Use of Three-Dimensional Reflectivity Structure for Automated Detection and Removal of Nonprecipitating Echoes in Radar Data

Matthias Steiner Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey

Search for other papers by Matthias Steiner in
Current site
Google Scholar
PubMed
Close
and
James A. Smith Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey

Search for other papers by James A. Smith in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

This study aims at assessing the potential of anomalous propagation conditions to occur, reviews past attempts to mitigate ground clutter contamination of radar data resulting from anomalous signal propagation, and presents a new algorithm for radar data quality control. Based on a 16-yr record of operational sounding data, the likelihood of atmospheric conditions to occur across the United States that potentially lead to anomalous propagation of radar signals is estimated. Anomalous signal propagation may lead to a significant contamination of radar data from ground echoes normally not seen by the radar, which could result in serious rainfall overestimates, if not recognized and treated appropriately. Many different approaches have been proposed to eliminate the problem of regular ground clutter close to the radar and temporary clutter resulting from anomalous signal propagation. None of the reported approaches, however, satisfactorily succeeds in the case of anomalous propagation ground returns embedded in precipitation echoes, a problem that remains a challenge today for radar data quality control. Taking strengths and weaknesses of past approaches into consideration, a new automated procedure has been developed that makes use of the three-dimensional reflectivity structure. In particular, the vertical extent of radar echoes, their spatial variability, and vertical gradient of intensity are evaluated by means of a decision tree. The new algorithm appears to work equally well in situations where anomalous propagation ground returns are either separated from or embedded within precipitation echoes. Moreover, sea clutter echoes are identified as not raining and successfully removed.

Corresponding author address: Dr. Matthias Steiner, Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08540. Email: msteiner@princeton.edu

Abstract

This study aims at assessing the potential of anomalous propagation conditions to occur, reviews past attempts to mitigate ground clutter contamination of radar data resulting from anomalous signal propagation, and presents a new algorithm for radar data quality control. Based on a 16-yr record of operational sounding data, the likelihood of atmospheric conditions to occur across the United States that potentially lead to anomalous propagation of radar signals is estimated. Anomalous signal propagation may lead to a significant contamination of radar data from ground echoes normally not seen by the radar, which could result in serious rainfall overestimates, if not recognized and treated appropriately. Many different approaches have been proposed to eliminate the problem of regular ground clutter close to the radar and temporary clutter resulting from anomalous signal propagation. None of the reported approaches, however, satisfactorily succeeds in the case of anomalous propagation ground returns embedded in precipitation echoes, a problem that remains a challenge today for radar data quality control. Taking strengths and weaknesses of past approaches into consideration, a new automated procedure has been developed that makes use of the three-dimensional reflectivity structure. In particular, the vertical extent of radar echoes, their spatial variability, and vertical gradient of intensity are evaluated by means of a decision tree. The new algorithm appears to work equally well in situations where anomalous propagation ground returns are either separated from or embedded within precipitation echoes. Moreover, sea clutter echoes are identified as not raining and successfully removed.

Corresponding author address: Dr. Matthias Steiner, Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08540. Email: msteiner@princeton.edu

1. Introduction

Quality control is essential to meaningful radar-based rainfall estimation. Radar echoes may be contaminated by nonmeteorological echoes that need to be identified and removed before rainfall estimation. This is particularly true for operational applications such as precipitation nowcasting and (flash) flood warning. A well-trained person may successfully recognize nonmeteorological contamination in radar echoes, such as ground clutter or anomalously propagated ground returns (called “AP” or “Anaprop” echoes). For offline case studies, manual editing of the data may be feasible and appropriate; however, for operational applications automated procedures need to be used.

Harrison et al. (2000) present recent efforts under way in the United Kingdom that show how extensive quality control may effectively reduce the root-mean-square (rms) difference between gauge-measured and radar-estimated rainfall amounts. Joss and Lee (1995) discuss elaborate procedures in place for operational radar data processing in Switzerland, where extensive data quality control is combined with terrain-based visibility and vertical precipitation structure, and gauge adjustments to achieve the most reliable rainfall estimates. Fulton et al. (1998) report on the data quality control and rainfall estimation procedures of the operational radar network in the United States. Despite elaborate and sophisticated efforts in data quality assurance, however, evaluations by Smith et al. (1996) show that anomalously propagated ground returns remain a serious problem, especially for situations where AP is embedded in precipitation echoes.

The aim of this paper is threefold: to investigate the potential of anomalous propagation conditions to occur throughout the United States from an atmospheric perspective (section 2), review past efforts in dealing with ground clutter and AP contamination in radar data (section 3), and present and discuss a new approach for automated radar data quality control (sections 4 and 5).

2. Anomalous propagation of radar signals

a. Refractive index and signal propagation

At microwave frequencies, the propagation of electromagnetic signals is influenced by atmospheric conditions. A commonly used quantity to describe the propagation behavior of electromagnetic signals is the index of refraction n, or the refractivity N, which can be approximated by
i1520-0426-19-5-673-e1
where p is the barometric pressure in millibars, e the partial pressure of water vapor in millibars, and T the absolute temperature in kelvins (Gossard 1977; Skolnik 1980; Babin 1996; Fabry et al. 1997).

It is the vertical gradient of the refractivity within the lowest several hundred meters above the ground that is especially important for characterizing radar signal propagation (e.g., Pratte et al. 1995). A decrease in atmospheric refractivity with altitude, dN/dh, tends to bend the radar rays so as to extend coverage beyond that expected with a uniform atmosphere. This abnormal propagation of electromagnetic waves is called anomalous propagation. Four basic modes of propagation are distinguished:

  • subrefractiondN/dh > 0 m−1,

  • normal refraction0 > dN/dh > −0.0787 m−1,

  • superrefraction−0.0787 > dN/dh > −0.157m−1, and

  • trapping or ducting dN/dh < −0.157 m−1.

Trapping or ducting is the most severe case of anomalous signal propagation, and results in ground returns (AP echoes) from locations where the radar beam intersects the ground or objects at the earth's surface.

In order to propagate energy within the duct, the angle the radar ray makes with the duct should be small, usually less than a degree (e.g., only the lowest elevation scans of surface-based radar are affected). Only those radar rays launched nearly parallel to the duct will be trapped. Atmospheric ducts are generally of the order of tens to hundreds of meters (Gossard 1977; Cook 1991; Babin 1996; Brooks et al. 1999). A simplified approximate model of propagation in atmospheric ducts (Skolnik 1980) predicts a maximum wavelength λmax that can be propagated in a surface duct of depth d as given by
λmaxdndh1/2d3/2
where λmax, dh, and d are in the same units (e.g., meters). For an operational Weather Surveillance Radar-1988 Doppler (WSR-88D) of the Next Generation Weather Radar (NEXRAD) network (Heiss et al. 1990; Baer 1991; Crum et al. 1998) with wavelength of 10 cm (S band), the duct must be at least 22 m thick in order for trapping to occur. Often only parts of the radar beam may be trapped.

A duct is produced when the index of refraction rapidly decreases with height. In order to achieve this, the temperature must increase and/or the humidity (water vapor content) must decrease with height. Temperature inversions must be very pronounced in order to produce superrefraction, while water vapor gradients are more effective than temperature gradients alone (Fabry et al. 1997). A common cause of ducting is the movement of warm dry air from land over cooler bodies of water (e.g., ocean), causing a temperature inversion in the boundary layer. At the same time, moisture is added by evaporation from the water surface, producing a moisture gradient (Skolnik 1980; Puzzo et al. 1989). Evaporation ducts are common just above the surface of the sea, where air may become saturated by evaporation from the sea surface. Over land, ducting is often caused by radiational cooling during clear nights, particularly in the summer when the ground is moist. Thus, over land, ducting is most noticeable at night and tends to disappear during the warmest part of the day (e.g., Moszkowicz et al. 1994). Superrefraction or ground ducts may also be produced by the diverging downdraft under a thunderstorm and resulting gust fronts. The relatively cool air, which spreads out from the base of a thunderstorm, may produce a temperature inversion within the lowest, possibly several hundred meters. The moisture gradient along the outflow boundary is also appropriate for the formation of a duct. The conditions favorable for the formation of a thunderstorm duct are relatively short-lived and have timescales on the order of 30 min to 1 hr, although in extreme cases such conditions may last for hours (Weber et al. 1993).

b. Climatological assessment of vertical refractivity gradients

Using a 16-yr record of operational sounding data (1973–88), the potential of anomalous propagation conditions to occur throughout the continental United States is assessed climatologically. Similar studies, for example, have been conducted by Bech et al. (2000) using soundings for Mediterranean coastal sites, Babin (1996) using helicopter-based refractivity measurements off the coast of Wallops Island (Virginia), and Gossard (1977) using airmass analyses. For each sounding of the dataset, the average refractivity gradient within the lowest 500 m above ground level (AGL) is determined. (The maximum gradient might be more relevant for the radar signal propagation problem; however, the limited and variable vertical resolution of the operational sounding data may result in questionable maximum gradient values.) These values are then compiled into a climatology of average refractivity gradients for each operational sounding station and used to study the likelihood of atmospheric conditions across the United States that are susceptible to anomalous propagation of radar signals. The sounding-based climatology will highlight large-scale temperature inversions and moisture gradients, yet only by chance capture conditions favorable to anomalous propagation produced by thunderstorm outflow boundaries. Moreover, relatively thin layers of strong vertical gradients may cause anomalous propagation, but the operational sounding data (variable resolution of one to several hundred meters) do not resolve tens of meters in the vertical. High-resolution refractivity profiles may be obtained, for example, from detailed atmospheric measurements aboard an airplane [e.g., Babin (1996) used a helicopter] or combined profiler and radio acoustic sounding systems (RASS) (Gossard et al. 1995). Lidar-based observations of temperature and water vapor (e.g., Eichinger et al. 1993, 1999) may provide alternative methods for obtaining high-resolution refractivity information. Fabry et al. (1997) also showed that radar-based phase measurements of ground targets can be used to reveal the spatial, near-surface structure of the index of refraction.

There are many areas of the continental United States that have conditions at least favorable for anomalous propagation conditions to occur on a regular basis, as will be discussed below. For example, the southwestern part of the United States, particularly southern California, is especially prone to this problem (Gossard 1977; Pappert and Goodhart 1977; Babin and Rowland 1992; Burk and Thompson 1997). Significant local, regional, diurnal, and seasonal differences are found among the different sounding stations.

1) Analyses of 1200 UTC soundings—Nighttime

The 1200 UTC soundings taken across the contiguous United States reflect nighttime conditions, ranging from 0400 LST at the west coast to 0700 (about sunrise) at the east coast. Figure 1 shows the distribution of average refractivity gradients within 500 m AGL for each operational sounding station. The critical categories of “superrefraction” and “trapping” are shaded in gray and black, respectively, and the fraction of the box covered indicates the percentage of soundings that exhibit such atmospheric conditions. For simplicity, we discuss seasonal trends only, although we noticed significant variability from month to month.

The general nighttime pattern revealed by Fig. 1 is that conditions of superrefraction may occur anywhere throughout the United States. During the winter season (Fig. 1a), superrefractive propagation conditions occur most likely south of 40° latitude, and particularly along the coastlines. For those areas, on average, superrefractive conditions may occur at least once a week, and maybe twice or three times that for southern California. The maximum likelihood of superrefractive conditions is displayed during the summer (Fig. 1c), when chances for such conditions to occur exceed 20% throughout most of the United States and are larger than 30% for southern California and most of the eastern seaboard states. The spring (Fig. 1b) and fall seasons (Fig. 1d) also show widespread conditions of superrefraction at least once a week, with local maxima in excess of 30% generally south of 40° latitude and throughout states bordering the ocean. Significant trapping conditions (at least once a week) occur mainly in southern California (e.g., San Diego), essentially throughout the year.

2) Analyses of 0000 UTC soundings—Daytime

The 0000 UTC soundings reflect conditions in the late afternoon, ranging from 1600 LST at the west coast to 1900 (about sunset) at the east coast. The general daytime pattern revealed for the four seasons, as shown in Fig. 2 (similar to Fig. 1), is more structured than the nighttime conditions depicted in Fig. 1. Conditions of superrefraction at least once a week are seen throughout the year in the coastline states. Again, southern California displays a maximum likelihood of superrefractive conditions to occur twice, if not three times a week, and trapping conditions approximately 20% of the time (e.g., San Diego).

An interesting seasonal feature is an enhanced likelihood (>20%) of superrefractive conditions spreading up the Missouri, Mississippi, and Ohio River valleys, reaching its maximum spatial extent in the summer (Fig. 2c), before retreating again. The corresponding minimum is observed in the winter season (Fig. 2a). In concert with this seasonal fluctuation of superrefractive conditions in the Missouri, Mississippi, and Ohio River valleys, there is also an increase in superrefractive conditions throughout the states along the east and west coasts of the United States.

The soundings launched from Huron, South Dakota, display an unusually high percentage of superrefractive and trapping conditions during the summer (Fig. 2c) and fall (Fig. 2d) seasons compared to their neighboring stations. The operational soundings from this location were discontinued in November 1994 and since moved to Aberdeen, South Dakota.

3. Review of approaches to mitigate clutter problems

There are various levels where the problem of ground clutter and AP echo contamination in radar data may be approached (e.g., Joss and Wessels 1990; Keeler and Passarelli 1990; Pratte et al. 1995), namely:

  • the radar installation (site, hardware),

  • the data processing (before and/or after recording), and

  • through comparison with other data sources.

The first may be considered a static approach, while the latter two are dynamic and more easily modified. The main focus of this study is on the processing of archived data; however, we review a variety of approaches for mitigation of ground clutter and AP echoes.

Visual inspection of radar volume scan reflectivity data reveals that ground returns resulting from anomalous signal propagation have little vertical extent and tend to clutter the lowest elevation sweeps only (generally below 2° elevation), depending on the vertical gradient of the refractive index (Sekhon and Atlas 1972; Johnson et al. 1975; Sirmans and Dooley 1980; Moszkowicz et al. 1994; Lee et al. 1995). Anomalously propagated echoes, similar to regular ground clutter close to the radar, decorrelate rapidly in space and are spatially heterogeneous to the greatest degree (Joss and Wessels 1990; Joss and Lee 1993; Pratte et al. 1993). Thus, AP signatures may be recognized in reflectivity data by their larger spatial variability than precipitation echoes. In contrast to regular (stationary) ground clutter, which exhibits a longer time correlation than weather echoes (Tatehira and Shimizu 1978, 1980; Sirmans and Dooley 1980), AP ground returns can appear much like precipitation, exhibiting growth, decay, and motion similar to that of rainstorms (Johnson et al. 1975; Weber et al. 1993).

For ground-based radar, signals returned from radar beams intersecting the ground should exhibit radial Doppler velocity values close to zero (not for sea clutter though), dependent on vegetation cover, wind, and antenna rotation (Rinehart 1991). The Doppler spectrum width should be small as well (Hamidi and Zrnic 1981; Joss and Wessels 1990). From visual inspection of radar volume scan velocity data, land-based AP returns in the lowest elevation sweep can clearly be recognized by their near-zero Doppler velocity. Contamination of AP in the second-lowest elevation sweep, however, may exhibit velocities similar to precipitation echoes, although the reflectivity signature indicates AP. The distribution of spectrum width data is significantly broader than the radial Doppler velocity for AP echoes (e.g., see Fig. 1 of Steiner et al. 1999b) and thus appears to be less useful for identification purposes.

a. Radar installation

1) Choice of radar site and hardware

A clever siting of the radar may be very effective in minimizing clutter contamination in the desired field of view (Smith 1972; Mann et al. 1986; Joss and Wessels 1990; Joss and Lee 1995). However, the placing of the radar antenna involves a compromise between extending the horizon and minimizing clutter contamination. Preference may be given to an elevated radar site (e.g., high tower or top of a mountain), because modern signal processing techniques are increasingly capable of mitigating many problems caused by clutter contamination, but nothing can be done to detect precipitation blocked from the radar view at a low site. A smart choice of radar characteristics (wavelength, antenna, polarization, Doppler, system stability, scan strategy, etc.) may also help in reducing the clutter problem (Skolnik 1980; Pratte and Keeler 1986; Joss and Wessels 1990; Keeler and Passarelli 1990; Rinehart 1991; Smith 1993, 1998). However, to balance application needs versus costs, one may have to compromise further. For example, use of longer wavelengths avoids problems caused from attenuation by precipitation; however, the relationship between precipitation and ground clutter worsens with increasing wavelength. Moreover, for a given size of antenna, a shorter wavelength will result in a narrower beam and thus better spatial resolution and reduced ground clutter. Also, the cost increases roughly proportionally to the weight of the antenna, that is, with the third power of the wavelength (Joss and Wessels 1990).

2) Signal processing before data recording

For noncoherent radar, a check on the temporal variability (fluctuation rate) of echoes from pulse to pulse (i.e., “Doppler simulation”) or the (auto)correlation in time have been suggested by Reid (1970), Johnson et al. (1975), Geotis and Silver (1976), Aoyagi (1978), Tatehira and Shimizu (1978, 1980), Sirmans and Dooley (1980), Passarelli (1981), Joss and Wessels (1990), Coveri et al. (1993), Michelson and Andersson (1995), and Haddad et al. (2000). For semicoherent radar, Andersson (1993) recommends a check on the agreement of velocity estimates based on staggered pulse repetition frequency (PRF) processing (at least two different PRFs are necessary). Pulse-pair processing (Anderson 1981; Hamidi and Zrnic 1981), time domain filtering (Mann et al. 1986; Michelson and Andersson 1995; Pratte et al. 1995), and frequency domain filtering (Passarelli et al. 1981; Schmid et al. 1991; Torres and Zrnic 1999) are choices for a coherent radar. Joss and Lee (1993, 1995), Lee et al. (1995), and Archibald (2000) advocate “clutter detection” (and rejection) rather than “clutter suppression” (which may result from time or frequency domain filtering) by means of a sophisticated decision tree, making use of high spatial resolution radar information.

b. Processing of archived data

For many applications, the user may not be able to influence the data recording and, therefore, has to resort to an intelligent processing of the archived data. Many different approaches have been suggested; for example, checks on the spatial (horizontal and vertical) and temporal continuity of reflectivity echoes (Hogg 1978; Smith 1990), and analysis of the horizontal and vertical reflectivity gradients (Mueller and Sims 1975; Riley and Austin 1976; Collier et al. 1980; Lee et al. 1995), including echo tops (Johnson et al. 1975; Moszkowicz et al. 1994; Rosenfeld et al. 1995). Other approaches focus on the texture (spatial variability) of echo patterns based on signal-to-noise ratio, reflectivity, Doppler velocity and spectral width, or differential reflectivity fields (Hall et al. 1984; Smith 1990; Joss and Wessels 1990; Joe 1991; Giuli et al. 1991; Pratte et al. 1993; Cornelius 1994). Probabilistic analyses, using multiple parameters as input to neural network or fuzzy logic procedures, have been explored more recently (Cornelius 1994; Grecu and Krajewski 1999, 2000; VanAndel 2001; Kessinger et al. 2001). The advent of multiple polarization radar observations, particularly the correlation coefficient between horizontally and vertically polarized backscatter signals and differential phase shift parameters, brought considerable skills to the detection of AP and ground clutter contamination (Blackman and Illingworth 1993; Zrnic and Ryzhkov 1996; Ryzhkov and Zrnic 1998; Collier 2000). Some of these data quality tests based on polarimetric parameters could be implemented also at the signal processing level before the data are recorded, similar to Doppler-based procedures.

c. Comparison with other data sources

Some more elaborate approaches embrace additional information from independent sources for assessing quality of the radar data. For example, Mann et al. (1986), Schmid et al. (1991), Lee et al. (1995), and Joss and Lee (1995) suggest using adaptive clutter or clutter residue maps, albeit as a last resort rather than a primary check. Moores and Harrold (1975), Hogg (1978), Delrieu et al. (1995), and Archibald (2000) propose using digital elevation data and code to predict the beam pattern for normal and anomalous propagation. Johnson et al. (1995), Pratte et al. (1995), and Fabry et al. (1997) suggest assessing the atmospheric conditions through direct or indirect measurements of the refractive index gradient.

Atkinson and James (1991) and Klingle-Wilson et al. (1995) evaluate data from multiple radar covering the same area, incorporating also rain gauge and/or satellite data (Fiore et al. 1986). Pamment and Conway (1998) discuss a probabilistic scheme used in the United Kingdom that combines synoptic reports, satellite infrared data, lightning data, and AP echo climatology for radar data quality control.

4. A new algorithm

The new radar data quality control algorithm, designed for single volume-scanning radar, makes use of the three-dimensional reflectivity structure. Radial Doppler velocity and spectrum width information, although readily available for many modern radar systems, is not used as part of the algorithm, which will keep the amount of data processing to a minimum (particularly relevant for operational applications). The algorithm builds upon three key parameters, namely, the vertical extent of radar echoes (ECHOtop), the spatial variability of the reflectivity field (SPINchange), and the vertical gradient of reflectivity (vertGRAD). Steiner et al. (1999b) evaluated several other parameters as well, but these three appear to be the most useful ones. In addition, gaps in rainfall echo areas that are potentially created by the algorithm in situations with AP echoes embedded in precipitation will be filled using echo information from above. The algorithm is executed in polar space to remain at the level of the data recording (i.e., how the radar sees its environment). Interpolation of data to a Cartesian grid may simplify subsequent data management and processing, but would likely introduce undesired range-dependent artifacts (e.g., Trapp and Doswell 2000).

The ECHOtop parameter indexes the highest elevation sweep (tilt) that contains reflectivity echoes in excess of a minimum intensity (REFLthresh = 5 dBZ) either directly above the base scan (first tilt) pixel of interest or its surrounding eight neighbors. This check on the pixel's neighbors is included to reduce edge effects accounting for potentially tilted storm cells.

The SPINchange parameter indicates the number of reflectivity fluctuations larger than 2 dBZ within an 11 (azimuth) by 21 (radial) pixel window, expressed as a percentage of all possible “spin” changes. A reflectivity increase (decrease) in radial direction from one pixel to the next by more than 2 dBZ would cause the spin to point up (down). Reflectivity fluctuations smaller than 2 dBZ are deemed insignificant and thus have no effect on the spin setting. The window is investigated in radial direction, beam by beam. The SPINchange parameter represents the sum of spin changes found within a window centered on the scrutinized pixel, normalized by all potentially possible spin changes.

The vertGRAD parameter gauges the vertical reflectivity difference between pixels directly above each other of the lowest two elevation sweeps, normalized by the elevation angle difference. The dimension of the vertGRAD parameter is dBZ degree−1.

Figure 3 shows the decision tree of the algorithm. The sequence of tests is applied to each pixel of the polar-spaced base scan to examine whether that pixel should be kept, removed, or potentially replaced. The algorithm removes echoes that are weaker than REFLthresh as a first step. Then, echoes without a significant vertical extent (i.e., ECHOtop equal base scan) are removed as well. (This might be a problem at far ranges, where the radar potentially overshoots precipitating cloud systems and only the lowest elevation sweep contains significant echoes.) The remaining pixels (i.e., echoes exhibiting some vertical depth) are subsequently checked on their SPINchange parameter and kept if that value would be less than an intensity-dependent SPINthresh, defined as SPINthresh = 8 − (Zpixel − 40)/15, where Zpixel is the reflectivity (in dBZ) of the pixel evaluated. (SPINthresh was fine-tuned such that this filter would be more aggressive for higher intensities but less so for weaker echoes.) Reflectivity pixels that fail the SPINchange test are investigated further before removal. If the vertGRAD for these pixels is less than or equal to 10 dBZ degree−1, they are kept despite a large SPINchange parameter. Such a threshold is very similar to the 12 dBZ degree−1 suggested by Lee et al. (1995), and also consistent with the results of Mueller and Sims (1975), who found 80% of the vertical gradients observed in severe storms to be less than 12 dBZ km−1.

The sequential order of the tests applied in the decision tree takes care of the most obvious unwanted echoes first and subsequently focuses on the more problematic areas that may require additional computational effort to be tackled. The fine-tuning of the parameter settings (e.g., for SPINchange and SPINthresh) was done empirically and does not bear any physical basis.

The ECHOtop test will remove most of the AP and sea clutter echoes that are separated from precipitation, because of a lack of vertical extent of those echoes. However, boundary layer features, such as gust fronts, will be removed as well. The SPINchange parameter grasps the spatial variability of AP echoes well, which is particularly useful for AP echoes embedded in precipitation, where the ECHOtop parameter is clueless. The SPINchange parameter tends to be too erosive at the storm cell boundaries, which is why the vertGRAD test was introduced to limit echo removal that is likely precipitation.

In this form, the algorithm successfully identifies and removes AP echoes. However, AP echoes that were embedded in precipitation will be removed, leaving a gap behind. Therefore, as a final processing step, the algorithm checks if there are echoes directly above, in the next higher (second tilt) elevation sweep, that exceed the REFLthresh, and if so, will use those pixel values to fill the gaps in the base scan. There is an option, to check on the third tilt instead of the second, which works better, particularly in situations of severe AP contamination, where even the second tilt may be affected. However, use of this option may result in loss of precipitation echoes at far ranges, where the third tilt might overshoot cloud tops. This gap-filling procedure is a first approximation only, and future improvements will have to consider vertical reflectivity profile information to downward extrapolate reflectivity measured aloft (see, e.g., Joss and Pittini 1991; Kitchen et al. 1994; Andrieu and Creutin 1995; Andrieu et al. 1995; Joss and Lee 1995; Amitai 1999; Vignal et al. 1999, 2000; Seo et al. 2000), or potentially otherwise obtain it sideways (e.g., Sánchez-Diezma et al. 2001), and fill the base scan gaps.

5. Example case studies and sensitivity analysis

Five typical situations were chosen to exemplify the performance of the algorithm:

  • AP echoes near or separated from precipitation (Figs. 4 and 5),

  • AP echoes embedded in precipitation (Fig. 6),

  • sea clutter (Fig. 7),

  • strong clear air and boundary layer echoes (Figs. 7 and 4), and

  • pure precipitation (Fig. 8).

For each of these examples, the unedited velocity and reflectivity fields of the base scan (first tilt) are shown together with the reflectivity of the second tilt. In addition, three different stages of the edited reflectivity field are shown: namely, before filling the gaps in rainfall echo areas caused by the removal of pixels (stage 4), and after filling those gaps by either using reflectivity information from the second (stage 5a) or third tilt (stage 5b) above.

The examples shown in Figs. 4, 5, and 6 are related to anomalous propagation of radar signals caused by thunderstorm outflow boundaries, which are not captured by the sounding climatology presented in section 2. Figure 7 shows a typical example of anomalous signal propagation caused by a nighttime temperature inversion, which is well captured by the sounding climatology. Figure 8 shows normal propagation conditions, also represented by the climatology discussed in section 2.

For a quantitative evaluation of the effect of the quality control algorithm on radar-based rainfall estimates, reflectivity Z (mm6 m−3) is converted to rain rate R (mm h−1) using the relationship Z = 300R1.4 of the NEXRAD precipitation processing scheme (Fulton et al. 1998). We do not consider potential biases compared to rain gauge accumulations or problems with hail contamination for this study. These factors are extensively discussed in Steiner et al. (1999a, and references therein). Rainfall estimates are based on the base scan reflectivity field without consideration of the vertical profile for extrapolating reflectivity measured aloft down to the surface. Vertical profile corrections are extensively discussed in Vignal et al. (2000, and references therein).

a. AP echoes near or separated from precipitation

A thunderstorm passing over the WSR-88D site at Amarillo, Texas (KAMA), on 25 May 1994 caused widespread severe AP echoes to occur behind the storm. Figure 4 shows several storm cells triggered along a gust front that was moving in a southerly direction. The outflow from these storms left atmospheric conditions behind that caused the radar signals at the lowest elevations to be trapped in the boundary layer, resulting in widespread severe AP echoes (Fig. 5). The AP echoes can easily be recognized by their zero velocity signature, high spatial variability in reflectivity, and essentially no vertical extent of the echoes—for example, those seen to the north of the radar in Fig. 4 and covering most of the area centered on the radar in Fig. 5.

The quality control algorithm succeeds in removing most of the AP echoes but keeps the rainfall echoes. Some smaller contaminated echo areas remain, owing to a severe AP situation that caused the second tilt to be affected. Visually, using the third tilt option to fill the gaps in the base scan would be slightly more successful in this particular situation. From a quantitative perspective (see Table 1), the removal of the AP echoes reduced the unconditional, area-average rainfall rate from 1.0 and 1.1 mm h−1 for the unedited data to 0.4 mm h−1 and essentially zero for the edited data shown in Figs. 4 and 5, respectively. The effect of the gap-filling procedure was not significant.

b. AP echoes embedded in precipitation

A major thunderstorm passing over the WSR-88D site at St. Louis, Missouri (KLSX), on 7 July 1993 resulted in severe AP contamination embedded within precipitation echoes (Fig. 6). The AP echoes can be recognized by their near-zero velocity and highly variable reflectivity pattern, beyond a range of 100 km to the east from the radar. Because they are embedded within rainfall echoes, the vertical extent of the AP contamination is obscured. The reflectivity pattern of the second tilt (Fig. 6c) hints that the AP contamination may extend up to this level, although the velocity signatures are no longer zero (not shown). This remains the most challenging situation to deal with, particularly for operational applications such as flash flood forecasting and warning (Smith et al. 1996). The 7 July 1993 storm was a major rain event of the Mississippi River flood episode during the summer of 1993 (Kunkel et al. 1994; Gumley and King 1995; Giorgi et al. 1996; Arritt et al. 1997).

The quality control procedure successfully flags and removes the most severe AP echoes (Fig. 6d). However, because the sweeps above the base scan may be affected by AP as well, filling in echoes from above will not entirely solve the problem. From a quantitative perspective, AP contamination severely affects the rainfall estimates, as shown by the dramatic reduction of area-average rain rate from the unedited to the edited data (Table 1). Moreover, filling the gaps created by the pixel removal is crucial under these circumstances. The detail of whether the second or third tilt is used is less significant.

c. Sea clutter

A strong nighttime inversion on 9 May 1998 caused the signals emitted at the lowest elevation angle by the Houston, Texas, WSR-88D (KHGX) to be trapped in the boundary layer over the ocean, resulting in severe sea clutter to the southeast of the radar at ranges exceeding 100 km (Fig. 7). The radial Doppler velocity is not zero, as expected for land-based ground clutter, reflecting instead the wave motion at the ocean surface. The reflectivity pattern of sea clutter is much smoother than for ground clutter or AP echoes over land, resembling more closely real precipitation echoes. However, the second tilt clearly shows that these echoes over water have no depth (Fig. 7c) and are thus not representing precipitation but rather sea clutter.

The quality control algorithm picked up on the lack of vertical depth and successfully removed those sea clutter echoes. There were essentially no gaps to be filled and thus it did not make any difference whether the second or third tilt option was used. The removal of the sea clutter echoes clearly affected the rainfall estimates, reducing them to near zero (Table 1).

d. Strong clear air and boundary layer echoes

The nighttime inversion depicted in Fig. 7 resulted not only in trapping of radar signals over the ocean (sea clutter), but also caused strong clear air returns over land to the northwest of the Houston, Texas, WSR-88D. These widespread clear air echoes exhibited little variability in reflectivity (other than wavelike structures around 100-km range); however, they reached appreciable intensity that would result in noticeable surface rainfall estimates if not identified as clear air echoes and removed.

The quality control algorithm was not able to completely remove all of the clear air echoes, because the second tilt contained a significant amount of such echoes. The remaining echoes, however, contribute very little to the estimated surface rainfall (Table 1).

The high sensitivity of the WSR-88D may reveal significant structures, such as gust fronts (Fig. 4) or wind-induced rolls (e.g., sea breeze), in the otherwise clear boundary layer. These features of the boundary layer generally have little vertical depth and thus are removed by the ECHOtop criterion of this quality control algorithm.

e. Precipitation

Figure 8 shows a typical summer mid-afternoon rainfall pattern, with scattered multicellular thunderstorms, as observed by the Melbourne, Florida, WSR-88D (KMLB). The quality control algorithm should leave these echoes untouched. From a visual perspective, this is the case (except for weak clear air echoes close to the radar). Close inspection, though, may reveal a few isolated spots that were retouched by the algorithm. Quantitatively, application of the quality control algorithm caused a slight reduction of the unconditional area-average rain rate but left the conditional rain rate the same (Table 1). This situation with scattered, multicellular storms represents a worst-case scenario in terms of potential “edge effects” for the algorithm.

f. Sensitivity analysis

The algorithm builds upon several quality control filters that are sequentially applied by means of a decision tree. Table 1 highlights the effect of adding various stages of data editing to the radar echo area and rain-rate estimates. Thresholding of the reflectivity field (stage 1) reduced the echo area (and, consequently, the unconditional rain rate) of the examples discussed by approximately 10%, but left the conditional rain rate essentially untouched. Application of the vertical echo depth filter (stage 2) clearly affected the examples with clutter separated from precipitation echoes, removing up to 30% or more of the echo area. The spatial variability check (stage 3) flagged another significant amount of echo area; however, some of that was prevented from removal by the vertical echo gradient check (stage 4). The SPINchange parameter test (stage 3) clearly demonstrated skill in tackling the embedded AP contamination for the St. Louis example (Fig. 6), as seen from the drastic reduction in average rain rate (Table 1). The process of filling in gaps (stage 5), potentially created by the previous stages, had some, albeit minor, effect as well.

And, as a final note, the effect of the different filters applied to the radar data was much more significant than the effect of fine-tuning the various parameter settings for the filters. For example, a wide range of value settings tested for the SPINchange parameter affected the echo area by a few percent only (not shown).

6. Summary and conclusions

This study was aimed at assessing the potential of anomalous propagation conditions to occur, reviewing past attempts to mitigate a ground return contamination in radar data resulting from anomalous signal propagation, and presenting a new algorithm for radar data quality control.

A 16-yr record of operational sounding data collected across the United States has been analyzed to assess the likelihood of atmospheric conditions that potentially lead to anomalous propagation of electromagnetic radar signals. This evaluation provides a lower bound on the susceptibility to anomalous signal propagation, because the soundings likely represent synoptic-scale temperature inversions and moisture gradients, and capture thunderstorm outflow boundary conditions only by chance. The analyses show that there are many locations across the country that experience atmospheric conditions leading to anomalous signal propagation on a regular basis. Coastal areas, including southern California, are particularly prone to this problem. For other areas, anomalous propagation conditions may be limited to nighttime or vary by season.

Anomalous propagation ground returns may resemble precipitation echoes, exhibiting growth, motion, and decay similar to that of rain storms. However, a contamination of radar data by anomalous propagation echoes is mostly limited to the lowest scan or two (e.g., below 2° elevation). Moreover, these echoes may easily be recognized on the basis of their great spatial variability of reflectivity (including differential reflectivity) and near-zero radial Doppler velocity. This is particularly true for anomalous propagation echoes over land, but not so over water.

Many different ways have been attempted to clean up radar data from ground clutter and particularly anomalous propagation echo contamination. Several of these past efforts were able to remove ground clutter and echo contamination caused by anomalous signal propagation, as long as those echoes were not embedded within rainfall echoes. This latter, worst-case scenario remains a challenge today.

A new algorithm has been presented that makes use of the three-dimensional radar reflectivity structure. The key parameters of this radar data quality control algorithm are based on the vertical structure and extent of the echoes, and their spatial variability. By means of a decision tree, the various parameters are evaluated for each pixel of the polar space base scan to determine whether that pixel would represent real rainfall or not. The new algorithm works equally well in situations where anomalous propagation ground returns are either separated from or embedded within precipitation echoes. Moreover, sea clutter echoes are identified as not raining and removed. However, the algorithm does not handle strong and widespread clear air echoes well. In addition, boundary layer features, such as gust fronts (although not raining), will be removed as well, which may be unfortunate for applications that are geared toward forecasting new storm cell formation that may be triggered along gust fronts (e.g., Wilson and Schreiber 1986; Wilson et al. 1998).

The algorithm works well with WSR-88D data, but may need fine-tuning for data collected by other radars or a different scanning strategy. An extensive comparison of several radar data quality control algorithms is under way (Robinson et al. 2001).

Acknowledgments

The sounding data used in this study were compiled and quality controlled by A. Allen Bradley of the University of Iowa. The comments and suggestions provided by Chris Porter of the National Severe Storms Laboratory, and three anonymous reviewers, were greatly appreciated. The manuscript was carefully proofread by Mary D. Steiner of Princeton University. This project was supported by the National Aeronautics and Space Administration (NASA) Grant NAG5-7744 and by the National Weather Service and the NEXRAD Operations Support Facility under Cooperative Agreement NA87WH0518.

REFERENCES

  • Amitai, E., 1999: Relationships between radar properties at high elevations and surface rain rate: Potential use for spaceborne rainfall measurements. J. Appl. Meteor., 38 , 321333.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anderson, J. R., 1981: Evaluating ground clutter filters for weather radars. Preprints, 20th Conf. on Radar Meteorology, Boston, MA, Amer. Meteor. Soc., 314–318.

    • Search Google Scholar
    • Export Citation
  • Andersson, T., 1993: Clutter suppression with Doppler radar—A case study. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 241–243.

    • Search Google Scholar
    • Export Citation
  • Andrieu, H., and Creutin J. D. , 1995: Identification of vertical profiles of radar reflectivity for hydrological applications using an inverse method. Part I: Formulation. J. Appl. Meteor., 34 , 225239.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andrieu, H., Delrieu G. , and Creutin J. D. , 1995: Identification of vertical profiles of radar reflectivity for hydrological applications using an inverse method. Part II: Sensitivity analysis and case study. J. Appl. Meteor., 34 , 240259.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Aoyagi, J., 1978: Ground clutter rejection by MTI weather radar. Preprints, 18th Conf. on Radar Meteorology, Atlanta, GA, Amer. Meteor. Soc., 358–363.

    • Search Google Scholar
    • Export Citation
  • Archibald, E., 2000: Enhanced clutter processing for the UK weather radar network. Phys. Chem. Earth, 25B , 823828.

  • Arritt, R. W., Rink T. D. , Segal M. , Todey D. P. , Clark C. A. , Mitchell M. J. , and Labas K. M. , 1997: The Great Plains low-level jet during the warm season of 1993. Mon. Wea. Rev., 125 , 21762192.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Atkinson, N. C., and James P. K. , 1991: Quality control of data from overlapping weather radars. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 206–209.

    • Search Google Scholar
    • Export Citation
  • Babin, S. M., 1996: Surface duct height distributions for Wallops Island, Virginia, 1985–1994. J. Appl. Meteor., 35 , 8693.

  • Babin, S. M., and Rowland J. R. , 1992: Observation of a strong surface radar duct using helicopter acquired fine-scale radio refractivity measurements. Geophys. Res. Lett., 19 , 917920.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Baer, V. E., 1991: The transition from the present radar dissemination system to the NEXRAD Information Dissemination Service (NIDS). Bull. Amer. Meteor. Soc., 72 , 2933.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bech, J., Sairouni A. , Codina B. , Lorente J. , and Bebbington D. , 2000: Weather radar anaprop conditions at a Mediterranean coastal site. Phys. Chem. Earth, 25B , 829832.

    • Search Google Scholar
    • Export Citation
  • Blackman, M., and Illingworth A. J. , 1993: Differential phase measurement of precipitation. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 745–747.

    • Search Google Scholar
    • Export Citation
  • Brooks, I. M., Goroch A. K. , and Rogers D. P. , 1999: Observations of strong surface radar ducts over the Persian Gulf. J. Appl. Meteor., 38 , 12931310.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Burk, S. D., and Thompson W. T. , 1997: Mesoscale modeling of summertime refractive conditions in the Southern California bight. J. Appl. Meteor., 36 , 2231.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collier, C. G., 2000: COST-75 final report—A summary. Phys. Chem. Earth, 25B , 817822.

  • Collier, C. G., Lovejoy S. , and Austin G. L. , 1980: Analysis of bright bands from 3-D radar data. Preprints, 19th Conf. on Radar Meteorology, Miami Beach, FL, Amer. Meteor. Soc., 44–47.

    • Search Google Scholar
    • Export Citation
  • Cook, J., 1991: A sensitivity study of weather data inaccuracies on evaporation duct height algorithms. Radio Sci., 26 , 731746.

  • Cornelius, R., 1994: Detecting ground clutter contamination of weather radar data with neural networks. Preprints, World Congress on Neural Networks, San Diego, CA, International Neural Network Society.

    • Search Google Scholar
    • Export Citation
  • Coveri, C., Caldini P. , Vezzani G. F. , and Lee R. , 1993: Statistical ground clutter filter. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 244–246.

    • Search Google Scholar
    • Export Citation
  • Crum, T. D., Saffle R. E. , and Wilson J. W. , 1998: An update on the NEXRAD program and future WSR-88D support to operations. Wea. Forecasting, 13 , 253262.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Delrieu, G., Creutin J. D. , and Andrieu H. , 1995: Simulation of radar mountain returns using a digitized terrain model. J. Atmos. Oceanic Technol., 12 , 10381049.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eichinger, W. E., Cooper D. I. , Parlange M. , and Katul G. , 1993: The application of a scanning, water Raman-lidar as a probe of the atmospheric boundary layer. IEEE Trans. Geosci. Remote Sens., 31 , 7079.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eichinger, W. E., Cooper D. I. , Forman P. R. , Griegos J. , Osborn M. A. , Richter D. , Tellier L. L. , and Thornton R. , 1999: The development of a scanning Raman water vapor lidar for boundary layer and tropospheric observations. J. Atmos. Oceanic Technol., 16 , 17531766.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fabry, F., Frush C. , Zawadzki I. , and Kilambi A. , 1997: On the extraction of near-surface index of refraction using radar phase measurements from ground targets. J. Atmos. Oceanic Technol., 14 , 978987.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fiore, J. V., Farnsworth R. K. , and Huffman G. , 1986: Quality control of radar-rainfall data with VISSR satellite data. Preprints, Joint Sessions 23d Conf. on Radar Meteorology and Conf. on Cloud Physics, Vol. 3, Snowmass, CO, Amer. Meteor. Soc., JP15–JP18.

    • Search Google Scholar
    • Export Citation
  • Fulton, R. A., Breidenbach J. P. , Seo D. J. , Miller D. A. , and O'Bannon T. , 1998: The WSR-88D rainfall algorithm. Wea. Forecasting, 13 , 377395.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Geotis, S. G., and Silver W. M. , 1976: An evaluation of techniques for automatic ground-echo rejection. Preprints, 17th Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 448–452.

    • Search Google Scholar
    • Export Citation
  • Giorgi, F., Mearns L. O. , Shields C. , and Mayer L. , 1996: A regional model study of the importance of local versus remote controls of the 1988 drought and the 1993 flood over the central United States. J. Climate, 9 , 11501162.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giuli, D., Gherardelli M. , Freni A. , Seliga T. A. , and Aydin K. , 1991: Rainfall and clutter discrimination by means of dual-linear polarization radar measurements. J. Atmos. Oceanic Technol., 8 , 777789.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gossard, E. E., 1977: Refractive index variance and its height distribution in different air masses. Radio Sci., 12 , 89105.

  • Gossard, E. E., Strauch R. G. , Stankov B. B. , and Wolfe D. E. , 1995: Measurement of property gradients and turbulence aloft with ground-based Doppler radars. NOAA Tech. Memo. ERL 453-ETL 67, Environmental Technology Laboratory, Boulder, CO, 31 pp.

    • Search Google Scholar
    • Export Citation
  • Grecu, M., and Krajewski W. F. , 1999: Detection of anomalous propagation echoes in weather radar data using neural networks. IEEE Trans. Geosci. Remote Sens., 37 , 287296.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grecu, M., and Krajewski W. F. , . 2000: An efficient methodology for detection of anomalous propagation echoes in radar reflectivity data using neural networks. J. Atmos. Oceanic Technol., 17 , 121129.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gumley, L. E., and King M. D. , 1995: Remote sensing of flooding in the U.S. upper Midwest during the summer of 1993. Bull. Amer. Meteor. Soc., 76 , 933943.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Haddad, B., Adane A. , Mesnard F. , and Sauvageot H. , 2000: Modeling anomalous radar propagation using first-order two-state Markov chains. Atmos. Res., 52 , 283292.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, M. P. M., Goddard J. W. F. , and Cherry S. M. , 1984: Identification of hydrometeors and other targets by dual-polarization radar. Radio Sci., 19 , 132140.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hamidi, S., and Zrnic D. S. , 1981: Considerations for the design of ground clutter cancelers for weather radars. Preprints, 20th Conf. on Radar Meteorology, Boston, MA, Amer. Meteor. Soc., 319–326.

    • Search Google Scholar
    • Export Citation
  • Harrison, D. L., Driscoll S. J. , and Kitchen M. , 2000: Improving precipitation estimates from weather radar using quality control and correction techniques. Meteor. Appl., 7 , 135144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heiss, W. H., McGrew D. L. , and Sirmans D. , 1990: NEXRAD: Next Generation Weather Radar (WSR-88D). Microwave J., 33 , 7998.

  • Hogg, W. D., 1978: Quality control and analysis of an archive of digital radar data. Preprints, 18th Conf. on Radar Meteorology, Atlanta, GA, Amer. Meteor. Soc., 150–154.

    • Search Google Scholar
    • Export Citation
  • Joe, P., 1991: Classification of radar imagery using texture analyses. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 107–110.

    • Search Google Scholar
    • Export Citation
  • Johnson, G. N., Smith Jr. P. L. , Nathanson F. E. , and Brooks L. W. , 1975: An analysis of techniques for dealing with anomalous propagation. Preprints, 16th Radar Meteorology Conf., Houston, TX, Amer. Meteor. Soc., 374–377.

    • Search Google Scholar
    • Export Citation
  • Joss, J., and Wessels H. , 1990: Ground clutter suppression for weather radar data. COST Tech. Rep. 73/WD/130, 6 pp.

  • Joss, J., and Pittini A. , 1991: Real-time estimation of the vertical profile of radar reflectivity to improve the measurement of precipitation in an Alpine region. Meteor. Atmos. Phys., 47 , 6172.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Joss, J., and Lee R. W. , 1993: Scan strategy, clutter suppression, calibration and vertical profile corrections. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 390–392.

    • Search Google Scholar
    • Export Citation
  • Joss, J., and Lee R. W. , . 1995: The application of radar–gauge comparisons to operational precipitation profile corrections. J. Appl. Meteor., 34 , 26122630.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keeler, R. J., and Passarelli R. E. , 1990: Signal processing for atmospheric radars. Radar in Meteorology, D. Atlas, Ed., Amer. Meteor. Soc., 199–230.

    • Search Google Scholar
    • Export Citation
  • Kessinger, C., Ellis S. , and VanAndel J. , 2001: NEXRAD data quality: The AP clutter mitigation scheme. Preprints, 30th Int. Conf. on Radar Meteorology, Munich, Germany, Amer. Meteor. Soc., 707–709.

    • Search Google Scholar
    • Export Citation
  • Kitchen, M., Brown R. , and Davies A. G. , 1994: Real-time correction of weather radar data for the effects of bright band, range and orographic growth in widespread precipitation. Quart. J. Roy. Meteor. Soc., 120 , 12311254.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Klingle-Wilson, D., Mann E. , and Boldi R. , 1995: An algorithm to remove anomalous propagation clutter returns from ASR-9 weather channel data using pencil beam radar data. Preprints, Sixth Conf. on Aviation Weather Systems, Dallas, TX, Amer. Meteor. Soc., 366–371.

    • Search Google Scholar
    • Export Citation
  • Kunkel, K. E., Changnon S. A. , and Angel J. R. , 1994: Climatic aspects of the 1993 upper Mississippi River basin flood. Bull. Amer. Meteor. Soc., 75 , 811822.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lee, R., Bruna G. Della , and Joss J. , 1995: Intensity of ground clutter and of echoes of anomalous propagation and its elimination. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 651–652.

    • Search Google Scholar
    • Export Citation
  • Mann, D. R., Evans J. E. , and Merritt M. W. , 1986: Clutter suppression for low altitude wind shear detection by Doppler weather radars. Preprints, 23d Conf. on Radar Meteorology, Vol. 1, Snowmass, CO, Amer. Meteor. Soc., R9–R13.

    • Search Google Scholar
    • Export Citation
  • Michelson, D. B., and Andersson T. , 1995: Identification and suppression of anomalous propagation echoes in two-dimensional radar images. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 656–658.

    • Search Google Scholar
    • Export Citation
  • Moores, W., and Harrold T. W. , 1975: Estimating the distribution and intensity of ground clutter. Preprints, 16th Radar Meteorology Conf., Houston, TX, Amer. Meteor. Soc., 370–373.

    • Search Google Scholar
    • Export Citation
  • Moszkowicz, S., Ciach G. J. , and Krajewski W. F. , 1994: Statistical detection of anomalous propagation in radar reflectivity patterns. J. Atmos. Oceanic Technol., 11 , 10261034.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mueller, E. A., and Sims A. L. , 1975: Statistics of high radar reflectivity gradients. Preprints, 16th Radar Meteorology Conf., Houston, TX, Amer. Meteor. Soc., 401–403.

    • Search Google Scholar
    • Export Citation
  • Pamment, J. A., and Conway B. J. , 1998: Objective identification of echoes due to anomalous propagation in weather radar data. J. Atmos. Oceanic Technol., 15 , 98113.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pappert, R. A., and Goodhart C. L. , 1977: Case studies of beyond-the-horizon propagation in tropospheric ducting environments. Radio Sci., 12 , 7587.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Passarelli, R. E. Jr,, 1981: Autocorrelation techniques for ground clutter rejection. Preprints, 20th Conf. on Radar Meteorology, Boston, MA, Amer. Meteor. Soc., 308–313.

    • Search Google Scholar
    • Export Citation
  • Passarelli, R. E. Jr,, Romanik P. , Geotis S. G. , and Siggia A. D. , 1981: Ground clutter rejection in the frequency domain. Preprints, 20th Conf. on Radar Meteorology, Boston, MA, Amer. Meteor. Soc., 295–300.

    • Search Google Scholar
    • Export Citation
  • Pratte, J. F., and Keeler R. J. , 1986: Sensitivities of operational weather radars. Preprints, Joint Sessions 23d Conf. on Radar Meteorology and Conference on Cloud Physics, Vol. 3, Snowmass, CO, Amer. Meteor. Soc., JP333–JP336.

    • Search Google Scholar
    • Export Citation
  • Pratte, J. F., Gagnon R. , and Cornelius R. , 1993: Ground clutter characteristics and residue mapping. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 50–52.

    • Search Google Scholar
    • Export Citation
  • Pratte, J. F., Keeler R. J. , Gagnon R. , and Sirmans D. , 1995: Clutter processing during anomalous propagation conditions. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 139–141.

    • Search Google Scholar
    • Export Citation
  • Puzzo, D. C., Troxel S. W. , Meister M. A. , Weber M. E. , and Pieronek J. V. , 1989: Appendix I: Methods for filtering clutter as a result of anomalous propagation. ASR-9 Weather Channel Test Report. Project Rep. ATC-165, Lincoln Laboratory, MIT, Lexington, MA, 155–170.

    • Search Google Scholar
    • Export Citation
  • Reid, J. D., 1970: Decorrelation time in weather radar signals. Preprints, 14th Radar Meteorology Conf., Tucson, AZ, Amer. Meteor. Soc., 419–420.

    • Search Google Scholar
    • Export Citation
  • Riley, G. F. Jr,, and Austin P. M. , 1976: Some statistics of gradients of precipitation intensity. Preprints, 17th Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 492–497.

    • Search Google Scholar
    • Export Citation
  • Rinehart, R. E., 1991: Antenna rotation-produced moving sidelobe echoes from ground clutter with a Doppler radar. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 339–342.

    • Search Google Scholar
    • Export Citation
  • Robinson, M., Steiner M. , Wolff D. B. , Kessinger C. , and Fulton R. A. , 2001: Radar data quality control: Evaluation of several algorithms based on accumulating rainfall statistics. Preprints, 30th Int. Conf. on Radar Meteorology, Munich, Germany, Amer. Meteor. Soc., 274–276.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., Amitai E. , and Wolff D. B. , 1995: Classification of rain regimes by the three-dimensional properties of reflectivity fields. J. Appl. Meteor., 34 , 198211.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryzhkov, A. V., and Zrnic D. S. , 1998: Polarimetric rainfall estimation in the presence of anomalous propagation. J. Atmos. Oceanic Technol., 15 , 13201330.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sánchez-Diezma, R., Sempere-Torres D. , Delrieu G. , and Zawadzki I. , 2001: An improved methodology for ground clutter substitution based on a pre-classification of precipitation types. Preprints, 30th Int. Conf. on Radar Meteorology, Munich, Germany, Amer. Meteor. Soc., 271–273.

    • Search Google Scholar
    • Export Citation
  • Schmid, W., Högl D. , Vckovski A. , and Joss J. , 1991: Test of clutter suppression techniques in the Swiss Alps. Preprints, 25th Int. Conf. on Radar Meteorology, Paris, France, Amer. Meteor. Soc., 875–878.

    • Search Google Scholar
    • Export Citation
  • Sekhon, R. S., and Atlas D. , 1972: Doppler radar structure of the melting zone and ground clutter effects. Preprints, 15th Radar Meteorology Conf., Champaign–Urbana, IL, Amer. Meteor. Soc., 222–227.

    • Search Google Scholar
    • Export Citation
  • Seo, D. J., Breidenbach J. , Fulton R. , Miller D. , and O'Bannon T. , 2000: Real-time adjustment of range-dependent biases in WSR-88D rainfall estimates due to nonuniform vertical profile of reflectivity. J. Hydrometeor., 1 , 222240.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sirmans, D., and Dooley J. T. , 1980: Ground clutter statistics of a 10 cm ground based radar. Preprints, 19th Conf. on Radar Meteorology, Miami Beach, FL, Amer. Meteor. Soc., 184–191.

    • Search Google Scholar
    • Export Citation
  • Skolnik, M. I., 1980: Introduction to Radar Systems. 2d ed. McGraw-Hill International Editions, 581 pp.

  • Smith, J. A., Seo D. J. , Baeck M. L. , and Hudlow M. D. , 1996: An intercomparison study of NEXRAD precipitation estimates. Water Resour. Res., 32 , 20352045.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, P. L. Jr,, 1972: Siting considerations for weather radars. Preprints, 15th Radar Meteorology Conf., Champaign–Urbana, IL, Amer. Meteor. Soc., 99–100.

    • Search Google Scholar
    • Export Citation
  • Smith, P. L. Jr,, . 1990: Precipitation measurement and hydrology: Panel report. Radar in Meteorology, D. Atlas, Ed., Amer. Meteor. Soc., 607–618.

    • Search Google Scholar
    • Export Citation
  • Smith, P. L. Jr,, . 1993: An update on weather radar system sensitivity. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 384–386.

    • Search Google Scholar
    • Export Citation
  • Smith, P. L. Jr,, . 1998: On the minimum useful elevation angle for weather surveillance radar scans. J. Atmos. Oceanic Technol., 15 , 841843.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steiner, M., Smith J. A. , Burges S. J. , Alonso C. V. , and Darden R. W. , 1999a: Effect of bias adjustment and rain gauge data quality control on radar rainfall estimation. Water Resour. Res., 35 , 24872503.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steiner, M., Smith J. A. , Kessinger C. , and Ferrier B. S. , 1999b: Evaluation of algorithm parameters for radar data quality control. Preprints, 29th Int. Conf. on Radar Meteorology, Montreal, QC, Canada, Amer. Meteor. Soc., 582–585.

    • Search Google Scholar
    • Export Citation
  • Tatehira, R., and Shimizu T. , 1978: Intensity measurement of precipitation echo superposed on ground clutter—A new automatic technique for ground clutter rejection. Preprints, 18th Conf. on Radar Meteorology, Atlanta, GA, Amer. Meteor. Soc., 364–369.

    • Search Google Scholar
    • Export Citation
  • Tatehira, R., and Shimizu T. , . 1980: Improvements in performance of ground clutter rejection. Preprints, 19th Conf. on Radar Meteorology, Miami Beach, FL, Amer. Meteor. Soc., 176–179.

    • Search Google Scholar
    • Export Citation
  • Torres, S. M., and Zrnic D. S. , 1999: Ground clutter canceling with a regression filter. J. Atmos. Oceanic Technol., 16 , 13641372.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trapp, R. J., and Doswell III C. A. , 2000: Radar data objective analysis. J. Atmos. Oceanic Technol., 17 , 105120.

  • VanAndel, J., 2001: Radar echo classifier algorithm development using Python. Preprints, 30th Int. Conf. on Radar Meteorology, Munich, Germany, Amer. Meteor. Soc., 704–706.

    • Search Google Scholar
    • Export Citation
  • Vignal, B., Andrieu H. , and Creutin J. D. , 1999: Identification of vertical profiles of reflectivity from volume scan radar data. J. Appl. Meteor., 38 , 12141228.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vignal, B., Galli G. , Joss J. , and Germann U. , 2000: Three methods to determine profiles of reflectivity from volumetric radar data to correct precipitation estimates. J. Appl. Meteor., 39 , 17151726.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weber, M. E., Stone M. L. , and Cullen J. A. , 1993: Anomalous propagation associated with thunderstorm outflows. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 238–240.

    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., and Schreiber W. E. , 1986: Initiation of convective storms at radar-observed boundary-layer convergence lines. Mon. Wea. Rev., 114 , 25162536.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., Crook N. A. , Mueller C. K. , Sun J. Z. , and Dixon M. , 1998: Nowcasting thunderstorms: A status report. Bull. Amer. Meteor. Soc., 79 , 20792099.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zrnic, D. S., and Ryzhkov R. V. , 1996: Advantages of rain measurements using specific differential phase. J. Atmos. Oceanic Technol., 13 , 454464.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Average refractivity gradient within 500 m AGL across the continental United States, based on a 16-yr climatology of operational soundings taken at 1200 UTC and shown by season (winter = DJF, spring = MAM, summer = JJA, fall = SON). The fraction of a box shaded either gray (superrefraction) or black (trapping) indicates the percentage of soundings of that particular station exhibiting the respective atmospheric conditions

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 2.
Fig. 2.

Same as Fig. 1, but for soundings taken at 0000 UTC

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 3.
Fig. 3.

Decision tree of the new radar data quality control algorithm

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 4.
Fig. 4.

Radar data collected by the Amarillo, Texas, WSR-88D (KAMA) on 25 May 1994 at 0034 UTC. Shown are the (a) unedited radial Doppler velocity and (b) reflectivity of the base scan, together with (c) the reflectivity of the second tilt. In addition, three different stages of the edited base scan reflectivity field are shown: (d) before filling the gaps in rainfall echo areas caused by the removal of pixels, and after filling those gaps by either using reflectivity information from the (e) second or (f) third tilt above

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 5.
Fig. 5.

Same as Fig. 4, but for data collected by the Amarillo, Texas, WSR-88D (KAMA) on 25 May 1994 at 0231 UTC

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 6.
Fig. 6.

Same as Fig. 4, but for data collected by the Saint Louis, Missouri, WSR-88D (KLSX) on 7 Jul 1993 at 0404 UTC

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 7.
Fig. 7.

Same as Fig. 4, but for data collected by the Houston, Texas, WSR-88D (KHGX) on 9 May 1998 at 0902 UTC

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Fig. 8.
Fig. 8.

Same as Fig. 4, but for data collected by the Melbourne, Florida, WSR-88D (KMLB) on 7 Jul 1998 at 2201 UTC

Citation: Journal of Atmospheric and Oceanic Technology 19, 5; 10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2

Table 1. 

Comparison of radar echo area, and unconditional and conditional area-average rain rate based on the unedited and edited (various stages) radar reflectivity fields

Table 1. 
Save