clutter contamination within radar data has two sources: NP ground clutter from stationary targets such as buildings, trees, or terrain, and AP ground clutter that arises from particular atmospheric conditions within the planetary boundary layer (i.e., temperature inversions and strong humidity gradients) that duct the radar beam to the ground. Because NP targets are stationary and "unchanging," they are relatively easy to remove with the use of clutter background maps. However, AP clutter can evolve and dissipate as atmospheric conditions change. Automatic detection and removal of the AP ground echoes is needed since a background map generally cannot include AP clutter locations. In the UAE, conditions conducive to anomalous propagation are most severe during the summer months.

Radar base data contamination from the sea surface occurs when the radar beam grazes the sea surface or when strong return is captured through the radar sidelobes. The amount of sea clutter detected by radar is dependent upon the sea state. The sea state is influenced by various factors that include the wave height, the wind speed, the wind direction relative to the direction of the radar beam, the fetch over which the wind has been acting, the direction of the waves relative to the direction of the radar beam, whether the tide is incoming or outgoing, and the presence of swells or waves, among others (Skolnik, 1980). Radar system characteristics can also have some effect on the amount of sea clutter detected. Further, when AP conditions occur, the distance over which sea clutter is detected can be elongated considerably.

The REC (Kessinger et al. 2003) is a data fusion system that uses "fuzzy-logic" techniques (Kosko, 1992) to classify the type of scatterer measured by Doppler radar systems. Using the three moments of radar base data (reflectivity, radial velocity and spectrum width) as input, various detection algorithms are formulated to make a classification. Two REC algorithms have been installed and tested on the UAE EEC radars: the AP detection algorithm (APDA) that detects regions of AP ground clutter return and the sea clutter detection algorithm (SCDA) that detects regions of sea clutter return.

Methodology

To develop the algorithms for the REC, several steps are required to prepare the data. The first step is to "truth" the data such that the characteristics of each scatterer type can be examined to find those that discriminate one scatterer from another. The characteristics of each scatterer are examined through the "feature fields". The feature fields are quantities (e.g., mean, median, standard deviation) that are derived from the radar base data fields of reflectivity, radial velocity and spectrum width. Next, histograms of the feature fields are made to examine the distribution of values between scatterer types. Finding those features that discriminate one scatterer type from another is a key task. Once the feature fields are selected for each scatterer type, "membership functions" are devised, based upon the distribution of values.

The REC Algorithms

Anomalous Propagation Detection Algorithm (APDA)
The feature fields used by the APDA include: the texture of the reflectivity field (TDBZ), the median radial velocity field (MVE), the median spectrum width field (MSW), the standard deviation of the radial velocity field (SDVE), the SPIN field, the vertical difference of the reflectivity (GDZ), and the GDZ field divided by the sine of the difference of the elevation angles (RSINZ). Equal weights are used. The APDA is applied over both land and sea regions.

Sea Clutter Algorithm (SCDA)
The feature fields used by the SCDA include: the texture of the reflectivity field (TDBZ), the median spectrum width field (MSW), the standard deviation of the radial velocity field (SDVE), the SPIN field, the vertical difference of the reflectivity (GDZ), the range-weighted GDZ field, and the GDZ field divided by the sine of the difference of the elevation angles (RSINZ). The SCDA is applied only over oceanic regions.