HomeUpSearchMail
NEW

Validating measurements at Rijn en Lek

Validating measurements at Rijn en Lek by Victor Reijs is licensed under CC BY-NC-SA 4.0

Introduction

• Conclusions
• References
• Acknowledgements

Rijn en Lek info

The height of the anemometer (wind speed and direction) is around 25m from the road level (gallery: 9m + space gallery and sail: 1m + sail length:12m + flagpole: 3m) and some 28m from the river's floodplain level.

Rijn en Lek is in a very open environment (certainly at shaft height). The trees at the castle Duurstede estate in the west and the two churches in the NNW can have influence. The rest is very open in a large portion of the wind rose (N to SW), and will have a likely z_0m=0.03m (the river-delta), while the rest (town direction) might be z_0m=0.5m or 1m.

A picture made by Justin Coombs ((c) 2024)

Alignment error

Photometric method

Comparing with meteorological stations

Location of meteorological stations

Wind speed

The average wind speed over the year can be used to normalise the meteorological stations' wind speed towards the Wijk bij Duurstede location:

The yearly average wind speeds are [m/sec]:

As the wind speed of the meteorological stations to reference to a height of 10m, we need to to determine the speed at the mill's shaft height (which is 25m) to be able to compare with the shaft height of Rijn en Lek.
It is assumed the roughness length at meteorological stations is z_0a=0.03m and assuming the roughness length at the mill's anemometer is z_0m=0.5m.
The speed at anemometer height (25m) will have an additional increase of [Beljaars, 1979, page 4]:
u_shaft=u₁₀*(ln(60/0.03)/ln(10/0.03))*(ln(25/0.5)/ln(60/0.5)) = u₁₀* 1.07

Remark: The z_0m is not really known as there might not exist an logarithmic ABL at the mill (now used a z_0m=0.5m being: numerous obstacles, x/H ≈ 10).

Wind direction

The average wind direction over the summer (red / black arrows) and winter (blue) can be used to normalise the meteorological station's wind direction towards the Wijk bij Duurstede location:

Sanitising the smartmolen database

• rows with a wind speed below 1.5m/sec (1Bft) and higher than 13.9m/sec (7Bft) are removed
  
• rows that have an MILL_ORIENTATION_CONFIDENCE of lower than 20 are removed
  
• remove rows with an absolute direction angle difference compared to Rijn en Lek of larger than 90 degrees.
  
• between 6/12/2024 10:00 and 7/12/2024 10:00 the anemometer speed was constantly 2.3m/sec and direction around 296deg (see also below speed graph and in the green circle of this graph). These were also removed.
  This happened gain in april 2025. New firmware would be needed in the weather station, but we would need a cherry picker to install it using an USB cable. The software of smartmolen has been updated, so it recognises this fault.
  
• Replaced the EHRD data with HERW data.
  

Analysis of the measurements

Comparing with Rijn en Lek

Wind speeds

The wind direction of the meteorological stations can be seen below with the wind direction at Rijn en Lek mill on the x-axis (for 1Bft upto and including 7Bft). The meteorological station's speeds have been compensated for the height of the anemometer and normalised with the average yearly wind speed.

Speed comparison at the mill cap

Expectations

What is to be expected when looking at the wind speed effects around the anemometer, which is 2m above the mill cap? The wake will have influence, by increasing or decreasing the windspeed at the anemometer. The following effects could happen (negative angle of attack [AoA] means a backed wind and positive AoA means veered wind):

• If the wind comes from the back (AoA lower -150 or higher than150deg).
  It is expected that the wind speed of the meteorological station will be similar (close to 100%) to the anemometer's if wind is from the NNE to SSW; or the meteorological station's could be higher than the anemometer's if the wind comes over the town (SSW to NNE).
  This situation is not happening in reality a lot. In the below analysis this is could not be checked, as there were not enough measurements for this situation.
  
• If the wind comes from the front (AoA between -30 and 30deg).
  It is expected that the wind speeds of the meteorological station will be higher or smaller than at the anemometer's if wind is from the NNE to SSW. The higher or smaller depends on the effect of the wake due to the mill cap (see also). The meteorological station's speed would likely be higher than the anemometer's if the wind comes over the town (SSW to NNE).
  This situation happens a lot.
  
• If the wind comes from the side (AoA between -45 and -135deg or between 45 and 135deg).
  This will have an in between effect of the earlier two bullets.
  This situation happens a lot between 45 and 105deg or -45 and -105deg. In the below analysis angles outside this range could not be checked, as there were not enough measurements for that range.

Evaluation

1. The South to West sector (most winds are from that direction) is from floodplains of Rijn en Lek rivers. So perhaps the z_0m is closer to 0.03m (Open flat terrain) for those directions
2. The floodplains are lower than road level, some 3m lower. This increases the wind speed (at 28m) at the meteorological station.
3. The accuracy of the anemometer is around 5%.
4. Would wind channeling (due to churches) cause a high SpeedFactor for Rijn en Lek?
5. To what do we compare to:
   
• the z_0a of the meterological station (0.03m) and the average z_0m at Rijn en Lek (0.1m)
  This concept is used.
  
• a more specific z_0m in a certain direction from Rijn en Lek (say between 0.05 and 1m)
  
• only compensate the height of the meteorological station and its z_0a (0.05m)
  This has a difference of some 9% compared to the 1^st concept.
  
6. The anemometer might be in the wake of the mill cap; this would introduce a higher or lower wind speed, depending where the anemometer is in the wake.
   Rijn en Lek's anemometer is at a height of 1.09*H_cap, it is not 100% sure what would happen at this height, will it reduce or increase the wind speed.
   To be verified with a CFD of the mill-body.

The influence of the first three bullets could be around 15%. So this could rectify the Rijn en Lek speed comparison, but this all is still uncertain.

How are speeds distributed

If we determine how a AoA-sector influences the speed difference between the meteorological station and Rijn en Lek:

In above graph each AoA-sector is 15degrees wide (and the sector middle is noted on the x-axis). Above the red dotted line the meteorological station's wind speeds are higher than at Rijn en Lek. The gray line (at around 91%) shows the average speed difference looking at all AoA directions.
An example: An AoA between mill cap and wind of +45deg (a range +37.5 to +52.5deg), gives for EDHL ~82% (orange) of Rijn at Lek; and for HERW ~100% (purple) of Rijn at Lek.

As most wind comes from SW; most negative AoA are coming from the river bed, while most positve AoA come from the town. Over the river bed (negative AoA) one would expect speeds closer to 100% than coming over the town (positive AoA) which speed is then expected to be lower than 100%. This is the opposite behavior as seen in above graph.
Was an error made in determine from what AoA the backed/veered wind comes from? Or is something else playing?

Wind direction

The relative wind direction of the meteorological stations can be seen below (relative to the wind direction at Rijn en Lek mill). The meteorological station's directions have been compensated for the average winds direction in winter. Rijn en Lek wind direction (on x-axis) has been compensated for the alignment error.
The distribution of directions can be seen below:

The standard deviation of relative meteorological stations' direction is some 25deg. Assuming that the standard deviation of the meteorological stations' and anemometer's wind direction are similar, they are both around 15deg.

Changing mill cap direction

From 24/11/2024 05:00 until 25/11/2024 01:00 one sees a slowly moving direction of the mill cap (from 198 to 204deg), while no manual turning happened (in the red circle):

This is at the time that storm Bert (in particular its warm sector) raged over The Netherlands.
Explanation could be: a) the cap was pushed to the side of the curb due to the wind force or b) the magnetometer calibration process caused this.

Wind distribution seen over town

KNMI meteorological stations Deelen and Herwijnen (orange dots, looks that the majority has a resolution of some 5deg) has been used below, as these two have the smallest difference with wind direction at Rijn en Lek (red dots, 1deg resolution) and they are close meteorological stations to Rijn en Lek.

Herwijnen, HERW	Deelen, EHDL

The looks to be some realistic mapping between Rijn en Lek (LOCAL), Deelen (EHDL) and Herwijnen (HERW). Although the KNMI meteorological stations Deelen and Herwijnen still looks to have lower wind speed than Rijn en Lek.

Matching mill anemometer readings with calculations

Rijn en Lek

The above compensated direction and wind speed data have been used. The speedfactor has been derived by dividing the Rijn en Lek (LOCAL) anemometer's wind vector by the meteorological (HERW or EHDL) station's wind vector. From these, the (non-weighted) average speedfactor magnitude per 10deg sector has been derived.

<right click a picture, to see a larger version in another browser window>

Rijn en Lek&HERW	Rijn en Lek&EHDL	Rijn en Lek&Open-Meteo

• Looking at the anemometer measurements
  
• Mesurements are for the period 8 November 2024 to 13 March 2025, so winter time (November upto and including March). The trees are leaveless.
• Leaved trees are between May and upto inlcuding September
  
• It looks that the anemometer speeds are most of the time higher than at Herwijnen. That would not have been expected.
• Non weighted sectors were used.
• The speedfactors (anmometer/meteorological) have been decreased with a scaling factor of ~0.82 to match somewhat the calculations.
  
• The wind did not blown a many times from North North West), so the uncertainty (the yellow dots around the white curve) of the Speedfactor is larger.
• Open-Meteo data (in Winter time) matches reasonably well with the calculations (of Summer time)
  We have around 80days of measurements, this migth not be enough to reduce uncertainty. Longer period of measuring is needed,. As we need to keep season in minds; so over multiple (2???) years.
  
• There might to be a blob on the East, is this due to possible wind channeling (high trees)?
• In one graph there looks to be a blob in the North-West, could this be wind channeling between the churches?
• Remember there is a scale factor that mapped speedfactor on the calculations.
  
• Looking at the Rijn en Lek calculations (aVR_nag and aVR_belj)
  
• Difference between speedfactor with anemometer and the calculation is smallest for aVR_belj. So aVR_belj is being used in further analysis.
  
• Non weighted sectors were used.
  
• The trees are calculated as leafed, so summertime.
  
• Wind channeling is not covered by the calculations (one would need CFD for that).

Other mills

• Checking the scaling factor at two other mills (Impington&Mildenhall, Impington&Open-Meteo, Upminster&London City, Upminster&Open-Meteo, and Labbus&Open-Meteo); one gets a scaling factor between 68% to 108% (say around ~80%). Open-Meteo uses the the ECMWF-IFS 9km model. z_0a is being used within its weather models of Open Meteo (pers.comm. Open-Meteo, 2025), see also.
  Remark: In below the z_0m of mill and z_0a of Meteo have been set equal using the value in this table.
  
• The preferred period is May to September, but in many cases not enough measurements were available and then a larger period has been taken (Impington and Labbus).
  

  Impington&EGUN	Impington&Open-Meteo	Upminster&EGLC	Upminster&Open-Meteo	Labbus&Open-Meteo

  
  
  • There are differences between measurements and calculations, so needs to be understood why there is a difference!!!!
    
  • For Impington: The anemometer is at the fan tail platform, so slightly behind the cap, but don't think this can explain this difference (as the scale factor would compensate this). I also don't think that the nearby Western trees could produce backpush for wind from the East (that would decrease the windfactor of Eastern directions instead of increasing).
    There could be wind channeling on the East (which would increase the speedfactor as seen).
    
  • For Impington when looking at Open-Meteo the differences are smaller with the calculations.
    Could this be due to seasonal effects?
  • For Upminster: Both EGLC and Open-Meteo compare not that well in the South-East with the calculations.
  • For Labbus: the differences are all around the speedfactor radar...
    
  • Remember there is a scale factor that mapped speedfactor on the calculations.
  • The uncertainty (the yellow dots around the white curve) of the Speedfactor can not really explain the differences.
  • The z_0a has been calculated (using a logarithmic ABL) from the wind speed at 10, 80, 120 and 180m derived from Open-Meteo Weather Forecast. This gives the following data (the Latitude and Longitude are Open-Meteo's grid points closest to the Mill):
    

    Mill	Latitude	Longitude	Environment	z0a
    Impington	52.24°	0.12°	quite open landscape, 400m from Impington and Cambridge	0.65m
    Labbus	52.68°	8.84°	just outside Sulingen village	0.41m
    Upminster	51.56°	0.24°	London suburb	0.93m

    
    

Some conclusions

We need to remember that a stable calibration of the magnetometer has now happened for some 17 weeks (Nov. 8, 2024 to March 1, 2025). Longer averaging might change the results in the future. If we need to have seasonal numbers, a two year period migth be needed.
There were battery problems with one or two magnetometers between 21 January and 17 February 2025.
<more to be added>

References

Acknowledgements

Disclaimer and Copyright
HomeUpSearchMail