Rainfall Measurement Details

Arable Rainfall Detection

- Danielle Watts, PhD | Vice President of Data Science at Arable

Rainfall is generally the most variable hydrologic element over any space, and its characterization one of the most commonly needed and difficult to rectify. Rainfall has multiple modes for non-stationarity-- the small-scale variability induced by processes like cloud formation, droplet formation, and wind, and larger scale variability induced by the movement of rainfronts, local and regional geography, and, in some cases, plant canopies.

Comparing very localized data, such as you receive from the Arable Mark, to either a forecast or a distant weather station may show some of this variability. Forecasts are usually looking at the likely conditions over a large area; individual points may see substantially more or less rainfall. The same is true for the nearest weather station; many weather stations in the US are at airports with conditions that may be quite different from the ones you see in your fields. The spatial variability can be remarkable, research has demonstrated

  • Up to 100% variability between rain gauges within 500 meters of each other (Jensen and Pedersen 2005)
  • Up to 26% (or more) variability between gauges within 250 meters of each other  (Pedersen et al. 2010)
  • And the variability is strongly influenced by the type of rainfall that is occurring (Emmanuel et. al. 2012)

You may also witness some spatial variability when comparing the rainfall among your devices if you have them in many fields. The value of a network of devices is that you get the truest picture of your weather conditions!


Mokondoko.png
Figure 1. There is significant spatial variability in rainfall even across small distances. Figure from Mokondoko et al 2018.

 

Calibrations

Arable rainfall is detected through a patented acoustic disdrometer. The disdrometer is effectively “listening” for raindrops to hit the top dome and then transforming the sound through to an energy bin, which we can then map to individual rain droplet sizes. The accumulation of each droplet provides us our rainfall rates. We are constantly working to improve our rainfall calibrations with field stations in Hawaii, USA and British Columbia, Canada. We are opening a third in Colorado, USA. Each of these is equipped with an OTT Parsivel2 laser rainfall sensor.

 

Our current calibration is accurate within 0.2 mm/hour for rain droplets >1 mm (>0.1 cm) in diameter. In Table 1 we can see that for most rainfalls we are gathering the majority of the rainfall distribution. However, if the rainfall is dominated by a very fine drizzle (often characterized as rainfall rates < 2.5 mm/hr) then we may miss a proportion of the rainfall. Upcoming algorithm releases will include updated calibration that will allow us to more accurately capture that fine misting rain.

Table 1. Laws-Parsons drop size distribution for various precipitation rates.

Drop diameter (cm)

Rainfall rate (mm/hr)

0.25

1.25

2.5

5

12.5

25

50

100

150

0.05

28.0%

10.9%

7.3%

4.7%

2.6%

1.7%

1.2%

1.0%

1.0%

0.1

50.1

37.1

27.8

20.3

11.5

7.6

5.4

4.6

4.1

0.15

18.2

31.3

32.8

31.0

24.5

18.4

12.5

8.8

7.6

0.2

3.0

13.5

19.0

22.2

25.4

23.9

19.9

13.9

11.7

0.25

0.7

4.9

7.9

11.8

17.3

19.9

20.9

17.1

13.9

0.3

 

1.5

3.3

5.7

10.1

12.8

15.6

18.4

17.7

0.35

 

0.6

1.1

2.5

4.3

8.2

10.9

15.0

16.1

0.4

 

0.2

0.6

1.0

2.3

3.5

6.7

9.0

11.9

0.45

   

0.2

0.5

1.2

2.1

3.3

5.8

7.7

0.5

     

0.3

0.6

1.1

1.8

3.0

3.6

0.55

       

0.2

0.5

1.1

1.7

2.2

0.6

         

0.3

0.5

1.0

1.2

0.65

           

0.2

0.7

1.0

0.7

               

0.3

 

The upcoming releases will also allow us to filter out extraneous noises that are inadvertently picked up by the internal microphone. Birds landing on the dome and passing tractors are the two most common noise contaminations, and can lead to the Arable Mark reporting more rainfall that it should. If you are observing much higher rainfall rates than you expect for one or more of your devices, we recommend moving the device away from any heavy machinery, and putting up bird deterrents (if your device is showing signs of bird perching).

Comparison to other rain gauges

It is common for disdrometers to have variable sensitivity to low rainfall rates (see Tokay 2013 or Angulo-Martinez et al. 2017 for comparison studies), and for rainfall gauges in general to perform variably to a reference (for an exhaustive study on this, please see WMO 2009; summarized in Table 2, here).

 

Tipping bucket rain gauges are perhaps the most commonly deployed gauge type in irrigation and agricultural settings. These types of gauges are often excellent for small-to-medium rainfall events, and are valued for their reliability in remote locations. However, tipping bucket gauges are known to underreport large rainfall events, particularly those events with wind, and the better gauges are physically designed to reduce evaporative & wind losses. The necessary dynamic calibrations are often not performed for years after they are due, and so errors induced by fowling and drift in the tipping mechanisms (often induced by the expansion of water) are compounded. The % error between gauges, even when well calibrated, can range up to 10% (UK EA 2004). The drift from calibration can be as much as 3-8% within the first year in the field, although sensors vary in their resistance to drift.

We strongly recommend customers who are evaluating Arable Mark data to other gauges consider distance between sensors and the time since any in-house rain gauges were calibrated. We are, ourselves, constantly working on our sensor calibrations; calibrations are everything to us!

Table 2. Rainfall rates by model compared to a rainfall reference. Source WMO 2009.

Rain gauge model

Measurement type

rainfall = a*rainfall referenceb

a

b

R2

RIM7499020-McVan

tipping bucket

1.31

0.9

0.68

AP23-PAAR

tipping bucket

1.15

0.96

0.85

R01 3070-PRECIS MECANIQUE

tipping bucket

1.08

0.95

0.77

PT 5.4032.35.008-THIES

tipping bucket

1.01

0.99

0.85

R 102 -ETG

tipping bucket

1.01

0.99

0.88

DQA031-LSI LASTEM

tipping bucket

1.06

0.96

0.72

UMB7525/I-SIAP-MICROS

tipping bucket

0.92

1.02

0.73

PMB2 -CAE

tipping bucket

0.78

1.05

0.87

RAIN COLLECTOR II -DAVIS

tipping bucket

1.16

0.92

0.73

LB-15188-LAMBRECHT

tipping bucket

1.21

0.96

0.81

PP040-MTX

tipping bucket

0.96

1

0.79

ARG100-EML

tipping bucket

1.21

0.92

0.75

MRW500-METEOSERVIS

weighing gauge

1.01

0.98

0.74

VRG101-VAISALA

weighing gauge

1.12

0.75

0.12

PLUVIO-OTT

weighing gauge

0.98

1

0.9

PG200-EWS

weighing gauge

0.98

1

0.81

T200B -GEONOR

weighing gauge

0.96

1

0.89

TRwS-MPS

weighing gauge

1.09

0.95

0.59

PWD22-VAISALA

laser optics

0.81

0.94

0.51

PARSIVEL-OTT

laser optics

0.82

1.1

0.77

LPM-THIES

laser optics

0.93

1.07

0.8

WXT510-VAISALA

impact detection

1.72

0.91

0.74

ANS 410/H-EIGENBRODT

pressure sensor

1.09

0.96

0.67

Electrical raingauge-KNMI

level sensor

1.05

0.97

0.82

 

Hilo_OTT_Setup.JPG

Figure 2. The Arable Mark and the OTT Parsivel2 at HILO in Hawaii, USA.

OTT_in_CO.png

Figure 3. The Arable Mark and a OTT Parsivel2 at the National Center for Atmospheric Research, Colorado, USA.

 

For more reading

Angulo-Martinez, M., Begueria, S., Latorre, B., & Fernandez-Raga, M. (2017). Comparison of precipitation measurements by Ott Parsivel and Thies LPM optical disdrometers. Hydrological Earth Systems Science Discussions, (in review - 672).

 

Emmanuel, I., Andrieu, H., Leblois, E., & Flahaut, B. (2012). Temporal and spatial variability of rainfall at the urban hydrologic scale. Journal of Hydrology, 420-431: 162-172.

 

Jensen, N.E., & Pedersen, L. (2005). Spatial variability of rainfall: Variations within a single radar pixel. Atmospheric Research, 77(1-4): 269-277.

 

Mokondoko, P., Manson, R. H., Ricketts, T. H., & Geissert, D. (2018). Spatial analysis of ecosystem service relationships to improve targeting of payments for hydrological services. PLoS ONE, 13(2).

 

Pedersen, L., Jensen, N.E., Christensen, L.E., & Madsen, H. (2010). Quantification of the spatial variability of rainfall based on a dense network of rain guages. Atmospheric Research 95(4): 441-454.

 

Tokay, A. (2013). Comparison of raindrop size distribution measurements by collocated disdrometers. Journal of Atmospheric and Oceanic Technology, 30: 1672-1690.

 

United Kingdom Environment Agency. (2004). Evaluation of tipping bucket rain gauge performance and data quality. Science Report: W6-084/SR. 63pgs.

 

World Meteorological Organization (2009). WMO field intercomparison of rainfall intensity gauges. Instruments and Observing Methods Report No. 99. 290pgs.