Where (not) to measure rainfall Neil I. Fox University of Missouri - Columbia.

Where (not) to measure Where (not) to measure rainfallrainfall

Neil I. FoxNeil I. Fox

University of Missouri - ColumbiaUniversity of Missouri - Columbia

Three partsThree parts

One problem with radar and three (possible) One problem with radar and three (possible) solutionssolutions– Microwave linksMicrowave links– TV station radarTV station radar– CorrectionsCorrections

Using gauges?Using gauges?

Problems with radarProblems with radar

Lots of problems with getting accurate Lots of problems with getting accurate estimates of precipitation rateestimates of precipitation rate

Using dual-polarization should make Using dual-polarization should make estimates a lot better but…estimates a lot better but…

We are observing precip above the surfaceWe are observing precip above the surface

Many things can happenMany things can happen

Between the height of the beam and the Between the height of the beam and the surface surface

Precip can evaporatePrecip can evaporate It can growIt can grow It can be advectedIt can be advected We may even be above the precip and not We may even be above the precip and not

see it at allsee it at all

Dealing with thisDealing with this

Near to surface instruments Near to surface instruments More radarsMore radars Radar correctionRadar correction

Nexrad coverageNexrad coverage

Illustration of Wind-driftIllustration of Wind-drift

1-dimensional view 1-dimensional view with constant shear with constant shear profile in u-directionprofile in u-direction

Notice that when wind-Notice that when wind-drift is applied the drift is applied the original pixel is spread original pixel is spread to 2 neighboring pixelsto 2 neighboring pixels

In the example none of In the example none of the original rainfall the original rainfall contributes to the pixel contributes to the pixel directly below itdirectly below it

Data Sets Used for Case StudyData Sets Used for Case Study

From the Sydney 2000 Games C-POL radarFrom the Sydney 2000 Games C-POL radar

Wind Component Retrieval Methods from Wind Component Retrieval Methods from Sun and Crook (1994).Sun and Crook (1994).

CAPPI height 1500m, 45x45 grid, 2.5km CAPPI height 1500m, 45x45 grid, 2.5km grid spacinggrid spacing

U and V wind componentsU and V wind components

Results from Single Time StepResults from Single Time Step

The following images are taken from the The following images are taken from the midpoint of the event near Sydney.midpoint of the event near Sydney.

The comparison between the given The comparison between the given reflectivity and the wind drift correction will reflectivity and the wind drift correction will be shown.be shown.

There is a noticeable dispersion of There is a noticeable dispersion of reflectivity in the correction schemereflectivity in the correction scheme

Reflectivity vs. CorrectionReflectivity vs. Correction

Corrected Reflectivity MovieCorrected Reflectivity Movie

Accumulation vs. CorrectionAccumulation vs. Correction

Accumulation ErrorAccumulation Error

There is a noticeable There is a noticeable area of area of underestimation on the underestimation on the western edge of storm western edge of storm as it tracks as it tracks northeastward, northeastward, coupled with an area coupled with an area of overestimation, on of overestimation, on the order of 5-10kmthe order of 5-10km

Accumulation Error MovieAccumulation Error Movie

Introduce TopographyIntroduce Topography

Wind drift plus topographyWind drift plus topography

Incorporating Incorporating topography into the topography into the wind-drift scheme wind-drift scheme simply reduces the simply reduces the error between the error between the original and wind drift original and wind drift correction, esp. in correction, esp. in higher elevationhigher elevation

Reduces the distance Reduces the distance the wind impacts the wind impacts falling dropsfalling drops

Conclusions of thisConclusions of this

Errors predicted by this study are significantErrors predicted by this study are significant– Up to 30 millimeters in this case over the span Up to 30 millimeters in this case over the span

of 6 hours (~100% error in some regions on the of 6 hours (~100% error in some regions on the edge of the precipitation)edge of the precipitation)

– Convective cases result in higher accumulation Convective cases result in higher accumulation errors than stratiformerrors than stratiform

– Higher elevation reduces these errorsHigher elevation reduces these errors

Implications for assimilation into hydrological Implications for assimilation into hydrological and meteorological modelsand meteorological models

Getting closer to the groundGetting closer to the ground

More radarsMore radars Microwave attenuationMicrowave attenuation GaugesGauges

Use of broadcast radarsUse of broadcast radars

Regular low-level PPIRegular low-level PPI Flexibility when desiredFlexibility when desired

– But must put broadcasters’ needs firstBut must put broadcasters’ needs first

Relative Beam heightsRelative Beam heights

St Louis: KLSX

Fulton Pleasant Hill, KEAX

Benefits to be gainedBenefits to be gained

Low-level cold season precipLow-level cold season precip Low-level modification of warm season Low-level modification of warm season

precipitationprecipitation Low-level and small-scale (high-resolution) Low-level and small-scale (high-resolution)

rotationrotation BoundariesBoundaries

LimitationsLimitations

Quality ControlQuality Control CalibrationCalibration Staffing considerationsStaffing considerations Only PPIOnly PPI

Benefit to BroadcastersBenefit to Broadcasters

PublicityPublicity Community serviceCommunity service Radar is not really used most of the timeRadar is not really used most of the time

S-band

1 Beamwidth

250kW

25m tower

Example from the tornado outbreak Example from the tornado outbreak of 4 -10 May 2003of 4 -10 May 2003

Numerous rotations spotted (and shown on Numerous rotations spotted (and shown on TV) using the radar, but not detected by TV) using the radar, but not detected by NWS radarsNWS radars

Some (small) hook echo features clearly Some (small) hook echo features clearly observable were not so clear with NWS observable were not so clear with NWS radar imagery radar imagery

CharitonChariton County is roughly 105 km from the radar site in County is roughly 105 km from the radar site in Fulton and the beam height is roughly 1550 m.  KEAX is 150 Fulton and the beam height is roughly 1550 m.  KEAX is 150

km: 2750 m.  KLSX is 210 km away: 4300 m.km: 2750 m.  KLSX is 210 km away: 4300 m.

The Northern Audrain County Cell:The Northern Audrain County Cell:From Fulton :  56 km @ 670 mFrom Fulton :  56 km @ 670 m

From KEAX:   200km @ 4150 mFrom KEAX:   200km @ 4150 mFrom KLSX:   135 km @ 2300 m.From KLSX:   135 km @ 2300 m.

The Boone County Cell:From Fulton:  48 km @ 550m.From KEAX: 165 km @ 3100 m.From KLSX:  160 km @ 2850 m

Making a problem useful: Microwave Making a problem useful: Microwave attenuationattenuation

Microwave signals at the frequencies Microwave signals at the frequencies typically used for cellphones are strongly typically used for cellphones are strongly attenuated by precipitationattenuated by precipitation

Bolton ProjectBolton Project

Transmitter Receiver

Transmitter Receiver

Effect of dumb clipart rain: differential attenuation

From Rahimi et al. submitted to JGR

Using gaugesUsing gauges

Gauges give us a measure of surface Gauges give us a measure of surface precipitationprecipitation– At a pointAt a point– They have bigger errors than you thinkThey have bigger errors than you think

Especially when installed by council Especially when installed by council workmenworkmen

Using gauges 2Using gauges 2

If we’re carefulIf we’re careful If we understand the errorsIf we understand the errors If we recognize that we’re using gauges!If we recognize that we’re using gauges!

No cheese today GromitNo cheese today Gromit

Poor Gromit – didn’t check the Poor Gromit – didn’t check the forecastforecast