s r

ir • • - • / !

. • • . • • : • /

- APRL/95/1

INTERNATIONAL CENTRE FORTHEORETICAL PHYSICS

ATMOSPHERIC PHYSICSAND

RADIOPROPAGATION LABORATORY

INTERNATIONALATOMIC ENERGY

AGENCY

UNITED NATIONSEDUCATIONAL,

SCIENTIFICAND CULTURALORGANIZATION

RADIOWAVE PROPAGATION MEASUREMENTSIN SENEGAL

F. Postogna

I.H. Sarpun

S.M. Radicella

and

K.A. Hughes

MIRAMARE-TRIESTE

ii£,a:;r:::':":::: •

APRL/95/1

International Atomic Energy Agencyand

United Nations Educational Scientific and Cultural Organization

INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS

ATMOSPHERIC PHYSICSAND

RADIOPROPAGATION LABORATORY

RADIOWAVE PROPAGATION MEASUREMENTS IN SENEGAL

F, Postogna, I-H. Sarpun', S.M. RadicelSaAtmospheric Physics and Radiopropagation Laboratory,

International Centre for Theoretical Physics, Trieste, Italy

and

K.A. HughesRadiocommunication Bureau, International Telecommunication Union (ITU),

Geneva, Switzerland.

ABSTRACT

This report describes the results of over two years of radiowave propagation measure-ments undertaken in Senegal under the auspices of the ITU. The signal level of a TVtransmitter located in Ziguinchor was received in Bambey (approximately 140 km eastof Dakar). The frequency band was VHF and the path length, 237 km. Statistical anal-ysis has been made of the field strength behaviour, together with a critical comparisonbetween predicted values, using ITU-R recommended methods, and the actual measureddata. The results show that the measured values of field strength were significantly higherthan predicted, indicating that the occurrence of super-refractivity in the region is greaterthan previously thought. The necessity for predictions to take into account seasonal anddiurnal variations is stressed.

The analysis described in this report represents a collaborative study undertakenwithin the framework of the Memorandum of Understanding (1993), established betweenThe International Telecommunication Union (IT)) and The International Centre for The-oretical Physics (ICTP).

MIRAMARE - TRIESTE

July 1995

'Permanent address: Osmangazi University, Eskisehir, Turkey.

PREFACE

The ICTP-APRL reports consist of preprints relevant toresearch and development work done at the AtmosphericPhysics and Radiopropagation Laboratory of theInternational Centre for Theoretical Physics with theparticipation of visiting scientists.

More information can be obtained by contacting:

Prof. Sandro M. RadicellaAtmospheric Physics and Radiopropagation LaboratoryP.O. Box 58634100 TriesteItalyPhone: +39 40 2240331FAX: +39 40 224604E-mail: rsandro@ictp.trieste.it

1. INTRODUCTION

Planning of radio services needs a good knowledge of the propagation

characteristics of the area of interest. Such information is required, not only to ensure that

an adequate service will be provided, but also to help assess the potential for interference

between stations working in the same band or, in a worse case, stations sharing the same

frequency. To help engineers in the planning of the broadcast services, ITU has developed

propagation prediction methods that usually give a representation of field-strength values

exceeded at 50% of the locations for different percentages of time and path lengths [2,3].

The set of data upon which these propagation prediction methods are based usually

comes from cumulative frequency distributions of field-strength measurements considered

on a yearly basis. In the main, these measurements have been made in temperate regions of

the world. As a consequence, they are not very representative of areas subject to

anomalous propagation effects such as extreme super-refractivity and ducting which are

often prevalent in low latitude tropical areas. The aim of the current experiment, therefore,

was to acquire data from such a tropical region in Africa and, in turn, to obtain

corresponding statistics of field strength. The results indicate that, not only do the current

prediction methods generally under-estimate the field strength for the propagation path

used in this experiment, but also that yearly statistics are too broad, and that seasonal and

diurnal variations must be taken into account for a realistic prediction.

2. EXPERIMENT

2.1. GENERAL

Senegal is located in western Africa, bordering the North Atlantic Ocean between

Guinea-Bissau and Mauritania. The total area is 196,190 km1. Senegal's climate is tropical

i.e. hoi and humid; there are two main seasons: rainy season (approximately April to

October) with strong south-east winds; dry season (approximately November to March)

dominated by hot, dry harmaltan winds. The terrain is generally low rolling plains rising to

foothills in south-east [1].

2.2. VHF PROPAGATION MEASUREMENT EXPERIMENT

The aim of the experiment was to carry out continuous measurements of the field

strength resulting from a TV transmitter in the VHF band over a path length of about 240

km. Table 1 gives details of the experiment.

"•*"•« t ZE j j ^E J5. .

DETAILS OFSENEGAL EXPERIMENT

FrequencyTx ant. height

Rx ant. heightTx e.r.p.

Path distance

215 MHz

200 m

10m

10 kW

237 km

Table I

The installation of the experiment was undertaken in November 1992 by the

Deutsche Bundespost, under the guidance of the BDT and BR of the ITU. Local support

was provided by engineers from RTS (Radiodiffusion Television du Se'ne'gal) and from

SONATEL, and it was this team of engineers who continued to monitor the experiment

and to collect the data throughout the duration of (he experiment. Subsequent detailed

analysis of the data was undertaken by ICTP.

The field-strength equipment was provided by Rohde & Schwarz. The receiver

(ESV) was installed at the SONATEL building at Bambey (located about 140 km east of

Dakar), with the antenna (yagi) placed 10 m above local ground level. The transmitter was

a TV station at Ziguinchor (in the south of Senegal) with a quoted e.r.p. of 10 kW. Figure

I indicates the positions of the transmitter and receiver. Although a detailed terrain profile

has not been acquired, the propagation path can be considered as essentially flat, with no

significant terrain obstacles.

The experiment was initiated in December 1992 and was expected to have a

duration of at least two years. Unfortunately a break-down of the receiver happened in

October 1993 and only after five months the experiment started again. A failure occurred

also between September and November 1994, Table 2 indicates the number of days for

each month that field strength data were obtained.

2.3 DATA PROCESSING

The total database of measurements obtained from the experiment amounted to

some 27 MB. In order to handle a database of such dimensions, purpose-written software

was developed by ICTP for both data reduction and subsequent analysis. The main

analysis program allowed the user to view the measurement data and to calculate

cumulative distributions over selected time periods. Further programs were also produced

for more detailed statistical analyses and comparison purposes, supported by proprietary

software. To produce the results presented in this report, over 700 data files were

necessary.

Figure I

Map of Senegal showing locations

of Tx (Ziguinchor) and Rx {Bambey)

Number of days for which field strength data were obtained

January

February

March

April

May

JuneJuly

August

September

October

November

December

1992

-

-

_

_

-

23

1993

31

28

31

30

31

30

31

16

2

7

-

-

1994

-

7

30

31

22

31

31

6-

9

31

1995

31

28

31

30

23_

_

_

_

-

-

Table 2

Number of days per month for which field strength data were obtained

3. DATA ANALYSIS

3.1. FIELD STRENGTH CALCULATION AND DATA RECORDING

PROGRAMME

The measured signal level received in Bambey was recorded on floppy discs as

values of voltage, and these were subsequently converted to values of field strength for

statistical analysis on a PC. Great care was taken in the conversion process to ensure that

correct values of field strength were calculated.

Using information provided by RTS on the transmitted power from Ziguinchor, the

free-space field strength at Bambey was estimated to be 69.5 dBfiV/m. Taking into

account the possibility of smaJl variations in the transmitted power from time to time, and

also in the characteristics of the receiving equipment, it is estimated that the overall error

in the field strength values is no greater than + 2 dB and that comparisons with

predictions will be accurate to about ±4 dB.

According to RTS, the normal schedule of broadcasts from Ziguinchor was:

Monday to Friday, approximately 1600 to 2400 hours,

weekends and national holidays, approximately 1200 to 2400 hours.

Some variation in this schedule was apparent from time to time.

For analysis purposes, it was necessary to select measurement data only from those

periods corresponding to times of the broadcast transmission. Since the transmission time

varied slightly from day-to-day, a simple means of selecting (he required data was to set a

lower limit to the field strength values below which the data would be ignored in the

statistical analysis. This limit, 7 dBu.V/m, corresponded to a value just above the "noise

floor". Since the signal from the transmitter could be received in Bambey for most of the

time, very few "real" data have been lost from the analysis by using this cut-off value.

3.2 CUMULATIVE DISTRIBUTION FOR THE ENTIRE DATABASE

Figure 2 shows a cumulative frequency distribution for the entire database of

measurements in accordance with the data listed in Table 2. The figure shows the value of

field strength (dB(iV/m) exceeded as a function of the percentage of measurement time.

As stated earlier, the measurement time corresponded.to the evening hours (approximately

1600 - 2400 h) on week-days and to afternoon and evening hours (approximately 1200 -

2400 h) at weekends.

i

10.00

Percentage of time100.00

Figure 2

Cumulative frequency distribution for (he entire database of measurements

It is interesting to compare (he values for 50, 10 and I % of the time with (hose

given by established prediction methods. One well-know method is that of

Recommendation ITU-R PN.37O [2] which contains families of Held strength curves, for

the VHF and UHF bands, as a function of distance for various antenna heights and

percentages of time. Propagation curves are also contained in the Final Acts of the

Regional Administration Conference for the planning of VHF/UHF TV broadcasting for

Africa (GE89) [3]. For the region of west Africa containing Senegal, the appropriate

propagation curves from the two prediction methods are identical and correspond to those

for a warm maritime region. Table 3 shows a comparison between the measured data and

prediclions using (hese two prediction methods. (N.B. The value of the free-space field

strength for the path is estimated as 69.5 dB(lV/m.)

Senegal measurements

(Dec.!992

- May 1995)

Recommendation 1TU-

R PN.37O

(orGE89)

50% lime

25.5

13.0

10% lime

55.5

27.5

I % time

67

44.0

Table 3

Comparison of predicted and measured field strengths (dB|iV/m)

The results indicate that the predicted values are seriously under-estimated. The high

measured values are a clear indication of the frequent occurrence of super-refractive

behaviour in the region of Senegal,

3.3 MONTHLY CUMULATIVE FREQUENCY DISTRIBUTIONS AND

COMPARISONS WITH PREDICTIONS

For a more detailed investigation of the measured signal levels, cumulative

distributions have been derived for each month for which more than fifteen days of data

were obtained. These are shown in Figure 3. Two aspects are immediately apparent.

Firstly, when comparing the values with those predicted for time percentages of 50, 10

and t (using Rec. ITU-R PN.370 [2], or GE89 [3]), it is clear that in many rases the

predictions are seriously underestimated. Moreover, there are many examples where the

value at I % approaches, or even tends to exceed, the free-space value of 69.5 dBfiV/m.

The second observation is that considerable variation is present from one month to

another. Again, comparison with predictions also shows considerable variation, with some

months giving reasonable agreement with prediction (e.g. January 1995) and others, very

poor agreement (e.g. April 1993). This strong monthly, or seasonal, variation is indicative

of the varying degree of super-refractivity from one month to another which, in turn,

reflects the meteorological behaviour prevailing in the region.

Inspection of Figure 3 shows that the highest values tend to occur from about March

to June with the lowest values from about July to September. Whilst this does not

correspond precisely with the well-defined seasons, this result is in broad agreement with

that obtained from a similar experiment in Burkina Faso [4], undertaken from 1986 -

1989, insofar as the highest values were recorded during the rainy season. It is difficult to

be more precise about any such relationship without a detailed study of the prevailing

meteorology, and this is to be the subject of a further study. However, the association

between the high field strength values and the rainy season probably implies a strong

influence of the humidity component of the refractivity - an observation that has been

made elsewhere [41-

It should be noted that prediction methods such as that in Rec. ITU-R PN.370 can

make some allowance for the degree of terrain irregularity along the propagation path of

interest. The basic propagation curves given in these prediction methods are appropriate

for "gently rolling terrain" which would correspond to terrain exhibiting a greater degree

of irregularity than would be present on the path in this study. Corrections can be applied

to the predictions according lo the actual amount of terrain irregularity on the path of

interest and, for the present case, i( is estimated that such a correction would increase the

predicted field strength by some 3 dB, thus reducing the general discrepancy between the

predictions and measurements. Nevertheless, the underestimation in predicted field

strength is still obvious in most cases.

Dec-92

1 10Percentage of time

Fig. 3 (a)

Feb-93

100

1 10Percentage of time

Fig. 3 (c)

100

Jan-93

-——.

-

1Si

807060

.. I . .1

,E so • n+v.3 40g 3 0

20100

... ' i '

11.

- J- L

;'t

. 1

- ---

1 1

8070

'• 60: 1 50j = 40

?3020

^ 10U 0

10 100Percentage of time

Fig. 3 (b)

Mar-93

r T TI

•:.[:i 1 -

"IK

i-j—

i

^--

1 10 100Percentage of time

Fig. 3(d)

Apr-93

10 100Percentage of time

Fig. 3 (e)

Jun-93

1 10 100

Percentage of time

Fig. 3 (g)

Aug-93

1 10 100Percentage of time

May-93

....

|-"

....

^. |

r\\\-

\ s

10

Percentage of time

Fig. 3 (f)

Jul-93

100

0 -10

Percentage of time

Fig. 3 (h)

Apr-94

I

Sjs1

10

Percentage of time100

Fig. 3 (i) Fig- 3 (j)

May-94 Jun-94

1 10 100

Percentage of time

Fig. 3{k)

Jul-94

0.001.00 10.00 100.00

Percentage of time

Fig. 3 (m)

Dec-94

10

Percentage of time

Fig. 3 (o)

8070-60 |5040 "i3020100 -

10 100

Percentage of ttme

Fig. 3 (I)

Aug-94

10 100

Percentage of time

10 100

Percentage of time

Fig. 3(p)

10

- «LJ»'Jtiit.. IB. J

Feb-95 Mar-95

10 100

Percentage of time

Fig. 3(q)

Apr-95

10 100

Percentage of time

Fig. 3 (r)

May-95

m1 10 100

Percentage of time

Fig. 3 (s)

10 100

Percentage of time

Fig. 3 (t)

3.4 MONTHLY VARIATION DURING THE EVENING HOURS

To investigate further the monthly variation of the received field strength,

cumulative distributions on a monthly basis were calculated for the time period between

2000 and 2300 hours. We choose this time window because it corresponds to the hours of

the largest TV audience and therefore to when the impact of interference phenomena

would be the greatest. The following plot in Figure 4 shows the monthly variability of the

field strength at three percentages of time, over a nine month period.

From the figure it is also clear that the field-strength level for 50% of time, predicted

from Rec. ITU-R PN.37O [2] (or from GE89 [3]), is too low and sometimes the difference

between the predicted level and the one obtained from measured values is greater than 30

dBuV/m.

70

«)

Ml

S 40>

§ 10

21)

10

0r

c

k

1 ri

\ i \U-

/ " • " * • " - —

5 s ?'

Months

\

\

Jun-

93

r

i =*

50*oltimc

— — 10%ofomc

" - • - 70*0111™

Figure 4

Monthly varialion of field strength at three time percentages

In addition, we have taken three representalive months in the period from December

1992 and August 1993 (January, April, and July) to underline the strong seasonal varialion

observed during the experiment. The corresponding cumulative distributions are given

below in Figures 5 to 7.

Monthly Cum. Dist., In Jan 93 (2000 to 2300 h)

20.00

10.00

0.00

-----

s

!f

4

...

N

-

s

!

1.00 10.00

Percentage of time

100.00

Figure 5

12

70 00

60.00

SO.OO

| 40.00aM 30.00

20.00

10 00

0.001.<M

Monthly Cum

\ -

J

• • —

. Dist.

I

Parc

-

:e

in Apr 93 (2000 to 2300 h)

1

it

-

0 00

je of time

,

ss\

1C

I])0

.00

Figure 6

70,00

60.00

50.00 •

| 40.00

g 30.00

20.00

10.00

1.

Mont

30

hly Cum. I

i

—

l ist. ,

I

i

n Jul 93 (20DO t o :

1 „

300

•sS\

10.00 100.00

Percentage of time

Figure 7

3.5 HOURLY VARIATION OF FIELD STRENGTH

To investigate the diurnal varialion of the field strength, we have calculated the

cumulative frequency dislribution on a time window of 60 minutes centred on the hour

13

from 1200 to 2300 during every month for which data were available. Then we extracted

the percentage of time at every hour corresponding to ihe following field strength levels -

greater than 10, 20. 30, 40, 50 and 60 dBuV/m.

To highlight the year-to-year variation, we compared the diurnal cumulative

distributions in ihe same month, for different years, Figures 8 to 10 show the results for

the three representative months chosen before.

Compiri ton of Jin-93 and Jan-BS

12 13 14 15 16 17 18 19 20 21 21 23

- 9 3 . >10HBuV/m

- 93, 120 dBiiWm

- 93. >3C dBuVfm

- 93. >40 OBuV/m

-93 ,^60 dBuV'm

- 9 3 . >6O dBuWm

• 95. >10dBuVJm

95, >2O dSuWm

• 95, »30 dBuVJm

• 95. >40 dBuV/m

Figure 8

Compiflaon of Apr-fl3 and Apr-94

° 93

— — 9 3

—<— 93

• loaBuvm

>20 OBuV/m

>30dBuV;m

—° 93. *40 dBuV/m

0 93

• - • • > • • 9 4

. - . « • - . 94

• • • ' • • - 9 4

. . . o - . . g^

. . . . . . 94

•• — • 94

^ 0 dSuV/m

>l0dBuV/m

>20 dBuV/m

>30 dBuV/m

>50 dBuVfm

>M OBuV/m

Figure 9

12 13 14 15 16 17 18 19 20 21 22 23

93,>IOdBijWm

93, >20 dBuVlcn

93, >30 dBuV/m

93. J O dBuV/m

93, >S0 dBuV/m

• " - • • 94. »20aBuV/m

- o - . . 94 ,40 dBLjVym

Figure 10

From the cumulative distribution data, it is also possible to extract values that could

help in defining the seasonal and year-to-year variations. These are shown in Table 4.

Seasonal

Variation

Jan-93

Apr-93

Jul-93

Jan-95

Apr-94

Jul-94

50% of time at

1900 h

>19dBnV/m

> 38 dBiiV/m

> 24 dBMV/m

>15.5dBMV/m

> 32 dBgV/m

> 23.5 dB^V/m

50% of time at

2100 h

> 23 dBuV/m

> 54 dBgV/m

> 26 dBpV/m

>15dBnV/m

> 48 dBpV/m

> 26 dBgV/m

50% of time at

2300 h

> 26 dBMV/m

> 60 dBpV/m

> 27 dBpV/m

>21.5dBgV/m

> 60 dBMV/m

> 29 dBMV/m

Table 4

Seasonal and year-to-year variation expressed numerically

This table shows numerically that the variations can be as large as 38.5 dB(jV/m at

2300 h and as low as 22.5 dBuV/m at 1900 h. This result, as with those given earlier,

indicates that a cumulative distribution on a yearly basis provides too limited information

for a good prediction of the field strength behaviour.

Interesting information can be extracted from the diurnal variation of the signal

levels. The maximum interference can be expected always in the late hours of the period

1600 - 2300 h, the diurnal variation being less distinct in July.

15

3.6 WORST-MONTH STATISTICS

Using predictions for annual statistics, Recommendation ITU-R PN.841 [5] gives

an expression to convert annual to the worst-month statistics :

where p w is the average worst-month time percentage of excess;

p is the average annual time percentage of excess;

Q is the conversion factor.

This allows us to make an additional comparison. From Rec. ITU-R PN.841, let us

assume that Q = 2.5 and then the expression above gives, for the worst-month, p w = 25%

if the annual time percentage of excess is 10%. Figure 11 shows a plot of the values of

field strength exceeded for 25 % of each month over a nine month period; also shown, is

the value of field strength exceeded for 10 % over the same period. The figure clearly

identifies April 1993 as the worst-month and confirms that a value of 2.5 for Q is very

appropriate for the data sample selected.

60

50

40

m

20

10

0

I

/

/

i k 1 i

\

• •• 1 - - t t —

* 1 5Month

• • • 1 1

^ i

25 % momhly value

to % value for the 9 monthpa lied

Figure 11

Worst-month over a nine month period

3.7 REPRESENTATIVE DAY

In order to get a better knowledge of the field strength behaviour, we derived from

the whule data base a representative day for each monlh. Firsl we calculated the median of

16

the field strength over a lime window of 60 minutes centred at each hour and then

computed the monthly median, upper quartile, and tower quartile for each hour.

Figures 12 to 14 show the median, upper and lower quartile variation of field

strength for each representative day for the three representative months chosen earlier.

The results confirm the behaviour described above of the diurnal and seasonal variability

of the potentially interfering signal field strength during evening hours.

12•o

7.00E+01 j-

6.00E+01 |

5.00E+01 }

4.00E+01 :

3.00E+01

2 OOF+fll

l.OOE+OI

0.00E+00 _ .

1700

1 - -

1800

Jan-93 Standard Day

, _ . .._

1900 2000 2100

Hours

. . - • • "

— — -

1

2200

,———

,

2300

Figure 12

7.00E+OI

6.00E+01

5.O0E+0I

« 3.00E+0!

2.00E+0!

1.00E+0I

I70O

Apr-93 Standard Day

, . , - - - ' . "

. . . \ ^ ~ ^ * ~ ^ ^' " _ — . . . • '

1800 1900 2000 2100 2200 2300

Hours

Figure 13

17

Jul-93 Standard Day

3.5OE+U I

3.00E+01

2.5(1E+OI i

| 2.00E+01

* I.50E+0I !

l.OOE+OI Ii

5 Wlli+OO i

o.ooE+nu i

I7(X) 1800 IWX) 2000

Hours

2!00 2200 2300

Figure 14

4. CONCLUSIONSThe data obtained from the experiment clearly indicate that higher values of field

strength were received than would have been predicted by well-known, established

prediction methods. This result could have serious consequences for the planning of VHF

broadcasting in the area since it is likely that the incidence of interfering signals, associated

with small time percentages typically from 10 to 1 %, will be underestimated by using the

existing prediction methods. These high values of field strength, which approach the free-

space level on many occasions, (and sometimes even exceed il), are due to the extreme

refractivity gradients which are frequently present in the region of west Africa,

From a critical analysis of the data and the graphs shown in this report, it is clear

that the annual average field-strength level predicted by the established prediction methods

is too limited. Seasonal variations, as well as diurnal variations, are not usually taken into

account in such prediction methods although the results of this study indicate that these

variations are, however, very important. For future planning of services, however,

especially in regions prone to significant seasonal variation, such variations should be

taken into account in order that prediction methods give realistic estimates of the signal

level to be obtained. This means that more accurate information, by season and hour, must

be acquired from experiments such as the one described here.

18

5. ACKNOWLEDGEMENT

The authors wish to thank Rohde & Schwarz and Deutsche Bundespost TELEKOM

for the equipment and for the help with its installation. The collaboration of SONATEL

and RTS (Senegal) is gratefully acknowledged for looking after the experiment and in

collecting the data. Finally, thanks are due to the BDT and BR of ITU for coordinating the

activities relating to this work.

I.H.S. has been partially supported under the agreement between ICTP and the

Centre for Turkish-Balkan Physics Research and Applications.

6. REFERENCES

[ ] ] CIA Factbook from http://www.ic.gov

[2] Recommendation 1TU-R PN.37O; VHF and UHF propagation curves for the

frequency range from 30 MHz to 1000 MHz

[3] GE89 - Final Acts of the Regional Administrative Conference for Planning of

VHF/UHF Television Broadcasting in the African Broadcasting Area and

Neighbouring Countries; Geneva, 1989.

[4] Radio-wave propagation measurements in Burkina Faso; K. Low and Z.

Bonkoungou; Telecommunication Journal, Vol. 57, XI, 1990,

[5] Recommendation [TU-R PN.841; Conversion of annual statistics to worst-month

statistics.

All the specific software designed to calculate cumulative frequency distributions

and related statistics are available from F. Postogna (e-mail : postogna@ictp.trieste.it).

19