Long Term Project

The effects of building structure on wind measurements as measured by Sonic Anemometers, Cup Anemometers, and propeller and vane anemometers Larissa Reames Windy C Semester project for METR 3613, Fall 2007 Group members: Kevin Haghi, Sean Waugh, Kenneth Jackson

1

Abstract The purpose of this experiment is to discover the differences in measurement between a WMOstandardized observing site and two masts atop an irregularly shaped 6-story building. To do this, Sonic Anemometers, cup anemometers, as well as vane and propeller anemometers were used. Data was collected from three different observing stations, 1 at the WMO site and 2 on the top of the roof, and manipulated to be in the desired form. From these data, it was determined that the terrain (in the cast of the roof mast, the face of the building) downwind of a measurement site that most greatly affects the wind speeds and directions measured by the instruments at that mast. The observations made in this experiment exhibit the effects that large, oddly shaped structures have on the wind dynamics within their vicinity. This experiment also displays the importance of site selection in the accuracy of wind measurements, albeit in an extremely exaggerated case.

1. Introduction Wind is the driving force in meteorology. Whether it be a small-scale eddy caused by a small tree or a destructive category 5 hurricane with sustained winds of 160mph, understanding wind and how it behaves is essential to understanding nearly all meteorological concepts. Although this experiment focuses much more on the micro-scale and seemingly insignificant winds around the National Weather Center in Norman, Oklahoma or the Oklahoma Mesonet Norman site, the same basic principles that will be addressed in this project can be applied to meso- and macro-scale phenomena. This experiment seeks to discover the difference in recorded wind measurements between a WMO-standard observation site and two roof-top masts set ontop of a building in an urban area, and to explain why those differences occur. To do so, it is necessary to understand how the wind measurements are made, along with principles that govern wind flow, especially over and around obstructions. Also, it is important to elaborate on meteorological standards for naming and quantifying wind measurements as well as the standards for site selection and how this experiment can be used to show the importance of following those standards for site selection. The instruments used in this experiment were a three-directional sonic anemometer, a cup anemometer, and a combination propeller-vane anemometer. A sonic anemometer works on the

2

principle that wind direction and speed, along with temperature and humidity, effect the speed of sound over a small distance. The sonic anemometer contains three sets of sensors, each set aligned vertically so that the sensors point towards each other. One sensor releases an extremely short pulse of sound, which the other sensor in the set receives. From the difference it time that it took for that pulse to reach the second sensor compared to the normal speed of sound, the sonic anemometer computes the u-, v-, and w-component of the wind, along with temperature and humidity. A cup anemometer works on the principle of drag force. Three small, plastic bowlshaped “cups” are attached to a central spire at a short distance, all facing in the same clockwise direction. The spire is set vertically, and as the wind blows, it pushes the cups and the spire spins. One would think that the forces on both sides of the cup would be equal and thus prevent the spire from spinning, but the drag force on the concave side of the cup is much greater than on the concave side. This creates a force differential on the cup, thus causing the spire to spin. From this spin, the cup anemometer can determine wind speed ONLY. The propeller-vane anemometer is designed like a prop plane without wings: a propeller in the front and a high tail in the back. When the wind blows, it directs the entire anemometer in the direction of the wind speed and spins the propeller on the front. From this, the anemometer can compute the wind direction and speed. Many different forces can cause wind to blow. On a large scale, pressure gradients cause longlived winds. On a slightly smaller scale, the inflow and outflow from strong thunderstorms can cause very high winds over a short period of time. Extremely tight pressure gradients, such as those seen in tornadoes, can cause devastatingly high wind speeds as well. On an even smaller scale, and in a realm much more relevant to the discussions of this experiment, objects as small as trees and as large as skyscrapers can modify the wind that is already blowing as a result of an outside force, such as those mentioned above. Objects as far away as 400km and at an angle relative to the anemometer as small as 1° can cause eddies and modifications to the wind at a measurement station (Fujita et. al. 1981). Thus, when you consider objects as close at 50m and as tall as 5 story buildings, stationary, seemingly harmless objects can have a hefty effect on wind measurements.

3

It is no wonder, considering the information above, that the World Meteorological Organization (WMO) has very strict standards for site selection in order to minimize the effects that obstructions have on the measurements taken at a site. The most important specification that the WMO makes relative to this experiment is that all measurements be taken at a distance of 10times the height of an obstruction away from said obstruction (WMO 1989). When considering this experiment, it is important to keep in mind that the site at the National Weather Center is about as far away from WMO-standard as any site could be, while the Norman Mesonet site is a WMO-standardized site. The data presented hereafter thus show the importance of site selection, albeit using an extremely exaggerated case. It is important to make a final note on the meteorological standard for naming and quantifying wind measurements. Winds are always labeled and recorded by the direction from which they come, with a positive u wind from the west and a positive v wind from the south. If a wind direction is given in degrees, a 0/360° wind is from the north and the degree measurements increase clockwise (i.e. 90° wind is from the east).

2. Experimental Details and Methods The data used for this experiment were collected from two observing stations: the Oklahoma Mesonet site in Norman Oklahoma (simply Mesonet from here on) and two masts atop the National Weather Center building in Norman, Oklahoma (NWC from here on). The data collected from these stations were from 12:01am on September 25, 2007 to 11:59pm on October 2, 2007. The layout and overall description and operation of these two observing stations follow. In addition to site description, the specific methods of data analysis used in this experiment also follow. The Mesonet site consisted of a single mast set on a 100m2 plot of land equipped with a Campbell Scientific Inc. (CSI) CR23X-TD data logger, solar panel, radio transceiver, lighting rod, and other atmospheric and land sensors, of which the most important for this experiment were the R.M. Young 5103 wind monitor and the R.M. Young 3101 cup anemometer. The R.M. Young wind monitor was a combination propeller and vane anemometer and was placed at 10m 4

above ground level and, combined with the data logger, output 5 minute averages of wind speed, wind direction, standard deviation of wind speed, along with the maximum 3-second wind speed in those 5 minutes. Its low threshold for wind speed measurements was 1 ms-1 for the propeller and 1.1ms-1 for the vane. It could also measure wind speed velocities up to 60ms-1, and could measure -1

wind speeds at an accuracy of ±.3ms and wind

Fig. 1: Oklahoma Mesonet site in Norman, Oklahoma direction at an accuracy of ±3°. An R.M. Young

cup anemometer was placed at the 2m and 4m above ground level and measured only wind speed at an accuracy of ±.5ms-1. The cup anemometer reported only a 5 minute wind speed average (McPherson et al. 2006). The site design and instrumentation complied with the WMO standards; most importantly the site is at a distance of 10-times the height of any obstruction away from said obstruction. The Mesonet site can be seen in Fig.1. In this Fig. 2: Rooftop view of the NWC observation site looking northwest. Note the large structure north of the west mast

experiment, all data taken from the Mesonet were assumed to be the correct measurements against which all other data were compared.

The NWC site consisted of two masts, one mast to the east and one mast to the east, here on referred to east mast and west mast respectively. Both masts can be seen in Fig.2, with the west mast towards the left and the east mast on the right. Each identical mast was a 10ft non-penetrating roof mast equipped with an R.M. Young wind sentry set Fig. 3: Sonic anemometer (right) and cup anemometer and (including a cup and a vane anemometer), an

vane (left) on top of the west mach on the NWC roof

5

R.M. Young 8100 three-component sonic anemometer,

along

with

other

sensors

designed to collect radiation, temperature, and humidity data. Each mast was also equipped with a Campbell Scientific CR1000 data logger to which both wind-measurement sensors were connected. The cup anemometer in the wind sentry set had a range of 0-50ms-1, Fig. 4: Looking northeast at the east mast. Notice the sodar on the left of the image and all of the small poles scattered around the rooftop

the ability to measure up to 60ms-1 wind gusts, and an accuracy of ±.5ms-1. The wind vane of the wind sentry set could specify wind

direction with an accuracy of ±5° (Klein et al 2007). The two instruments used on the NWC roof masts can be seen side by side as constructed on the masts in Fig. 3. As can be in Fig. 2, the west mast was shielded on its north and west sides by building structure up to a height of 3 stories. There was a smaller structure of about one story in height to the southeast of the east mast, as well as large sodar approximately 8 feet in height to the northwest of the east mast which can be seen in Fig. 4. The most important property of the NWC building design in the scope of this experiment was the very irregular shape of the north side of the building as seen in Fig. A1. Additional images of the national weather center along with additional views as seen from the roof of the NWC can be found in Appendix A Figs. A1-A5. Also, in addition to gathering raw data from the Mesonet and the NWC, radar images were also compiled for the entire week for which the raw data was collected. These radar images were used in this experiment to investigate the possibility that not only the building design and placement of the stations on top of the NWC roof affected the wind data recorded, but that passing or nearby thunderstorms and the wind gusts that they created may also have had an effect on the wind data. These radar data were obtained from the National Oceanographic and Atmospheric Administration (NOAA) and National Climatic Data Center’s (NCDC) NEXRAD National Mosaic Reflectivity Images website.

6

Having obtained all of the data from the sources described above, it was necessary to do many steps of calculations and other data analysis methods in order to get all data in to the desired format for this experiment. The data from the Mesonet was in 5 minute data averages of wind direction in degrees and wind magnitude. This was the format that was most useful for the purposes of this experiment, thus no direct manipulation of the Mesonet data was needed. The data from the sonic anemometers of the NWC roof masts, however, were given in u (eastwest) and v (north-south) components. Thus, it was necessary to convert the components to degree and magnitude format. To do this, the following equations were used: (1) M = u 2 + v 2  (2) α = 90 −

360 v arctan  + α  2π u

where M is the wind speed and α is the wind direction. All data from both sites were entered in to a Microsoft Excel® spread sheet, and Eqs. 1 and 2 were applied to all sonic anemometer data from the east and west masts on the NWC roof. Once this was done, all manipulation of sonic anemometer data was completed and it was in 5-minute averages of degree and magnitude format. Although the data from the cup anemometers on the NWC roof were given in degree and magnitude format, they were given in 1-minute averages instead of 5-minute averages. Thus, the data had to be manipulated in order for it to be in a usable 5-minute average format. Having all cup anemometer data again entered into Microsoft Excel® spreadsheets, the following equation was applied to the wind speed data only: (3) AVERAGE(OFFSET($B$2, (ROWS($A$1 : A1) - 1) * 5,0,5)) where $B$2 is the top most cell containing wind speed data. The rest of Eq. 3 was edited in no way an $A$1 is purely a reference cell from which no data is drawn for the function of the equation. This gave 5-minute averages of wind speed. To obtain wind direction data from the cup anemometers, the data had to be converted to u-v component form. This was necessary because taking the average of degree measurements can be dangerous and incorrect in a few select occasions (for example the average of 0° and 360° would be computed as 180° instead of the 0°

7

or 360° as it should be). Thus, the following equations were applied to all cup anemometer wind direction data: (4) u = − M sin(α ) (5) v = − M cos(α ) where M is again the wind speed magnitude and α is again the wind direction in degrees. Once these components were computed, Eq. 3 was used to create 5-minute wind component averages. and Eq. 2 was applied to these averages to find the wind direction for each 5-minute average. Thus, all direct manipulations of data for this experiment were complete.

3. Results and Discussion A graph of wind speed over the entire two

Wind Direction (degrees)

350

300

week period for all sensors can be seen in

250

Fig 5., where Mesonet data markers are seen

200

in red and all other marker colors are data

150

from the NWC roof masts. For greater detail

100 Sonic East Sonic West Cup East Cup West Mesonet

50

0 0

2000

4000

6000

8000

10000

12000

Time (min)

Fig. 5: Graph of wind speed over the entire one week period. Mesonet wind speeds are shown in red.

of this graph, refer to Appendix B Fig. B1. A graph of wind direction over the entire two week period for all sensors can be seen in Fig. 6 where Mesonet data markers are red and all other marker colors represent data

from the NWC roof masts. Again, for a larger version of this graph, refer to Appendix B Fig. B2. When

examining

all

of

the

data

10.0

presented in Fig. 5 and 6, some general

8.0

conclusions can be made. It can be seen series inconsistencies in wind degree measurements atop the NWC occur when the wind, as recorded at the Mesonet, was most directly out of the

7.0

Windspeed (m/s)

that the greatest inaccuracies and time

Sonic East Sonic West Cup East Cup West Mesonet

9.0

6.0

5.0

4.0

3.0

2.0

1.0

0.0 0

2000

4000

6000 Time (min)

8000

10000

Fig. 6: Graph of wind speed for the entire one week period. Mesonet wind speeds are shown in red 8

12000

north, as can be seen in Fig. 5 between. The wind direction data at these times was very sporadic for the instruments on the east mast, while they were fairly consistent at due south (180°) for the west (shielded) mast. At most other wind directions, the NWC measured wind directions were fairly consistent with the Mesonet wind directions. Some outliers can be seen, however, in the east NWC mast instrument data at varying, seemingly random points throughout the week of data. In terms of general trends in wind speed data, mesonet wind speeds were generally significantly higher than those measured at the NWC. Before this experiment was performed, it was thought by the author that greater wind speeds would be measured atop the NWC than at the Mesonet, but this experiment proved that hypothesis incorrect. This trend can be seen most obviously when Mesonet recorded wind directions were most directly out of the north, as exemplified in the time between ~1000-2500s, as can be seen when comparing Fig. 6 to Fig. 5. This trend appears to be least evident when the Mesonet recorded winds were both low speed and out of an approximate easterly direction, as can be seen in the time between ~2500-4000s. It also appears, in analysis of Fig. 6 that, if any NWC instrument measures a wind speed anywhere close to that measured by the Mesonet, it was one of the cup anemometers, and most often the cup anemometer mounted on the east mast. After analyzing these general trends over the entire week-long period of data acquired, the author found the time period between ~1000-3000s to be the most interesting, thus the rest of this analysis will focus on the analysis and explanation of the data during this period. Graphs focusing on the wind directions and wind speeds of this specific period can be seen in Appendix C Figs. C1 and C2, respectively. In addition to these graphs, graphs relating the wind speed deviation of the various NWC instruments placed individually along side Mesonet recorded wind direction can be seen in Appendix C Figs. C3-C6. A new, more specific trend in regards to wind speeds measured at the NWC can be seen when analyzing wind speeds seen in Fig. C2 in relation to corresponding wind directions seen in Fig. C1. When wind directions were most directly out of the north during this period in conjunction with very high wind speeds, the wind speeds measured by the NWC masts were nearly zero for

9

all instruments, though the wind speeds as measured by the east instruments were slightly higher. This trend can be explained for the west mast by noting again the large building structure directly to the north of the west mast. Because the building structure to the north of the mast (as seen in Fig. 2) is so much greater in size than the mast and is so close to the mast, it blocks all winds that place the structure upwind of the mast, namely winds from the north. This trend would also be seen if winds were out of the west or northwest, but, because prevailing winds in Oklahoma are out of the south and the east, west or northwest winds did not occur for any significant length of time during this smaller period or over the entire week of data. Obstructions can also account for the very low wind speeds as measured at the east mast. The sodar to the northwest (which can be seen in Fig. 4) of the east mast is on about the same scale in height as the mast and much larger in horizontal size, thus it is likely that this obstruction had a huge effect on the wind speeds measured during this specific time. A lot can also be said for both masts in terms of general setting apart from the building itself as compared to that of the Mesonet site. As can be seen in Fig. 1, and as required by the WMO for proper meteorological observations, the Mesonet site is in a wide open field with very little obstructions in any direction to hinder winds. The NWC masts, on the other hand, are surrounded not only by large obstructions, but by small ones as well, such as the 2 foot poles that are placed all over the roof of the NWC as seen in Fig 4. Also, the fact that the NWC roof masts are atop a 6 story building in the middle of a metropolitan area says a lot for the amount of difference in the setting between the Mesonet and the roof of the NWC and the effect that that setting will have on the wind speed measurements from the NWC Another very interesting trend can be seen in the wind directions recorded by the west mast during the times in which the Mesonet wind directions were most directly out of the north. During this period, as can be seen in Fig. C1, the wind directions recorded by the west mast are fairly consistent at 180°, or due south. This most likely occurred because of some specific interaction of the strong north wind with the structure to the north of the mast, causing the winds to curve around the building at a wide angle and be returned to the mast as a very light south wind. Perhaps the most interesting result in this experiment came from the wind direction data of the NWC east mast during pronounced north winds. As can be seen in Fig. C1, during the time

10

between ~1200-2500s, Mesonet wind directions are out of the due north, while wind directions measured by the east mast are totally erratic and random. Unlike the west mast that shows a consistent but incorrect wind direction, the wind direction measurements of the east mast have no apparent pattern besides chaos. This is largely due to the structural design of the north face of the NWC. The north face of the building, as can be seen in Fig. A4 is by far the most irregularly shaped face of the NWC building (consider the shape of the north face in Fig. A4 compared to the southeast face in Fig. A5). These irregular jutting-outs and curves create many random microscale eddies that make their way up the building and on to the roof. These eddies create erratic and unpredictable wind directions on top of the roof which can be seen by the chickenpock pattern of wind direction data obtained from the east mast. During this shortened measurement period, which corresponded to the morning of September 25 to Early in the morning of September 27, a cold front

passed

(mid

afternoon

on

September 25) and brought with it storms and certainly increased wind during and for a certain period after passage.

Although

thunderstorms Fig, 7: Radar of the Southern Plains area for 2:00am CST (8Z) on September 26, 2007. Notice the storms over central Oklahoma.

did

a pass

line

of

around

10:00am CST on September 25, there is not evidence for this in the

data. In the wake of that storm system, however, more isolated storms formed at 2:00amCST on September 26, as can be seen in the radar image of Fig. 7. Although the strongest of these storms may not have passed directly over the measurement sites, it is certain that outflow and inflow from these storms is likely to have interfered with measurements made at the two sites studied. These effects can be seen at approximately 1500s. Here, a small spike in measured wind speeds of all instruments as well as a short-lived change in wind direction from due north to more north east. This change in wind direction caused an odd looking pattern in the data from the west mast. Where as before, it was recording a steady southerly wind, at the time when storms are in the

11

vicinity, there is a short downward spike in the wind direction down to a more accurate reading of north east. This is the result of wind blowing in at an angle such that it was picked up correctly by the mast and no longer blocked by the structure to its north.

Summary and Conclusions The purpose of this experiment is to discover the differences that arise essentially from good and poor site selection as well as to discover why these differences arise. The “poor” site is two masts atop the roof of a 6-story building in an urban setting while the “good” site is a WMOstandard Oklahoma Mesonet location in the local vicinity (within 10 miles) of the “poor” site. This experiment found that large differences arose between the sites, especially when the winds were out of the north. This was due to one site on the roof being shielded by a large 3 story structure and the second mast was exposed to the strong eddied created by the odd shape of the north face of the building upon which it sat. This experiment also discovered that storm passages or storms in the vicinity of a measurement station can affect the measurements recorded at said sites and create very interesting patterns in the observed data that might otherwise be perplexing if the observer had no knowledge of the storms. It would be useful to do further similar experiments perhaps during late April and May to observe the effects of extremely strong thunderstorms and supercells on the measurements made at these two different locations.

5. References Fujita, T. Theodor and R.M. Wakimoto, 1981: Effects of Miso- and Mesoscale Obstructions on PAM Winds Obtained during Project NIMROD. Appl. Meteor., 21, 840-858. WMO, 1989: Guide on the Global Observing System. Wmo 488, Geneva, Switzerland. Available online at http://khmi.nl/~laatdej/TMP/WMO488.pdf Mcpherson, Renee A., et al., 2006: Statewide Monitoring of the Mesoscale Environment: A Technical Update on the Oklahoma Mesonet. J. Atmos. Oceanic Technol., 24, 310-321. Klein, P.K. et al, 2007: Wind Measurements Using Anemometry: Long Term Project II.

12

6. Appendices A. Images of the NWC Site

Fig. A1: View of the west mast looking northeast

Fig. A2: Terrain looking south from the roof of the NWC

13

Fig. A3: NWC roof masts looking east. Notice the small one-story building to the east of the east mast

Fig. A4: View from the ground of the north face of the NWC

Fig. A5: Computer simulated model of the southwest face of the NWC

14

B. Enlarged Full Graphs Over the Entire One-week Period of Data Collection

350

300

250

Wind Direction (degrees)

200

150

100 Sonic East Sonic West Cup East Cup West Mesonet

50

0 0

2000

4000

6000

8000

10000

Time (min) Fig. B1: Large scale image of the wind direction data over the entire one-week collection period

15

12000

10.0 Sonic East Sonic West Cup East Cup West Mesonet

9.0

8.0

7.0

Windspeed (m/s)

6.0

5.0

4.0

3.0

2.0

1.0

0.0 0

2000

4000

6000

8000

Time (min) Fig. B2: Graph of wind speed data for the entire one-week observation period

16

10000

12000

C. Graphs For More Focused Time Period Between 0-3000s 350

Sonic East Sonic West Cup East Cup West Mesonet

300

Wind Direction (degrees)

250

200

150

100

50

0 0

500

1000

1500

2000

2500

3000

Time (min)

Fig. C1: Wind Direction data from the time period between 0-3000s 7.0 Sonic East Wind Speed Sonic West Wind Speed Cup East Wind Speed Cup West Wind Speed Mesonet Wind Speed

6.0

Wind Speed (m/s)

5.0

4.0

3.0

2.0

1.0

0.0 0

500

1000

1500 Time (s)

Fig. C2: Wind speed data for the time period between 0-3000s

17

2000

2500

3000

6.0

400 Sonic East Deviation Mesonet Wind Direction 350

5.0

300

Wind Deviation (m/s)

250 3.0 200 2.0 150

Wind Direction (Degrees)

4.0

1.0 100 Time (s)

0.0 0

500

1000

1500

2000

2500

50 3000

-1.0

0

Fig. C3: The deviation in the wind speed measurements made by the sonic anemometer of the east mast from the wind speed measurements of the Mesonet over the time period from 0-3000s. Wind direction is graphed against the right Y-axis and in pink.

6.0

400 Sonic West Deviation

5.0

350

4.0

300

3.0

250

2.0

200

1.0

150

0.0 0

500

1000

1500 Time (s)

2000

2500

100 3000

-1.0

50

-2.0

0

Fig. C4: The deviation in the wind speed measurements made by the sonic anemometer of the west mast from the wind speed measurements of the Mesonet over the time period from 0-3000s. Wind direction is graphed against the right Y-axis and in pink.

18

Wind Direction (Degrees)

Wind Speed Deviation (m/s)

Mesonet Wind Direction

6.0

400

Wind Speed Deviation (m/s)

Cup East Deviation Mesonet Wind Direction 5.0

350

4.0

300

3.0

250

2.0

200

1.0

150

0.0 0

500

1000

1500 Time (s)

2000

2500

100 3000

-1.0

50

-2.0

0

Fig. C5: The deviation in the wind speed measurements made by the cup anemometer of the east mast from the wind speed measurements of the Mesonet over the time period from 0-3000s. Wind direction is graphed against the right Y-axis and in pink.

6.0

400

Wind Speed Deviation (m/s)

Cup West Deviation Mesonet Wind Direction 5.0

350

4.0

300

3.0

250

2.0

200

1.0

150

0.0 0

500

1000

1500

2000

2500

-1.0

100 3000 50

-2.0

0 Time (s)

Fig. C6: The deviation in the wind speed measurements made by the sonic anemometer of the west mast from the wind speed measurements of the Mesonet over the time period from 0-3000s. Wind direction is graphed against the right Y-axis and in pink.

19