Clear air interference predictions in tropospheric radio wave propagation: Indian experiences

(1)

IndianJournalofRadio^&SpacePhysics Vol.25,December1996,pp 328-335

Clear air interference predictions in tropospheric radio wave propagation:

Indian experiences

MVS N Prasad&S K Sarkar

RadioScienceDivision,NationalPhysicalLaboratory,NewDelhi110012

Received14March1996; revisedreceived20May1996

Interference predictions are vital components of system design especially in broadcast services for predicting the coverage area of own transmitters as well as for assessing the signal strength of interfering transmitters. The problems associated with lTU-R interference predictions related to tropospheric radio wave ptopagation are discussed. Based on the experiences of analysing data of interfering signals at VHF and above, a new prediction procedure is advocated after taking into consideration all the propagation mechansims in demarcating coordination distances. This approach can be very useful in spectrum management applications for FM and TV transmitters in the VHF and UHF bands, particularly for tropical countries where abnormal propagation conditions prevail for most of the time.

I Introduction

The commercial services operating at frequencies from 30 MHz onwards utilize tropospheric medium for propagation. Most of these services ~ utilize ITU-R recommendations for design which include predicting interference levels from other transmitters and the coverage area of their own transmitters. The increasing usage of the spectrum by various services coupled with the anomalous propagation conditions prevailing over the subcontinent make interference predictions an important component for the design of new systems and also for assessing the reliability of the existing systems.

2 Results and discussion

The present paper deals with the problems en- countered in the prediction of interference levels over the Indian subcontinent using ITU-R recommendations relating to tropospheric radio wave propagation. These have surfaced in the light of our experienGe in analysing some cases of interfering signals. Based on a procedure suggested by Hewitt! for UK, an integrated prediction procedure for tropical countries where extreme propagation conditions prevail is advocated. In other words this paper presents some experimental evid- ence from the tropical countries for the support of the model proposed by Hewitt.

The ITU-R interference procedure2 is a collec- tion of loosely associated reports and the draw- backs of the ITU- R predictions are! :

(1) A comprehensive procedure is not available to predict the field strength of interfering signals for small and higher percentages for a given propagation mechanism.

(2) The results of the worst-month model fell short of the true worst-month results demanded by end users.

(3) The ITU-R method takes surface duct propagation mechanism as the main mode of anomalous propagation. The neglect of elevated layer reflection/ refraction over inland/coastal areas has resulted in underestimates of interference predic-·

tion. This is illustrated as follows.

Our limited experience at the National Physical Laboratory in analysing data of various iuterfering signals at VHF shows that this mechanism occurs for percentrages as high as 10%. The simultaneous analysis of the data of VHF TV signals from neighbouring countries recorded at New Delhi and the meteorological data collected at path midpoint of the interfering signals and neighbouring sites showed that elevated layers occurring at heights of 1 to 2 km with gradients of - 100 to - 150 N /km produced appreciable field strength values and interfering signals were very weak when surface ducting gradients were present at the same sites3•

The monitoring of Jalandhar TV signal (frequency, 208.75 MHz; distance, 360 km) at New Delhi and further path loss calculations showed that during daytime the signal propagates due to scattering and is very weak, and during nighttime the signal

'I

(2)

165.5 160.0 153.0

149.0 125.0

137.0 166.9

168.8 195.5

199.5 195.5

162.2 157.4

152.0 131.0

133.4 121.0

129.0 159.0

154.0 Table I-Predicted and observed values of path loss from different prediction methods

Pondicherry-Kokavil Tirunelveli-Mt. Pidurutalagala (Distance: Sea, 50 km; land, 270 km; (Distance: Sea, 160 km; land, 220 km;

total, 320 km) total, 380 km)

42 205

128.4 128.9

LOS distance for K=4/3 (km) Free space loss (dB)

Observed path loss (dB) 50%

10%

1%

Loss due to scatter (dB) (Tatarskii method)

Loss using Gannaway method (dB) Loss due to diffraction (dB) (ITU-R method) Loss due to reflection (dB)

Loss due to ducting (dB)

ITU-R method for mixed land-sea paths ]%

10%

Fig. I-Change of propagation mechanism observed on Sri Lankan TV signal (ChanneI.5).

predicted path loss for scatter mechanism (166.9 and 168.8 dB), and the path loss values for 10%

of the time (153.0 and 149.0 dB for channels 8 and 5 respectively) agreed with those predicted from reflection mechanism (157.4 and 152.0 dB).

The path loss values deduced for 1% of the time (125.0 and 137.0 dB for channels 8 and 5 respectively) are in close agreement with those predicted from ducting mechanism (131.0 and 133.4 dB).

The ITU-R method6 for mixed land-sea paths predicted values of 121.0 and 129.0 dB for channels 8 and 5 for 1% of the time and 159.0 and 154.0 dB for 10% of time. This method has overestimat- ed the loss by 5 to 6 dB especially for 10% of the time.

An example of how the reflection changes into ducting is shown in Fig. 1 in whjch Sri Lankan Television signal originating from' Mt. Pidurutala- gala on channel 5 was recorded at Triunelveli in South India. Figure 2 shows the post-sunset variation of Jalandhar TV signal. The Jalandhar signal strength increases (10 to 15 dB) due to reflection

from stratified elevatedlayers4•

Similarly the data collected at Triunelveli (trans- miter at Mt. Pidurutalagalal) and Pondicherry (transmitter at Kokavil) in South India from Ru- pavahini transmissions of Sri Lanka on channels 5 (174-181 MHz) and 8 (195-203 MHz) showed that scattering accounts for 50%, reflection from layers for 10%, ducting for 1% oftime5• These are mixed land-se~ paths and super-refractive and ducting gradients prevailed very frequently at higher heights. These signals used to be received up to distances of 700 km and surface duct propagation alone cannot account for this long distance propagation. Surface ducts occur over sea for several hundred kilometres. But over land the occurrence over such large distances is extremely rare. In the present case the sea path is less than the land path. The sea path is around 50 km and rest is land path. Hence the possibility of surface ducts occurring all over 700 km is extremely rare.

Propagation through elevated ducts or' reflection from elevated layers or a combination of different propagation mechanisms, i.e. partly by line-of-sight and partly by scattering etc., could have caused the observed signal levels. Sometimes signals as high as 10 dB below free space levels are observed. The path losses predicted from different prediction methods along with the observed losses are shown in Table 1. In the present study the path loss deduced from the observed signal strength for 50% of the time (165.5 and 160.0 dB for channels 8 and 5 respectively) tallied with the

.do •.•

-

I ^...

^••••

(3)

330 INDIAN JRADIO &SPACE PHYS, DECEMBER 1996

exhibits small amplitude rapid fluctuations of the order of 2-4 dB during daytime and afterwards it starts increasing exhibiting slow and deep fades of the order of 5-10 dB. The transition of atmospheric conditions is clearly seen in this figure where rapid fading changes to slow and deep fading. The fast fading is due to the scattering of radio wave from refractive index irregularities, and the increase in signal strength and formation of slow and deep fades are due to the reflections of signal from elevated layer structures of the atmosphere. These are the changes that are important for interference and communication engineer has to take them into account in system design.

Also the analysis of line-of-sight link data at 2 and 7 GHz over different regions of India showed that the rising layer produces a clear morning transition on LOS signals which is prominent in northern regions. This increases the signal levels

Fig. 2-Post-sunset variation of Jalandhar TV signal.

so

II

2124

-

E

'521II^"'20

~...•••...•II

...•

"

~ .1_0;

14l~

12noo '800

1900 2000 2100 2200

TIME ( hr. )

aoo

by as much as 20 dB and could be a source of interference if systems around this frequency are operated in the zone. An example of this observed on Delhi-Meerut line-of-sight link operating at 2 GHz in northern India is shown in Fig. 3.

(4) Definitions of radiometeorological climates used for troposcatter interference were very diffi- cult to associate with specific paths when they cross a mix of land, sea and coastal areas. This is more problematic in a country like India where different radio climatic zones exist. The north-west part of India experiences intense summer due to the proximity of desert of Rajasthan, whereas the north-eastern hilly regions of India record the highest rainfall. In North India the winter and summer seasons are well defined, whereas in southern plains the temperature difference between the two seasons is not appreciable. The western and the eastern coastal regions of India are dominated by land-sea interaction problems that lead to anomalous propagation conditions. In the Srinagar region, which is in the extreme north, substandard propagation conditions prevail due to cold dry boundary layer.

(5) In the case of ducting model, antenna height was not used and it resulted in unreliable predictions.

3 New prediction procedure

In view of the above problems associated with the ITU-R procedure, there is a dire necessity to develop a new prediction procedure encompassing different propagation mechanisms that occur for different percentages of time on a given path. This development should incorporate such key factors as radiometeorological database with fine resolution, propagation experiments, and propagation

I

1100

I

1000

0t00I

I

0Ill0 TIMF ( hr. I

I

0100

I

0600

I

0500

Fig. 3-Rising layer effect on Delhi-Meerut LOS signal.

II _II 'ii' I I 11 II ~1111~l I~III II I " 1111 IlllI'll loiilM il illi, IIIi

(4)

modelling. We have used tethered balloon flights at a few sites to study the fine structure meteorological data. But the problems associated with this like wind, etc. make this UDwie,Jdyand inconve- nient. This cannot be used for acquiring &tatistica1- ly significant samples of data.

In worst months like premonsoon months when s~rong super-refractive and ducting gradients occur frequently, special slow rising balloon flights have to be conducted at times. This can be done in col- laboration with meteorological departments of the respective countries. Propagation experiments have to be conducted particularly in premonsoon seasons and a continuous radio vigil has to be maintained for catching interfering signals at all frequencies from a few hundred MHz to 10 GHz.

Even the existing radars can be utilized to find out the horizontal extent of anomalous propagation conditions and the fine structure radiometeorological data coupled with this can throw light on the onset, growth and dissipation of anomalous propagation conditions. An example of how the existing meteorological radars can be utilized to map the super-refractive conditions is given below. In the year 1982, S-band cyclone warning radars belong- ing to Meteorological Department situated at Ma- chilipatnam, Madras and Karaikal in the Indian east coast were operated continuously and simul- taneously for 10 days in the month of May when summer reached its peak^7• During anomalous propagation conditons the range of the radars in- creased to as high as 500 to 900 km. In normal times it used to be 100 to 150 km. At Karaikal super-refraction echoes were observed mostly south of the station and Sri Lankan east coast was visi- ble most of the days between noon and evening hours with peak intensity between 1400 and 1600 hrs. From Madras super-refraction echoes were observed between afternoon. and midnight hours but sometimes these echoes were observed till next day morning hours and peak conditions were observed between evening and midnight hours. At this station super-refraction echoes were found to be extending frequently all along the coast from Sri Lanka to West Bengal and occassionally across the bay. At Machilipatnam these conditions start- ed appearing in the afternoon hours to till around midnight and then again in the morning hours from about 0500 to 0700 hrs or sometimes up to 1100 hrs but peak conditions were observed during evening hours. In these stations, the onset of the conditions coincided with the advent of sea breeze and at MachilipatnaJI.1 these super-refractive conditions were seen when light winds prevailed in east, south-east and southern directions.

It was also observed that on. these days when there was' no wind overall super-refractive conditions were weak. Hence .super-refJactive/ ducting conditions are most prevalent when winds are eastedy,south-eastedy or southernly, but strong winds as well as calm conditions are not favour- able for long distance anomalous propagation.

Hence wind direction can be taken as an indicator of ducting conditions; the exact direction is again dependent on local topography and meteorological conditions. As far as sea breezes are concerned, they-~t in by afternoon, persist for several hours and produce large lapses of humidity with height whjch favour the onset of super-refraction propagation along the coast. Also refractivity gradients at various levels have not shown good correlation with the occurrence of anomalous propagation conditions ..

Emphasis has to be placed on the following aspects8 in the prediction procedure: (a) Physical extent of atmospheric ducts over both land and sea surface; (b) Coupling between antenna and duct, including the effect of antenna beamwidth;

(c) Effects of terrain and sea roughness on surface and elevated ducts; (d) Relative importance of surface and elevated ducts over land, sea and coastal areas; and (e) Propagation characteristics of the paths operating at" frequencies less than 10 GHz which terminate at short distances beyond the horizon.

By coupling the radio data and radiometeorological data, propagation models can be developed. Predictions coming from such models can be more accurate and reliable as their inputs have come from fine resolution meteorological data.

The approach consists of providing individual models for each propagation mechanism and com- paring the observed field strength values with the predicted values and identifying the propagation mechanism of the particular interfering signal. The details of the calculations of signal levels for line- of-sight, diffraction, scatter, reflection and ducting are discussed below.

3.1 Line-or-sight

The well known two-ray' interference can be employed to deduce the loss due to ground ray reflection and this can be added to free space loss.

Sometimes the occurrence of ground-based and elevated layers can change the loss. In such cases the loss due to reflection is calculated as given below. If there are sub-refractive gradients, the rays reach the receiver after hitting some obstacle on the path. Then diffraction loss has to be added, which can be deduced using the methods dis-

(5)

332 INDIAN J RADIO &SPACE PHYS, DECEMBER 1996

= -12.95-20 loglOv for v~ 2.4 '" (4) where v is the well known terrain diffraction par- ameter. It is defmed as

cussed below. For the links operating above 10 GHz, water vapour absorption loss has to be added. In fact this is more important as the water vapour concentrations observed over Indian coasts are as high as 26 g/m3• At these frequencies con-

tribution due to reflection can be neglected. v= h[2Il(1/ a+ 1/b}J112 ... (5)

where EIEo is the ratio of field strength at the re- ceiving point to its free-space value. This is given by Fresnel integral .

This is added to the free space loss to get the path loss.

3.2.4 Edward and Durkins method-In this method also the terrain diffraction parameter is defined as in the Epstein-Peterson method. The total loss is the sum of free space loss and terrain diffraction loss. The terrain diffraction loss is given by

... (7) ... (6) J(v)=6.4

+

20 log(

fVZTI +

^v) dB

where h is the height of the obstacle. from the transmitter-receiver line, a and b are the distances of obstacle from transmitter and receiver respectively,

3.2.3 ITU-R method-In this method the terrain diffraction parameter is calculated as above. Uti- lizing that terrain diffraction loss is calculated from

.. . (1) 3.2 Diffraction

Smooth-earth diffraction loss can be deduced by employing ITU-R methods, whereas for single knife-edge diffraction, Epstein-Peterson, Deygout and ITU-R methods can be employed. These methods gave least prediction errors when tested against some VHF data in western India9• In this study the methods utilized are: (1) Blomquist and Laden, (2) Epstein-Peterson, (3) ITU-R, (4) Ed- ward and Durkins, (5) Deygout, and (6) Longley and Rice. The summary of the methods are briefly given below.

3.2.1 Blomquist and Ladell method-In this method the total attenuation is the sum of free space loss, a weighted sum of smooth spherical earth loss and obstacle diffraction loss, urban loss and vegetation loss. The last two factors are ap- plicable depending on the nature of the path whether it is passing through urban area or through area filled with vegetation. The path loss is given by

where ^~o is the free space loss, Uthe urban loss,

and Vthe vegetation loss. EIEo= 1₂

+ i

v,

... (8)

where, Fs is smooth spherical earth diffraction loss and Fdis the obstacle diffraction loss.

3.2.2 Epstein-Peterson method-In this method the ground reflections are neglected and the overall propagation loss can be calculated as the sum of the propagation factors for each obstacle

Fd=F(al, a2, h~)

+

F(a2,a3, h;)

+

F(a3, a4, h;~

... (3) where al is the distance from transmitter to first obstacle, a2 from first to second obstacle, a3 from second to third obstacle, a4from third to fourth obstacle, h~, h'2, h; are the heights of obstacles from line joining transmitter and receiver.

F(a, b, h) is the well known Fresnel-Kirchhoff diffraction propagation factor for a perfectly absorb- ing knife-edge and is given by

F(a, b, h)

= -

6.02 - 9.11 v+ 1.27 v2 forO:E;;v<2.4

where

f

is the frequency in MHz. For ^h

>

v~,the terrain diffraction loss is given by

This holds good for the values of vin the range 0< v< 2. For values of v> 2, EIEo is approximat- edas

... (9)

... (10)

... (11)

v=548 rIf(a+abb)]^1/2 EIEo=0.225 Iv

i

3.2.5 Deygout method-In this method also the attenuation is the sum of free space loss and terrain diffraction loss. The terrain diffraction parameter is defined as

F=20 10glO(hlv)+ 16

3.2.6 Longley and Rice method-This model requires frequency, polarization, ground conduc- tivity and dielectric constant, surface refractivity at ... (2)

FR=

-(F

s +F:.)112d

'I I II ~~III~I I~I II' II I

(6)

Table 2-Comparison of observed path loss values with predicted values for the 13 paths Name of the path

Distance Predicted path loss (dB)

Ian Blomquist

Epstein- ITIJ-R Edward andLongley andObserVedDeygout and LadeU

Peterson DurkinsRice Bombay region

1 Bombay-Bhiwandi

40.0 151.85116.80125.21135.53114.21126.50114.8

2 Bombay-Charoti

111.0 185.70136.25127.67131.08127.60154.34141.50

3 Bombay-Kamala

34.4 140.40130.14149.72128.20117.63135.50117.64

4 Bombay-JETE

36.8 136.52109.52108.80109.13134.16118.50119.52

5 Bombay-Kolad

82.4 156.12134.30137.83123.88130.79128.50122.17

PuneregioD6 Pune-Supe

105.0 135.36134.09134.43134.40140.19135.21128.00

7 Pune-Chincholi

31.0 122.21122.23122.18124.66121.60123.57116.00

8 Pune-Sitara

77.0 149.45139.30137.26139.70139.72144.43138.57

9 Pune-Panchagani

50.0 122.68121.36119.39119.69119.68118.38117.57

10 Pune-Yavat

56.0 121.55120.56122.67119.58120.63118.86118.57

11 Pune-Sural

52.0 129.04128.45128.49134.87128.57129.35124.57

Srinagar region12 Srinagar-Kangan

24.0 143.09135.06134.86135.07135.43137.78134.00

13 Srin~r-Manigam

25.0 134.48132.12132.13132.16135.15131.00131.42

Name of the path

Table 3-Comparison of prediction erors of different prediction methods for the 13 paths

Distance Prediction error (dB)

km

Blomquist Epstein-

ITU-R Edward andLongley andDeygout and LadeU

Peterson DurkinsRice 1 Bombay-Bhiwandi

+25.35-12.22-12.29-1.29-9.7+9.03 40.0 2 Bombay~Charoti

111.0 +44.20-13.83-10.42-13.90+ 12.84-5.25

3 Bombay-Kamala

34.4 +4.9-7.30

-17.86-5.36 -17.87+14.22 4 Bombay-JETE

36.8 + 18.02+ 15.66-9.05-9.70-9.37+ 1.02

5 Bombay-Kolad

82.4 +27.62-4.62+2.29-6.33+9.33+5.80

6 Pune-Supe

105.0 +7.36

+6.43+6.40 + 12.19+7.2]+6.09 7 Pune-chincholi

31.0 +6.21

+6.18+8.66+6.18+5.60+7.57 8 Pune-Sitara

77.0 + 10.80+0.73+ 1.13+ 1.15+5.86-1.31

9 Pune-Panchagani

50.0 +5.11+ 1.79+0.81+2.12+2.11+3.79

10 Pune-Yavat

+2.98+ 1.99+ 1.01+2.06+0.29+4.1 56.0 11 Pune-Surul

52.0 +4.47+ 10.30+3.88+3.92+4.78+4.00

12 Srinagar-Kangan

24.0 +9.11+ 1.08+0.88+ 1.09+1.45+3.78

13 Srinagar-Manigam

25.0 +3.48+0.42+ 1.12+ 1.12+ 1.16+4.15

ground level, transmitter and receiver ground clearances, terrain profile between transmitter and receiver. From all these parametes, diffraction loss is calculated.

Table 2 gives the predicted path loss values using the above methods and the observed path loss values. In this table the value of urban loss is added to Blomquist and Ladell method and Ep- stein-Peterson method for paths around Bombay transmitter due to· its degree of urbanization. In the case of ITU-R, Edward and Durkins and Dey- gout methods, the total path loss is the sum of free space loss and knife-edge diffraction loss for all the paths. The paths lying around Pune and Srina-

gar transmitters do not fall under urban category.

So urban loss is not added to any of the methods for these eight paths. Table 3 shows the prediction errors AP (=Pc - Pm) for all the 13 paths; here Pc

is the calculated path loss and Pm is the measured path loss. The table shows that ITU-R, Deygout and Epstein-Peterson methods can give realistic values if an urban path loss of 10 dB is added to predicted loss. In the Pune and Srinagar region no urban loss is added to any of the methods as the eight propagation paths pass through rural/ sem- iurban areas. Here Epstein-Peterson. ITU-R and Deygout methods give the least mean prediction errors.

(7)

334 INDIAN JRADIO &SPACE PHYS, DECEMBER 1996

3.3 Scatter

In this mode traditional methods like NBS which tend to emphasize on the lower signal levels, i.e. mainly on the reliability of tropo links, are not suitable. GannawaylO has developed an empirical method to predict signal levels from long distance transmitters at VHF using scatter mode. This can be tried and in the event of it not giving good prediction accuracies, the coefficients can be modified. This method lias been deduced on the basis of Yeh's method.

3.4 Reflection

In this model the power reflected from the elevated layers can be calculated from the model of Eklund and Wickertsll. The calculation involves the knowledge of the layer thickness, grazing angle, base and top of the layer heights, etc. This information on layer heights can be obtained by the earlier mentioned slow rising balloons or radi- osonde data and from the statistics of the existing acoustic sounders located at different places of the country. Over temperate climates Wickerts and Nilsson¹² investigated the propagation of 60, 170, 460, 5000 MHz signals over a 160 km path with the help of acoustic sounders and refractometers

over Balctic Sea. They observed that high signal levels at 170 MHz beyond horizon are mainly caused by reflection and duct propagation at 5000

MHz. Low signal levels are due to scattering from inhomogeneities in the refractive index field. They have also observed that duct propagation at 5000 MHz is about 70% of the time in summer and less than 10% during winter. At 170 MHz the opti- mum layer height to give strong reflections is just above the radio horizon at the path midpoint, i.e.

400 m on a 160 km path. They have adopted the expression of Eklund and Wickertsll to calculate the reflected power. It is given by

!!. n2

P,=PFS--- ...

⁽¹²⁾

4~4[1+(n:~rr

where

!!.nTotal change of refractive index through the layer

A- Grazing angle in radians B Thickness of the layer

They have given a nomogram for calculating loss due to reflection ^(Fr) as a function of ^B and ^A- for !!.n= 10-5 and

~=

10 m rad. This formula has been tested by us and can be used for calculating reflected power from elevated layers.

The signal levels calculated by lTU-R method for scatter and NPL prediction method taking scattering and reflection into account are shown in

..

36

--- lTU-R

- NPL

-.-.- EXPERIMENTAL

20 30

EXCEEDED

Fig. 4-Comparison of predicted and oberved signal levels for Delhi-Jalandhar path.

" NPL

---,.---.

---- - --- --

---

~

.~.

«,EXPERIMENTAL

-...---

'-.

-:; r - -_

^...•...

"

lTU-R •••••••

....•...••••.

~ ~ "

.

'.

o I

-

^{...• 24}^~^~^E^II)^~ 30¹⁸

-I

^W^W

-I >

12

-I ^Z

^<J:^l!>(/)

6

'I 'I "I' I II "11 II' Il~n!^I

(8)

Fig. 4 along with the observed signal levels of Jal- andhar signal. The figure shows that the ITU-R method underestimates the signal levels by 5 dB at 3% probability level and by 3.5 dB at 10% probability level. The agreement is better at higher probability levels. The NPL prediction method gives better agreement than the ITU-R at all probability levels. From this it is concluded that Jaland- har TV signal propagates through scattering and reflection, whereas in the nightime combined scattering and reflection is responsible for the increase in signal level. This illustrates statistically the poor agreement of the ITU-R method.

3.5 Ducting

This is an important source of interference at frequencies from VHF to 40 GHz. Various methods are present in literature for the prediction of field strengths in duct propagation. They are: (1) wave guide mode method, (2) phase integral or WKB method, (3) initial value method, (4) contin- uation method, (5) coupled mode method, (6) finite difference method and (7) hybrid ray and mode method. In fact none of these methods have been tried over our country by any group. It is ad- visable to try these methods against the observed values. From their theoretical backgrounds it appears that finite difference method is more suitable.

When the observed field strength values exceed the free space values one can say with confidence that duct propagation is the main mechanism be- cause in duct the losses are proportional to

d-

1as against d~² in free space. The problem arises when the observed values are around 10 to 20 dB below free space values. In these cases either it could be ducting or reflection. Sometimes if the duct is not intense, energy would be leaked -out and the observed values could be less than the predicted values in these cases. Sometimes the interfering signal can propagate by means of more than one mechanism. For example, after travelling in LOS mode for some distance, it can get scat- tered, reflected or ducted depending upon the antenna heights, distances between the transmitting and receiving sites, etc. All these complex propagation mechanisms can be understood only with the help of good models and radio data

4 Conclusions

The inadequacies in the present interference predictions of I1U-R are brought out in the light of our experience in identifying the propagation mechanisms of interfering signals at VHF and above. This calls for new integrated approach for interference predictions based on all possible tropospheric propagation mechanisms, viz

WS,

diffraction, scatter, reflection, ducting, etc. and should be tested by the various groups all over the world. As far as possible efforts should be made to collect fine structure radiometeorological data with the help of meteorological organisations in their respective countries at least in the worst peri- ods which are the premonsoon months in most regions. Such data will form reliable inputs for prediction techniques and the accuracy and applica- bility of these techniques can be validated.

References

1 Hewitt M T, Microwave inteiference propagation prediction in NW Europe, International Microwave Conference, 5MBO 91, Brazil, 1991, pp 246-251.

2 ITU-R Report 569-3, The evaluation of propagation factors in inleiference problems between· sttItions on the surface of the earth at frequencies above about 0.5 GHz, Reports and Recommendations of the 1TU-R, Study group 5, Geneva, 1990.

3 Prasad M V S N, Sarkar S K, Saxena R C &Reddy B M, Anomalous VHF propagation observed at New Delhi, NPL Tech Rep 88-C.5, 0016 (National Physical Laboratory, New Delhi 110 012), 1988,pp 1-17.

4 Prasad M V S N, Dua M K &Reddy B M, Indian JRa- dio &Space Phys, 15 (1986) 92.

5 Prasad M V S N, Sharma Suresh, Mangal Sain &Reddy B M,IEEE Trans Broadcast ( USA), 38 (1) (1992) 33.

6 ITU-R Report 239-6, Evaluation of path loss over mixed land-sea paths, Reports and Recommendations of ITU-R, Study group 5, Geneva, 1990.

7 Dabas R S, Prasad M V S N,Dutta H N & Sarkar S K, Indian JRadio &Space Phys, 25 (1996) 151.

8 General description of COST proj~t in the field of tele- comunications on "Influence of the atmosphere on interference between radio communication systems at frequencies above 1 GHz, Prepared by the working group of COST 205 during the meeting at Leidschendam (NL), Dec 1982.

9 Prasad M V S N, Sain M & Reddy B M, IEEE Trans Broadcast ( USA), 36(3) (1990) 234.

10 Gannaway JN, Radio Commun (UK), (1981) 710.

11 Eklund F&Wickerts S, RadioSci(USA), 3 (1968) 1066.

12 Wickerts S& Nilsson L, cited in Modem topics in micro- wave propagation and ai,..sea interaction, edited by A Zancla, 1973, pp 217-40.

Clear air interference predictions in tropospheric radio wave propagation: Indian experiences