National Weather Service Weather Forecast Office

All instruments have limitations that may result in inaccurate or conflicting data. Knowing this, forecasters scrutinize the data they use to determine if it is valid and ascertain it's meaning. By using information from a variety of sources, forecasters may identify problematic data and gain a more accurate and complete picture of a flash flooding situation. In this section, we will discuss the limitations of the observation network, weather radar, weather satellites, and human forecasters. These limitations can work together to help a flash flood evade timely detection.

Rain Gauges Used by the NWS

(Courtesy of the Arkansas-Red Basin River Forecast Center)

• Sparse observations rarely catch local maximums of precipitation
• Most drainage basins don’t have rain gages
• Not timely enough--most observations only arrive hourly or once per day
• Gage intercomparison and interoperability problems
  • Different operators and maintainers
  • Different collection and measurement techniques
  • Different accuracy and precision
• Few flash flood warning devices; most don’t report to the NWS
• Low population density over much of NM results in low chance of flash flood detection and/or verification

• All linearly polarized Doppler or conventional radars occasionally produce inaccurate rainfall estimates due to:
  • Rain falling too far from the radar
  • Mountains blocking rain from view
  • Reflectivity and rain rate conversion errors
  • Below beam effects
  • Ground clutter and superrefraction
  • Hardware problems
• Net impact: rainfall overestimation by a factor of around two in convective events (Fresch, 2004). Most notable exceptions:
  • Radars tend to underestimate rainfall at the farthest ranges
  • Radars tend to underestimate rainfall blocked by mountains
  • Radars tend to drastically overestimate rainfall near large hail

National Weather Service forecasters primarily use the Weather Surveillance Radar 1988 Doppler (WSR-88D). Currently, the national network of WSR-88Ds uses linear polarization, and a 3 dB angular beam width of 0.96°. Read on for more detail on each of the above mentioned weather radar limitations with illustrations from the WSR-88D.

Radar Limitation 1: Rain too Far from Radar

Cause 1: Beam Broadening

Beam Width Versus Range

(NWS ROC, 2004)

• Area covered by beam increases as range increases
• Beam grows too large compared to size of rain shaft
• Algorithm assumes targets completely fill beam, because radar can’t tell otherwise
• Power returned from rain shaft averaged over entire beam width
  • The farther the rain shaft is from the radar, the wider the beam and the more the rain rate is underestimated
  • Also causes radar to overestimate areal coverage of rain echo at long ranges

Radar Limitation 1: Rain too Far from Radar

Cause 2: Overshooting

• Height of radar beam increases as range increases, because beams travel away from the radar at an angle
• Lowest beam (0.5°) can overshoot or only partially sample shallow stratiform rain or low-topped convection beyond about 100 miles from a radar
• At long ranges from the radar, overshooting can be the largest single contributor to rain underestimates (Story, 2002)

Radar Limitation 2: Rain Blocked from Radar View

• Mountains can shield mid/low-level rain shafts from view causing rainfall underestimation beyond mountains
• Since lower elevation angles intercept more terrain, coverage decreases below 10,000 ft
• WSR-88Ds try to correct some of this error :
  • Using data from range bins next to the blockage, if blockage < 2° in azimuth
  • Using data from higher angles, if blockage > 2° in azimuth; however, higher angles overshoot more low-level rain resulting in continued underestimation

In the left photo above, notice the much more circular and complete radar coverage in Texas, where the terrain is flatter. The reflectivity image (above right) depicts beam blockage by mountains east and north of Albuquerque, northeast of Santa Fe, and between Albuquerque and Gallup.

Radar Limitation 3: Rain Rate and Reflectivity Conversion Errors

• Radar equations for deriving rain rate from reflectivity sometimes overestimate or underestimate the rain rate
  • Reflectivity/rain-rate relationship depends on drop-size distribution
  • Drop-size distribution varies for each storm and during each storm
• Reflectivity is proportional to the 6th power of droplet diameter
  • Liquid water coats hail (less dense than water) causing it to look like a giant water drop to radars, resulting in huge rainfall overestimates
  • Wet melting snow and sleet (both less dense than water) stick together near the freezing level and look like giant water drops to radars, resulting in rainfall overestimates

Radar Limitation 4: Below Beam Effects

• Strong horizontal winds below the lowest beam can blow precipitation from one range bin to the next, causing rainfall overestimation in one location and underestimation in an adjacent location
• Evaporation of precipitation as it falls below cloud base can lead to rainfall overestimation
• Coalescence of rain drops below the lowest radar beam causes rainfall underestimation
  • Numerous small drops of a tropical airmass stick together after they fall below the radar beam
  • Since radar reflectivity is proportional to the 6th power of the droplet diameter, small droplets produce much smaller rain rate calculations than the actual amount of rain deposited by the larger drops on the ground
  • Occasionally a problem in NM during summer

Radar Limitation 5: Ground Clutter and Superrefraction

• Power returned from non-weather targets causes precipitation overestimation
  • Clutter suppression prevents most of this, but misses some clutter
  • Clutter suppression can also supress reflectivity from weather targets
  • WSR-88D algorithms attempt to identify anomalously high reflectivities and remove them before calculating rain rate
• Superrefraction (more downward bending of the radar beam than normal) causes precipitation overestimation when the beam bends far enough to hit the ground
  • Induced by temperature inversions with relatively strong vertical gradients of humidity
  • Usually results in rainfall tally when there is no rain

Radar Limitation 6: Hardware Problems

• External or internal system noise can cause reflectivity to depart from optimal calibration; results in rainfall overestimation or underestimation
• Wet radome causes precipitation underestimation
  • Liquid moisture on the dome around the antenna attenuates the signal where it’s transmitted
  • Less energy reaches the storms, so storms return less energy to the radar

Weather Radar Limitations Summary

1. Rain falling too far from radar

• Beam broadening
• Overshooting

_

• Too low
• Too low

• Too low

3. Reflectivity to rain rate conversion errors

• No unique relationship
• Hail and mixed precipitation

• Too low or too high
• Too high

4. Below beam effects

• Strong horizontal winds
• Evaporation
• Coalescence

• Too low or too high
• Too high
• Too low

• Too high

6. Hardware problems

• Incorrect calibration
• Wet antenna dome

• Too low or too high
• Too low

The net result of these limitations is rainfall overestimation by a factor of around two in convective events (Fresch, 2004); however, radars tend to underestimate rainfall at the farthest ranges and when mountains block the radar beam from reaching the rain. Furthermore, large hail in a storm can cause rainfall overestimates significantly larger than a factor of two.

Satellite Precipitation Estimate Limitations

• Tend to overestimate the area and magnitude of rainfall for large, slow-moving, and cold-topped convective systems
• Tend to underestimate rainfall from warm topped convective systems
• Tend to mislocate rainfall in regions of strong vertical wind shear
• Tend to miss rainbursts early in convective system development
• Susceptible to other large errors because the relationship between rainfall rates and satellite-measured radiances varies

Thus, satellite precipitation estimates compliment radar precipitation estimates and the rain gage network, but they do not replace them (Scofield et. al., 2003).

During the mid to late 1990s, the NWS sent its meteorologists to an intensive, 4-week course on how to use the new national network of Doppler weather radars as well as the latest on convective storm structure and evolution. Since then, meteorologists continued to hone thier skills through experience, training on radar improvements, follow-on training on severe storms, and case studies from real weather events. Nonetheless, the flash flood environment presents a number of uncertainties that even the best training can not resolve.

• Storm threat depends on storm severity as well as the amount and type of human activity in a location
• Weather observations, radar and satellite data can sum to information overload for forecasters, who must also consider the human impact of even the most innocent looking storms
• Forecasters don’t know the amount of human activity in every location
• Ordinary storms can have a major impact on people in the wrong place at the wrong time
• A significant number NM’s flash flood injuries and fatalities are caused by ordinary storms in highly used and/or hydrologically favored areas
• Continued evolution of radar algorithms employing detailed Geographic Information System mapping could help

Now that you understand the flash flood detection capabilities and limitations, we will look at an example of a flash flood that evaded detection and caught some campers off guard. Then, we will cover how to prepare and react appropriately to the flash flood threat.