An investigation of factors affecting the accuracy of Thies disdrometers

An investigation of factors affecting the accuracy of

Thies disdrometers

TECO-2008 St Petersburg, November 29th, 2008 Graham Upton

Graham Upton

and

Dan Brawn

University of Essex, Colchester, UK

Thanks


• To the UK Met Office (for providing the data)

• To the EPSRC (Grant:EP/P500214/1) (for the money)

•To Dan Brawn (for letting me come to St. Petersburg!)

The Thies disdrometer


•Uses a 780 nm infrared parallel light-beam with a cross-section of 45.6 cm2.

•Particle diameters recorded using 22 bins

•Fall speeds recorded using 20 bins 1 7

Modern Optical DisdrometersThies Disdrometer

The Parsivel Optical Disdrometer has a similar operation

The test site: EskdalemuirUndertheseclouds!


Overview of the instruments


Looking North-West

The disdrometers (3 Thies & 1 Parsivel)


Looking towards the North-West

Close up of the three Thies


Looking towards the West

T1

T3

T2

Arrangement of the four disdrometers


Disdrometer Pole

Height of disdrometer

above ground

Orientation(pole end

given first)

T1 A 2m North-South

T2 B 1.75m North-South

T3 B 2m West-East

Parsivel C 2m West-East

Restrictions on comparisons


1. Counts in lowest Thies size bin ignored (to permit comparison with Parsivel: diameters considered are therefore between 0.25mm and 8mm)

2. Use only hours with complete data (there were some computer problems: this restriction ensures that individual minutes are correctly matched)

3. Use only minutes in which the average Thies count was 50 drops (to eliminate records not caused by rain)

Average drop counts (per minute)


1. Excellent agreement between the Thies

2. Number of drops recorded by the Parsivel is only about 75% of that recorded by a Thies

Rain rate T1 T2 T3 Parsivel

< 1 mm/hr 286 286 284 212

> 1 mm/hr 886 898 897 719

Comparison of Thies and Parsivel drop count distributions


1. Log scale on vertical axis2. Parsivel’s consistently higher (though small) counts of larger diameter drops3. Thies’s much greater count of the smaller drops

Notes:

Overall equality of counts between the three Thies does not mean that

they agree all the time!


Time T1 T2 T312:03 1618 1645 1290

: : : :12:11 2222 2279 1879Total 18090 18049 15428

Nine minutes of heavy rain during a storm at midday on June 30th, 2007

Overall equality of counts between the three Thies does not mean that

they agree all the time!


Time T1 T2 T312:03 1618 1645 1290

: : : :12:11 2222 2279 1879Total 18090 18049 15428

North-South North-South West-East Pole A Pole B Pole B 2m 1.75m 2m

Shielding: the effect of wind direction


Winddirection

110°- 130°- 150°- 170°- 190°- 210°- 230°- 250°- 270°- 290°-

T1 > T3(% of

minutes with wind in given

direction)

85% 64% 39% 27% 26% 31% 48% 56% 61% 50%

No. ofminutes

77 196 126 213 178 123 226 204 132 65

These are statistically significant variations from 50%



Winddirection

110°- 130°- 150°- 170°- 190°- 210°- 230°- 250°- 270°- 290°-

T1 > T3(% of

minutes with wind in given

direction)

85% 64% 39% 27% 26% 31% 48% 56% 61% 50%

No. ofminutes

77 196 126 213 178 123 226 204 132 65

The two disdrometers are at the same height above ground



Winddirection

110°- 130°- 150°- 170°- 190°- 210°- 230°- 250°- 270°- 290°-

With wind in given

direction,% of

minutes T1 > T3

85% 64% 39% 27% 26% 31% 48% 56% 61% 50%

No. ofminutes

77 196 126 213 178 123 226 204 132 65

The two disdrometers are at the same height above groundT1 is oriented North-South T3 is oriented West-East

Shielding and drop size


T3

T2

The next results compare counts for

T2 and T3

(very close in space)

A useful comparison statistic


mij = Median(Counti / countj)where counti denotes the one-minute drop count for

disdrometer iOnly minutes where both disdrometers count at least 50 drops are

included



Drop diameter (mm)

Wind from the South

m23

Wind from the Westm32

0.125- 0.81 0.85

0.25- 0.86 0.96

0.375- 0.99 0.98

0.5- 1.00 0.98

0.75- 0.94 0.96

1- 0.98 0.96T2 is oriented North-South (T3 is not)

Test statistic is m23 = Median(count2 / count3)



Drop diameter(mm)

Wind from the South

m23


0.125- 0.81 0.85

0.25- 0.86 0.96

0.375- 0.99 0.98

0.5- 1.00 0.98

0.75- 0.94 0.96

1- 0.98 0.96T3 is oriented West-East (T2 is not)

Test statistic is m32 = Median(count3 / count2)



Drop diameter(mm)

Wind from the South

m23


0.125- 0.81 0.85

0.25- 0.86 0.96

0.375- 0.99 0.98

0.5- 1.00 0.98

0.75- 0.94 0.96

1- 0.98 0.96

Smallest drops

Greatestloss

Test statistics are m23 = Median(count2 / count3) and m32 = Median(count3 / count2)

Shielding and wind speed


Wind speed(m/s)

Wind from the South

m23


0- 1.00 1.00

1- 0.97 0.97

2- 0.92 0.99

3- 0.88 0.91

4- 0.89 0.92

5- 0.86 0.78

Very light wind

No problem


Shielding and wind speed


Wind speed(m/s)

Wind from the South

m23


0- 1.00 1.00

1- 0.97 0.97

2- 0.92 0.99

3- 0.88 0.91

4- 0.89 0.92

5- 0.86 0.78Strongwind

Serious problem


The combined effectof drop diameter and wind speed


The worst case occurs with small drops in strong winds

Drop diameter(mm)

Wind speed<3m/s > 3m/s

< 0.25 mm 0.87 0.80

> 0.25 mm 0.98 0.93

Entries are m23 ratio

Fall-speed calibration

Values of m21 = Median(count2 / count1) Values of m23 = Median(count2 / count3)


Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.23 0.93 0.99

> 0.25 0.95 1.04 1.03

All 1.16 1.00 1.01

Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.55 0.87 0.95

> 0.25 0.95 1.02 1.01

All 1.33 0.97 0.99First result: All three Thies disdrometers are recording

roughly equal numbers of both small drops and large drops




Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.23 0.93 0.99

> 0.25 0.95 1.04 1.03

All 1.16 1.00 1.01

Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.55 0.87 0.95

> 0.25 0.95 1.02 1.01

All 1.33 0.97 0.99Second result: All three Thies disdrometers are recording

roughly equal numbers of the faster falling drops




Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.23 0.93 0.99

> 0.25 0.95 1.04 1.03

All 1.16 1.00 1.01

Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.55 0.87 0.95

> 0.25 0.95 1.02 1.01

All 1.33 0.97 0.99Third result: T2 assigns fewer of the fast-falling drops

to the smaller drop-size categories than do either T1 or, particularly, T3




Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.23 0.93 0.99

> 0.25 0.95 1.04 1.03

All 1.16 1.00 1.01

Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.55 0.87 0.95

> 0.25 0.95 1.02 1.01

All 1.33 0.97 0.99Fourth result: T2 assigns more drops to the slow-falling speeds

than either T1 or, particularly, T3




Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.23 0.93 0.99

> 0.25 0.95 1.04 1.03

All 1.16 1.00 1.01

Drop diameter

(mm)

Fall speed (m/s)

< 1 > 1 All

< 0.25 1.55 0.87 0.95

> 0.25 0.95 1.02 1.01

All 1.33 0.97 0.99Fifth result: T2 assigns more drops to the slow-falling speeds

than either T1 or, particularly, T3. This is especially true of the smallest drops.

Summary of results


1. Overall counts agree well from one Thies to another

2. Thies counts are appreciably greater than those from a Parsivel

3. Drop size distributions agree well from one Thies to another

4. The drop size distributions seen by Thies and Parsivel are markedly different (as are the water volumes)

5. Thies suffer from shielding with counts in the smallest drop sizes reduced by as much as 20% in strong winds

6. Fall speeds can vary appreciably from one Thies to another --- particularly in the case of the smaller drops

Summary of results








Summary of results








Summary of results








Summary of results








Summary of results








Summary of results








Thank you for your attention. Any questions?

Documents

An investigation of factors affecting the accuracy of Thies disdrometers