Radar, Part 2: Interpreting Radar Images

This is a sample lesson page from the Certificate of Achievement in Weather Forecasting offered by the Penn State Department of Meteorology. Any questions about this program can be directed to: Steve Seman


At the completion of this section, you should be able to list and describe the three precipitation factors that affect radar reflectivity. You should be able to explain why hail causes very large reflectivity values while snow tends to be under measured. You should also be able to explain the difference between "base reflectivity" and "composite reflectivity."


Now that you know how a radar works, we need to discuss how to properly interpret the returned radar signal. As with any remote sensing tool, we have to understand what factors influence the amount of radiation that is received by the instrument. As you recall, radar works via transmitted and returned microwave energy. The radar transmits a burst of microwaves and when this energy strikes an object, the energy is scattered in all directions. Some of that scattered energy returns to the radar and this returned energy is then converted to reflectivity (in dBZ). Ultimately, the intensity of the return echo (and therefore, reflectivity) depends on three main factors inside a volume of air probed by the radar "beam":

  • the size of the targets
  • the number of targets
  • the composition of the targets (raindrops, snowflakes, ice pellets, etc.)

Allow me to elaborate a bit on each of these factors impacting radar reflectivity. For starters, the size of the precipitation targets always matters. The larger the targets (raindrops, snowflakes, etc.,) the higher the reflectivity. By way of example, consider that raindrops, by virtue of their larger size, have a much higher radar reflectivity than drizzle drops (the tiny drops of water that appear to be more of a mist than rain). Secondly, the power returning from a sample volume of air with a large number of raindrops is greater than the power returning from an equal sample volume containing fewer raindrops (assuming, of course, that both sample volumes have the same sized drops). The saying that "there's power in numbers" certainly applies to radar imagery!

To see how the size and number of targets impact reflectivity, consider this example. Many thunderstorms often show high reflectivity on radar images, with passionate colors like deep reds marking areas within the storm with a large number of sizable raindrops. A large number of sizable raindrops falling from a cumulonimbus cloud also leads to high rainfall rates at the ground. Thus, high radar reflectivities are usually associated with heavy rain.

An image of radar reflectivity from 2255Z on June 1, 2012.
An image of radar reflectivity from 2255Z on June 1, 2012. The line of high reflectivity values denotes large numbers of large rain drops (characteristic of thunderstorms). Indeed, rainfall rates when this "squall line" passed through exceeded 3.5 inches per hour!
Credit: NCAR

The radar image above from 2255Z on June 1, 2012 shows a line of strong thunderstorms (called a "squall line") just to the west of State College, Pennsylvania (UNV on the map). Although the storm moved through the region very quickly, rainfall rates at the Penn State Weather Center exceeded 0.6 inches in a 10 minute period. This converts to a rainfall rate of 3.6 inches (91.4 mm) per hour! That said, let me caution you that inferring specific rainfall rates from radar images can be tricky business. A given reflectivity can translate to different rainfall rates, depending on, for example, whether there are a lot of small drops versus fewer large drops.

The presence of large hail in thunderstorms can really complicate the issue of inferring rainfall rates from radar reflectivity even more. Typically, radar reflectivity from a thunderstorm is greatest in the middle levels of the storm because large hailstones have started to melt as they fall earthward into air with temperatures greater than 0 degrees Celsius (the melting point of ice). Covered with a film of melt-water, these large hailstones look like giant raindrops to the radar and can have reflectivity values higher than 70 dBZ. So, when large, "wet hailstones" are present in thunderstorms, rainfall rates inferred from the very large reflectivity are typically overestimated. The bottom line is that higher reflectivity usually corresponds to higher rainfall rates, but the connection is not always neat and tidy.

Okay, lets move on to the final controller of radar reflectivity -- composition. The intensity of the return signal from raindrops is approximately five times greater than the return from snowflakes that have comparable sizes. Snowflakes have inherently low reflectivity compared to raindrops, so it's easy to underestimate the area coverage and intensity of snowstorms if you're unaware of this fact. It might be snowing quite heavily, yet radar reflectivity from the heavy snow might be less than from a nearby area of rain (even if the rainfall isn't as heavy) because the return signal from raindrops is more intense.

There's another way that moderate to heavy snow falling within the range of the radar can be camouflaged. Indeed, precipitating stratiform clouds are often shallow (not very tall), which means that the radar beam will sometimes overshoot snow-bearing clouds located relatively far away from the radar site (see radar beam on the left, below). To see what I mean, check out the meteogram at Islip, New York (on Long Island), from 14Z on January 26, 2011, to 14Z on January 27, 2011. Note the report of heavy snow at 07Z on the 27th. Now take a look at the 07Z reflectivity from the radar at Boston, and focus your attention on the very weak reflectivity at Islip. Clearly, Islip is almost out of the range (the white circle) of the Boston radar, so the radar beam really overshot the relatively shallow nimbostratus clouds producing heavy snow at Islip at the time. Fortunately, the radar at Upton, New York is located closer to Islip, and as you can see its 07Z image of reflectivity, it gives a much more realistic look for Islip. The moral of this story is that you need to be careful interpreting radar images in winter where snow might be falling.

A graphic to show the radar reflectivity of nimbostratus clouds. A radar beam shines across the image and extends above some clouds that are dropping snow.
The radar reflectivity of nimbostratus clouds (on the left) relatively far away from the radar can be too low because the radar beam probes its upper layers, where precipitation rates are relatively light (or the radar beam misses the cloud altogether).
Credit: David Babb

To further complicate interpreting radar images during winter, I point out that partially melted snowflakes present a completely different problem to weather forecasters during winter. When snowflakes melt, they melt at their edges first. With water distributed along the edges of the "arms" of melting flakes, partially melted snowflakes appear like large raindrops to the radar. Thus, partially melted snowflakes have unexpectedly high reflectivity.

For pretty much the same reason, wet or melting ice pellets (sleet) also have a relatively high reflectivity. During winter, radar images sometimes show a blob of high reflectivity embedded in an area of generally light rain. Often, this renegade echo of high reflectivity is "wet sleet". It's a good idea to look at surface observations to verify whether the relatively intense echo is indeed sleet or an area of moderate to heavy rain. For example, check out this thin band of sleet (marked by high reflectivity) over southeast New York and northeast Pennsylvania, which separated rain to the south and snow to the north. Those who did not bother to look at surface observations may have mistaken this area of "wet sleet" or partially melted snowflakes for intense rainfall.

Base Versus Composite Reflectivity

For a powerful thunderstorm that erupts fairly close to the radar, a scan at 0.5 degrees would likely intercept the storm below the level where the most intense reflectivity occurs. Such a single, shallow scan falls way short of painting a proper picture of the storm's potential. As a routine counter-measure, the radar tilts upward at increasingly large angles of elevation, scanning the entire thunderstorm like a diagnostic, full-body MRI.

The radar can tilt upward to angles of elevation as large as 19.5 degrees, as indicated in the figure below, which shows the elevation scans in a common "general surveillance" radar mode. But, the series of elevation scans shown below isn't the only option that National Weather Service NEXRAD units have; they are programmed with multiple scanning strategies to give forecasters the most useful data depending on the weather situation. A complete scan like the one shown below takes about 6 minutes, which means that under normal circumstances, forecasters must wait about 6 minutes to get a look at the newest radar scan at each elevation. But, during severe weather, forecasters desire more frequent low-elevation scans to better see what's happening in the lower parts of thunderstorms. So, the radar can be switched into "SAILS" mode, which causes the radar to interrupt its scanning progression to give more low-level scans, providing forecasters with more frequent updates on the lowest elevation scan.

The elevation scans of the WSR-88D (NEXRAD) radar. More explanation in text.
The elevation scans of the WSR-88D (NEXRAD) in general surveillance mode.
Credit: NOAA's Radar Operations Center

On the image above showing how the radar can tilt upward at increasingly large angles, the numbers at the top represent the standard angles included as part of the general surveillance scan. Also note the colorful "beams," which represent the approximate width and length of the radar scan as a function of distance from the radar site. Again, note how wide the "beam" becomes at great distances from the radar.

Meteorologists describe the radar reflectivity derived from a single scan as base reflectivity, and the most common base reflectivity corresponds to the scanning angle of 0.5 degrees. The National Weather Service also provides images of composite reflectivity, which represents the highest reflectivity gleaned from all of the individual scan angles.

The idea that reflectivity at one scan angle can be higher than at another angle might seem a little fuzzy to you, so I'll elaborate. Consider, for example, the case of a severe thunderstorm. The storm's updraft, which is a fast, rising current of moist air that sustains the thunderstorm, is usually strong enough (25 meters per second or faster) to suspend a large amount of rain (and hail) aloft. Meteorologists call the suspension of precipitation high in a thunderstorm precipitation loading. At this stage of the storm, the reflectivity high in the cumulonimbus cloud is much greater than the reflectivity lower in the cloud. So, a radar image created from composite reflectivity will likely display the higher dBZ level (more intense colors) than a radar image of base reflectivity. Eventually, of course, the rain intensity at lower altitudes (and the surface) will increase as rain and hail fall from the cloud (this will occur once the updraft can no longer support the weight of suspended water and ice).

For example, check out the image below. This graphic shows radar reflectivity plots of a garden-variety thunderstorm at four different scan angles. For the record, the storm erupted over southeast Pennsylvania and the radar data came from the NEXRAD near State College. The upper-left panel shows the radar reflectivity at a scan angle of 0.5 degrees, the upper-right displays the radar reflectivity at a scan angle of 1.5 degrees, while the lower-left and lower-right panels correspond to scan angles of 2.4 degrees and 3.4 degrees, respectively.

The radar reflectivity of a garden-variety thunderstorm at four different scan angles. More explanation in text.
The radar reflectivity of a garden-variety thunderstorm at four different scan angles. For the record, the storm erupted over southeast Pennsylvania and the radar data came from the NEXRAD near State College (KCCX). The upper-left panel shows the radar reflectivity at a scan angle of 0.5 degrees, the upper-right displays the radar reflectivity at a scan angle of 1.5 degrees, while the lower-left and lower-right panels correspond to scan angles of 2.4 degrees and 3.4 degrees respectively.
Credit: Mark Thornton

First, note that the core radar reflectivity on the upper-right panel (scan angle of 1.5 degrees) was higher than the core base reflectivity at 0.5 degrees (see upper-left panel). Comparing the two images, we conclude that the heaviest precipitation was higher up in the thunderstorm at this time. Note that the radar reflectivity markedly decreased at a scan angle of 2.4 degrees (lower-left panel). When the scan angle was set to 3.4 degrees (lower-right panel), the reflectivity all but vanished, indicating that there weren't many precipitation particles near the top of the storm.

Here's one last example of how composite reflectivity can be higher than base reflectivity due to precipitation loading in the updrafts of intense thunderstorms. On August 19, 2008, Tropical Storm Fay was moving very, very slowly over the Florida peninsula. The 1938Z base reflectivity (on the left, below) shows fairly high dBZ values (orange) in the immediate vicinity of the radar site at Melbourne, FL (KMLB). At the time, heavy thunderstorms were pounding the east-central coast of Florida. Now shift your attention to the image of composite reflectivity on the right (below). Notice the reddish colors near Melbourne, which indicate higher dBZ values (compared to the corresponding radar echoes on the image of base reflectivity).

The base and composite reflectivities of Tropical Storm Fay.
(Left) The base reflectivity of Tropical Storm Fay from the radar at Melbourne, FL (KMLB), at 1938Z on August 19, 2008. (Right) The composite reflectivity at the same time. For the record, the range associated with composite reflectivity is 285 miles, about twice the range corresponding to base reflectivity.
Credit: National Weather Service

Composite reflectivity may not be representative of current precipitation rates at the ground, but it can show the potential if the precipitation causing the highest reflectivity (often well up into the cloud) can fall to the surface.You might think that this discussion is probably too much "inside baseball," but composite reflectivity is the mode of choice on regional or national mosaics that you frequently see on the Web and on television. So, the bottom line is to make sure that you know which type of radar product you are looking at before performing any kind of analysis.

Now you know the basics of interpreting radar imagery, and we're just about ready to wrap-up our lesson. Before you finish up, however, you may be interested in the Explore Further section below, where you can find out more about some common radar products (precipitation-type images and satellite-radar composites) that you'll commonly encounter on television and online.

Explore Further...

Commonly, regional or national radar mosaics visually distinguish areas of rain from snow and mixed precipitation (any combination of snow, sleet, freezing rain, and/or rain) using different color keys. Note that rain, mixed precipitation, and snow each has its own color key in the national radar mosaic below. While the exact methods for creating such images vary, they all start with radar reflectivity, and often incorporate surface temperature and other observations to give a "best guess" of precipitation type. More specifically, while the radar data is used to tell where precipitation is falling, surface analyses of temperature and moisture variables are plugged into a mathematical and imperfect formula in order to make an "educated guess" at the type of precipitation reaching the ground. 

An example of the kind of radar imagery used in TV weathercasts. A map of the US, with precipitation shown in color. Precipitation is color-coded, so the average viewer can quickly see what type is expected where.
The mosaic of radar reflectivity at 12Z on December 20, 2011. Surface observations were artificially combined to "loosely infer" precipitation type (where echoes appear on the map). Echoes are then colorized to roughly distinguish between snow, rain and mixed precipitation.
Credit: WSI Corporation

For example, examine the radar reflectivity map above. Now look at the pink areas representing mixed precipitation in eastern Ohio into western Pennsylvania. Compare these pink areas with the corresponding 12Z observations of precipitation and obstructions to visibility. To help you get your bearings, I circled three observations of rain mixed with snow. So, yes, there was some mixed precipitation in eastern Ohio and western Pennsylvania, but on close inspection, it appears that rain was actually falling in some of the pink areas on radar near West Virginia (oops!). Radars now have some additional capabilities (beyond the scope of this course) to help discern precipitation type, but always keep in mind that such "precipitation-type" radar images aren't perfect (they don't always show the correct precipitation type).

Another common product is a satellite-radar composite, or "sat-rad image" (see image below). For the record, sat-rad images are mongrel superimpositions of radar imagery onto satellite images. Before using or interpreting this type of image, make sure that you're aware of a few key things. First, the satellite and radar data come from two completely different sources, although the "look" of the image doesn't necessarily make that obvious. As you know, WSR-88D radars are located on the ground (not aboard geostationary satellites), which has some major implications for data coverage.

An example of a sat-rad image, which is radar imagery superimposed onto an infrared satellite image. The US is shown here.
The 12Z infrared satellite image and the corresponding radar mosaic on May 10, 2012. Isobars (blue contours) were added for purposes of analysis.
Credit: NOAA and the Penn State e-Wall

Recall the range of the national array of WSR-88D radars? It does not extend very far out into the oceans nor very far north into Canada nor very far south into Mexico. Thus, to an uneducated user, sat-rad images can give the impression that some clouds are not producing precipitation when they really are. For example, examine the 12Z sat-rad image on May 10, 2012 (above). Note the low-pressure system centered along the New England Coast. According to this radar mosaic, there weren't any radar echoes over the St. Lawrence River Valley in Quebec, Canada. But, if you study this 12Z mosaic of reflectivity on May 10, 2012 (from Environment Canada), you'll see that precipitation was indeed occurring over the St. Lawrence River Valley. Obviously, these echoes were beyond the range of U.S. radars. It works both ways, of course: The rain over the New England Seaboard was beyond the range of the Canadian radars (even though most Canadian radars are congregated fairly close to the U.S.-Canada border).

In the hands of educated apprentice forecasters, these images can enhance your analysis of the current weather pattern, but they can really be misleading to someone who's not fully aware of what these images really show. So, make sure to use them with care!