Investigation of Anomalous Propagation Conditions in Central and West African Stations Using High-Resolution GPS Radiosonde Observations

J. Earth Syst. Sci. (2020) 129:220 Ó Indian Academy of Sciences

https://doi.org/10.1007/s12040-020-01477-y (0123456789().,-volV)(0123456789().,-volV)

Investigation of anomalous propagation conditions in Central and West African stations using high-resolution GPS radiosonde observations

1,6, 2 3,4 3 SAMUEL KAISSASSOU *, A LENOUO ,RSTANESSONG ,KAMSU TAMO , 3 3,5 6 AVONDOU ,WPOKAM and J KANKO 1Department of Electrical and Telecommunication Engineering, National Advanced School of Engineering, University of Yaounde 1, P.O. Box 8390, Yaounde, Cameroon. 2Department of Physics, Faculty of Science, University of Douala, P.O. Box 24157, Douala, Cameroon. 3LEMAP, Department of Physics, Faculty of Science, University of Yaounde I, P.O. Box 812, Yaounde, Cameroon. 4Faculty of Agronomy and Agricultural Sciences, School of Wood, Water and Natural Resources, University of Dschang, P.O. Box 786, Ebolowa, Cameroon. 5Department of Physics, Higher Teacher’s Training College, University of Yaounde 1, P.O. Box 47, Yaounde, Cameroon. 6Department of Meteorology, Climatology, Hydrology and Soil Sciences, National Advanced School of Engineering, University of Maroua, P.O. Box 46, Maroua, Cameroon. *Corresponding author. e-mail: [email protected]

MS received 13 December 2019; revised 23 May 2020; accepted 10 July 2020

A comprehensive study on the anomalous propagation (AP) conditions occurring over the central and west African stations was made from 2 years (January 2005–December 2006) high-resolution data measured by GPS (Global Positioning System) radio survey observations. Through data quality control and diagnostic analysis, the probability of AP occurrence and characteristic quantities of the three typical anomalous propagation conditions were given. The sub-refraction, super-refraction and ducting cases were investigated statistically using the vertical proBle of modiBed refractivity gradient. Strong diurnal variation in the percentage occurrence of the AP had its peak during the wet months, while the dry months had the lowest values. From 0600 to 1800 local time (LT) at day (1800–0600 LT at night), the total percentage occurrence of super-refraction, sub-refraction and ducting were 82.5% (78.5%), 11% (15.5%) and 6.5% (6%), respectively. Besides statistical results, local meteorological conditions prevailing over central and west Africa have also been discussed. Keywords. Anomalous propagation; west-central Africa; duct; radiosonde.

1. Introduction the tropopause. The AP conditions can conduct to a propagation of nonstandard electromagnetic Atmospheric refraction has a significant impact on telecommunication systems such as RADARs, the propagation of electromagnetic waves. The LIDARs and other wireless tools. It has been shown discovery of this eAect dates as far back as the that, the electromagnetic waves are bending emergence of the radio wave technology (Freehafer because of the spatial stratiBcation of the refractive 1988). We noticed that anomalous atmospheric index of air sometimes alter the wave direction and conditions for propagation occur generally below intensity. 220 Page 2 of 16 J. Earth Syst. Sci. (2020) 129:220

Amongst those damages are multipath fading height dependence (Fink et al. 2010; Cheng et al. and interference, diAraction due to the Beld 2015). Unfortunately, in west-central African obstacles and so on (Lavergnat and Sylvain 2000; regions, the radio survey stations (Faccani et al. Zribi et al. 2016). Thus awareness of anomalous 2009) are limited and their geographic locations propagation will significantly improve the opera- are not homogeneous. Thus, AP information that tional performance of telecommunication systems. generally requires high-resolution observation Information about radio propagation measure- database resource has almost been inexistent in ments in tropical and sub-Saharan regions are recent years. The AMMA (African Monsoon not well known. Some eAorts based on observa- Multidisciplinary Analysis) campaign was laun- tional data such as in-situ measurements are ched to enhance data in these regions. Further- needed to provide scientiBc explanation and to more, campaigns such as AMMA and other data satisfy engineers’ need. EAorts are also needed for measurements are expected to build a credible the conceptualization of telecommunication tools database in this part of the world. As mentioned such as RADAR or satellite basement in the earlier, neither the climatology nor the meteorol- region. AP detection has been scrutinized in dif- ogy of anomalous propagation in west-central ferent regions of the world based on the vertical Africa has been documented and discussed satis- proBle of atmospheric parameters. Most of the factorily due to unavailability of high-resolution previous works on anomalous propagation condi- data to date. Therefore, the main objectives of tions, at the troposphere, in west-central Africa our current work are to provide an overview of (Parker et al. 2008; Guy et al. 2011; Kaissassou the eAects of anomalous propagation as well as et al. 2015a,b) relied on accurate radio survey the place and time of its occurrence. This paper measurements. also aims to understand impacts of anomalous Previous studies include the works of Falodun propagation conditions and how frequent these and Okeke (2012) which were based on radio wave events occur, which is fundamental for the propagation measurements in Nigeria. Falodun understanding of radar data information and also and Ajewole (2006) have also published an article for the weather forecasting model (Mesoscale on the temporal variability of the refractive index Atmospheric Simulation System and WRF) in Akure city (7.15°N, 5.12°E), Nigeria. They validation, and avoiding misinterpretation of described seasonal variations of the vertical gra- erroneous data. dient index and made some correlation between In addition, it might serve to promote aware- the dry and rainy seasons in Akure. Attempts are ness amongst scientists and operational workers made to evaluate such phenomena in terms of using microwave propagation like telecommuni- statistics of duct conditions over the Wallops cation tools (Kaissassou et al. 2015a, b). The data Island, Virginia (Babin 1996) and using radio used here, collected from the AMMA campaign, survey observations in Barcelona, Spain (Bech were obtained using high-resolution balloon GPS et al. 2002). Other authors like Dalmaz (1977) radio surveys. These radio surveys were launched studied the inCuence of temperature on the several times daily (0000, 0600, 1200, and 1800 refractivity at Goztepe; Barla (1986) noticed UT) in order to provide a vertical atmospheric spatial and seasonal variability in Turkey; Mentes proBle for parameters such as pressure, tempera- and Kaymaz (2007) carried out the statistics of ture, relative humidity, and wind speed. GPS surface duct conditions over Istanbul, Turkey radio surveys have some advantages over classic (41°N, 29°E). The statistical study of microwave radio surveys: they have higher vertical resolu- refractivity is accurate for forecasting AP condition, which facilitates the detection of small AP tions like ducting. These results can be more dimensions. significant if work is done on a larger scale and for Section 2 enumerates the data and methodology. a longer period. An eAort to identify the meteorological conditions Investigations of AP occurrences across the prevailing in west-central African regions is being world were made public recently (Ao 2007; presented in section 3. Section 4 treats the seasonal Mentes and Kaymaz 2007; Lopez 2009; Kaissas- variations of AP over west-central Africa. In sec- sou et al. 2015a,b) using different techniques such tion 5, we discuss the impact of atmospheric as radio survey for an eAective detection of AP. refractivity on propagation of waves. Section 6 The radio survey measures atmospheric parame- summarizes the work followed by a brief ters (viz., temperature, pressure, humidity) with conclusion. J. Earth Syst. Sci. (2020) 129:220 Page 3 of 16 220

2. Data and methodology Furthermore, in the wet season during certain periods (15 July–15 September 2006 called special 2.1 Data observing periods (SOP3)), the number of surveys increased and became regular on a daily basis. This During the AMMA campaign, over 19 stations in was in order to monitor the diurnal cycle of the west-central African cities regularly released high- unusually intense monsoon (Parker et al. 2005). In resolution radio survey balloons (Vaissala RS-80, fact, the AMMA 2006 intensive observing periods RS-92, and MODEM). Figure 1 shows their geo- (IOPs) involved launching of many radio surveys graphic locations. This current study makes use of several times daily (up to 4 times). Thus, the radio 2 years data (January 2005–December 2006). Most survey network widened with new stations (Parker of these radio surveys were launched around 00:00 et al. 2008). and 12:00 LT. Temperature (T), pressure (P), There was collection and storage of radio relative humidity (RH), and horizontal wind pro- surveys in their original forms with a temporal Bles have been collected in different days, months resolution, averaging 5 sec, and a vertical reso- and seasons. lution of about 5 m less than the standard sur- Figure 1 illustrates the spatial distribution of vey of the World’s Meteorological Organization GTS radio survey network and the percentage (WMO). of surveys received at each station, valid at night Due to the increase in the number of daily (1800–0600 LT) and in the day (0600–1800 LT). It launches, several data from the same stations also shows that during the campaign, many dor- were available in the 3 or 6 hrs window, that mant radio survey stations were re-activated, while enable the quantity and the quality of data used other unreliable stations were renovated. In addi- in this study. Data has been separated into two tion, new stations were installed in the Gulf of subgroups: daytime data and night-time data. Guinea from the coast (Douala-Cameroon: For a better description of the vertical atmo- 04°010N, 09°420E; Abidjan-Ivory Coast; 05°150N, spheric structure, we divided the atmosphere into 03°560W) to inland (Niamey-Niger: 13°290N, large numbers of vertical levels. That is a pre- 02°100E; N’Djamena-Chad: 12.088N, 15.028E). requisite for an accurate description of anomalous The number of radio surveys released for most propagation patterns over these regions. Specifi- of the pre-existing stations were increased. cally, radio survey data, which is an important

Figure 1. Data reception through the GTS from west-central African stations from 2005 to 2006. A disk is used to identify each station, the upper half (in red) refers to day surveys and the lower half (in blue) is used for night surveys. The fractional shading of each semicircle denotes the proportion of data received. 220 Page 4 of 16 J. Earth Syst. Sci. (2020) 129:220 source of atmospheric parameters, are crucial to Z Â 106 M ¼ N þ ; our understanding of the thermodynamic pro- R ð2Þ cesses. Consequently, investigations carried out 77:6 4810e to perform and quantify errors resulted to the M ¼ P þ þ 0:157z: T T measurement of several atmospheric parameters from radio surveys. Careful analysis has been In the equation above, z altitude is in meters and made to homogenize the dataset that arose from R is earth’s radius. From equation (2), we random motions of balloons or due to various observe that temperature inversion usually fol- other reasons, for better results. lows duct prevalence accompanied by a rapid decrease in air moisture. Later studies have shown that most AP events appear below 3000 m 2.2 Methodology (Lopez 2009). That explains why the maximum Suitable propagation of electromagnetic waves height computed in our current work did not requires an excellent mastery of the variation of exceed 3000 m. Accurate checking was done to atmospheric condition (parameters) patterns eliminate worthless data measurement before during different seasons of the year, and a statistical calculations. good assessment of their potential inCuences Vertical proBle of the modiBed refractivity is especially on tools that operate on radio wave a valuable input parameter to characterize the paths. impact of atmospheric layers (from the ground Spatial variations in the atmospheric refractivity level to 2000 m) on the propagation of RADAR facilitate the observation of AP conditions. These signals (Willoughby et al. 2002). The increase in changes in refractivity are related to variations of modiBed refractivity of the atmosphere with dM the temperature lapse rate (gradient) and/or water height, dz , directly aAects the propagation of vapour pressure. The expression for the refractiv- electromagnetic waves as previously noted. Dif- ity, N (derived by Steiner and Smith 2002) is given ferent behaviour of the electromagnetic path can be by: observed for a set of given values of atmospheric refractive index. A decrease in its value with height 77:6 4810e in a temperature inversion layer or in the presence N ¼ P þ ; ð1Þ T T of a negative high vertical gradient of water vapour pressure can reCect the beam of radio waves towards the ground. In the case of RADAR, the where T is the absolute thermodynamic temper- reCected beam upon hitting the ground is scattered ature in Kelvin (K), P and e are the atmospheric totally or partially backwards towards the pressure and the water vapour pressure respec- RADAR, consequently the signals can be misin- tively, measured in hectopascal (hPa). Figure terpreted as spurious precipitation. 2(a–d) shows the refractivity proBles in some stations during wet and dry months. For practical purposes, it is appropriate to use a 3. West-central Africa climatology modiBed refractivity M, taking into account the eAectoftheearth’sradiusofcurvature 3.1 Meteorological conditions (R) with altitude z as mentioned by Bech et al. (2007). Oceans, rivers and huge lakes adjust weather in The determination of the modiBed refractivity neighbouring lands. Therefore, inland areas that index includes statistics, diurnal and seasonal are more distant from these large water bodies are variations which are done for the lowest 3000 m generally warmer than coastal areas. During dry height. months there is a significant diurnal increase in For identifying AP, our work is based on temperature and inland regions warm up faster evaluation of the vertical gradient of modiBed than coastal regions due to the phenomenon of refractivity when its value is not between breeze. As result, we observed that large amount of dM À1 79\ dz 157 M km (normal refraction). The heat absorbed into the thin layers of superBcial relationship between the following variables rocks are rapidly released to the environment. At (Bean and Dutton 1996; Turton et al. 1988)is night, the case is different as there is a drastic drop described as: in temperature and inland regions become very J. Earth Syst. Sci. (2020) 129:220 Page 5 of 16 220

Figure 2. Refractivity proBles during (a) dry month (January 2006) in coastal stations, (b) dry month (January 2006) in Sahelian stations, (c) wet month (August 2006) in Sahelian stations, and (d) wet month (August 2006) in coastal stations. cool during the dry season (DJF and MAM). regions. These modiBcations in the elements of Warm, hot and dry climates result during this climate inCuence all states located in the interior of period because rain rarely reaches the interior continents. Besides the Sahelian climate, there 220 Page 6 of 16 J. Earth Syst. Sci. (2020) 129:220

Figure 2. (Continued.) exists the maritime climate that is predominant in 3.2 InCuence of the inter-tropical convergence the coastal areas. In regions such as Tombouctou, zone (ITCZ) Ndjamena, and Agadez, the diurnal and annual rates in temperature are very high due to their The ITCZ is a prominent meteorological geographic location. phenomenon over west/central Africa characterized J. Earth Syst. Sci. (2020) 129:220 Page 7 of 16 220 by low pressure, high surface temperature, cloudi- conditions. Its position migrates between 14° and ness, rain, and convergent trade winds (McGregor 17°N in June and May, and the southern regions and Nieuwolt 1998). below the ITCZ are in the rainy season. Even To evaluate refractivity, radio-meteorological inland cities near the Sahara such as Nouakchott, data acquired by in-situ measurements (radio Tombouctou and Agadez, are under important and survey launched during AMMA campaign) at acceptable heavy rains due to the displacement of stations located in west-central African countries ITCZ northward. There is a decrease in tempera- are used. The locations of the stations were ture and the amount of water vapour, which chosen cautiously to include all the climatic fea- increases the refractivity in the troposphere tures of west-central African regions. In this case, thereby increasing the gradients. After attaining Douala, Abidjan, Dakar and Cotonou were its peak in the north around July end, it retreats designated as stations with coastal climates, down to the south in September, under the inCu- while Ngaoundere and Bamako were chosen for ence of the Harmattan. Thus, the regions north of Guinea Savannah and Sahelian components, 10°–12°N experience dry conditions, subsequently respectively. decreasing the dM/dz. West-central African regions have two dominant seasons. The rainy season starts around mid-April and ends around early November in 4. Seasonal variation of anomalous most of the region. The dry season usually dom- propagation over central and inated by the Harmattan, a dry wind usually west-central Africa characterized by dust, that originates from the south of the Sahara to these regions and is 4.1 Winter season (DJF) strongest during late autumn and winter (late November to mid-March). Figures 3 and 4 show the diurnal cycle of anoma- The seasonal characteristic of anomalous prop- lous propagation over west-central Africa for win- agation proBle and its correlated parameters may ter during the day (0600–1800 LT) and at night be related to the movement of the ITCZ, that (1800–0600 LT). According to these Bgures, AP migration provides and controls the amount of frequencies are generally low at night and high in water and rainfall in the region. Depending on the the day. There is high super-refraction over the displacement of ITCZ, central and west-central entire domain, intermediate values of sub-refrac- Africa may be under monsoon or Harmattan winds. tion and low rates of ducting especially at night. In During monsoon, warm and moist air blows from some West-Central African cities during winter, the Atlantic, consequently wet season prevails. AP conditions are almost absent at night (from However, Harmattan is related to the movement of 1800 to 0600 LT), as seen in Bgure 4. Activities of relative cool and dry wind from Saharan to sub- sub-refraction in Nouhadibou mostly aAected the Saharan areas. The ITCZ movement reaches its dry months of December, January and February peak northward between July and August, while it with the occurrence rate around 70%. This is due reaches its trough southward in January (Ojo 1977; to the diurnal warming, since Nouhadibou is loca- Tapio et al. 2014). When maximum, it goes up ted near the Sahara Desert. In the regions within to 6°N. Hence, the Harmattan dominates all Gulf of Guinea, the duct frequencies during winter regions northward. The inland stations of Niamey, are often lower than the other regions located Agadez, Tombouctou, Ndjamena and Bamako fall below 20°N, with more pronounced values as well. under the inCuence of the Harmattan. As the water Over the whole of central and west-central Africa vapour pressure lowers, both at ground level to altitudes as low as about 1000 m, the rate of reduction at the surface becomes more significant. Table 1. Four different propagation regimes. These eAects result in lowering of the dM/dz Characteristic dN (kmÀ1) dM (kmÀ1) values. A summary of the different propagation dz dz Ducting dN dM regimes is given in table 1. dz \ À 157 dz 0 By early March, the ITCZ is around 10 N and Super-refraction dN dM ° À157\ dz À79 0\ dz 79 the southern regions are exposed to rainy season as Normal À79\ dN 079\ dM 157 long as the ITCZ remains in that position. At this dz dz Sub-refraction 0\ dN 157\ dM moment, the northern areas are still under dry dz dz 220 Page 8 of 16 J. Earth Syst. Sci. (2020) 129:220 during winter, especially during the day, the mean 4.2 Spring season (MAM) AP occurrence is generally maximum over land areas (due to surface evaporation). The exceptions During spring (Bgure 5), the presence of duct rarely are Tombouctou, Tambacounda, Conakry, and aAect most of the cities compared to other seasons. Douala where AP is inexistent. That is because of the low air moisture over these

Figure 3. DJF day seasonal summary of anomalous propagation frequency from 2005 to 2006. These conditions of propagation dM dM are derived in terms of M-index and the gradient intervals are 0\ dz 79 for super-refraction, in green; 157 [ dz for dM sub-refraction, in blue and dz 0 for ducting, in red.

Figure 4. Night seasonal (DJF) summary of AP frequencies from 2005 to 2006. These conditions of propagation are derived in dM dM terms of M-index and the gradient intervals are 0\ dz 79 for super-refraction, in green; 157 [ dz for sub-refraction, in blue dM and dz 0 for ducting, in red. J. Earth Syst. Sci. (2020) 129:220 Page 9 of 16 220

Figure 5. Similar to Bgure 3, although it is for the MAM.

Figure 6. As in Bgure 4, but for MAM. regions. The AP conditions in the day are due to the meteorological conditions occur over the mountain dominant stable conditions as well as wind advec- regions in the night as shown in Bgures 5 and 6. tion from the nearby Atlantic, which usually brings about moisture. Figure 5 shows that super-refrac- 4.3 Summer season (JJA) tion is most significant, sub-refraction is not fre- quently observed, and ducts almost disappear at The pie charts in Bgures 7 and 8 show clearly, the night (because of decreased turbulence at low alti- percentage of occurrence of super-refraction, tude). The absence of ducts at night (Bgure 6), is also sub-refraction, and ducting events at each station observed in other seasons. Generally, the same during summer/rainy season. From Bgure 7, 220 Page 10 of 16 J. Earth Syst. Sci. (2020) 129:220

Figure 7. As in Bgure 3, but for JJA.

Figure 8. As in Bgure 4, but for JJA. results from Agadez, Tombouctou and Ndjamena that is often wet and very warm over the entire stations show that JJA is mostly aAected by region. Extreme ducting conditions occur along super-refractive conditions. The percentage of the coast in the Gulf of Guinea and in Sahelian appearance is within 79–81%. In the same season, cities round the year. In fact, the presence of duct cases have been recorded only for about 6% subsiding dry air at low-altitudes combined with occurrence. the intense evaporation from the wet surface Due to surface evaporation, there are usually provokes negative thin edge moisture gradient at surface ducts in the evenings and rainfall gener- lower altitudes and thus leads to high percentage ally reaches its peak in the morning (e.g., Kais- of anomalous propagation. Moreover, adiabatic sassou et al. 2015a, b) during the JJA season, heating combined with large-scale subsidence J. Earth Syst. Sci. (2020) 129:220 Page 11 of 16 220

Table 2. Probability of occurrence of AP phenomena in rainy above large lakes produce a low-altitude tem- and dry season. perature inversion that creates super-refractive Rainy season (%) Dry season (%) events. The AP diurnal cycle is very pronounced Season in day/night in day/night over all the stations with maximum duct rate at each station in summer. That was the case for Sub-refraction 15/20 09/13 other seasons as well as in winter where we Super-refraction 88/75 83/79 Ducting 9/7 4/5 recorded the lowest frequency, with little average diurnal variations.

Figure 9. As in Bgure 3, but for SON.

Figure 10. As in Bgure 4, but for SON. 220 Page 12 of 16 J. Earth Syst. Sci. (2020) 129:220

Figure 11. Summary of occurrences of ducting (red), super-refraction (green) and sub-refraction (blue) in Cotonou from 2005 to 2006.

Figure 12. As in Bgure 11, but for Ndjamena.

4.4 Autumn season (SON) formation of atmospheric layers. These conditions rise the humidity due to evaporation that occurs on wet In each station, the diurnal variation in refractivity surfaces. has the same proBle with the lowest values recor- In Bgures 9 and 10, ducting phenomena over the 19 ded in the dry season and the highest in the rainy stations remain pronounced in the diurnal cycle. The season. This is summarized in table 2. Despite the region under study is the most inCuenced, particu- progressive onset of the monsoon and the dry larly during autumn because of high reinforcement of season, there are also appearances of diurnal vari- moisture from the surface. Suitable conditions to an ations in AP conditions that are similar to those increase humidity in surfaces, due to the presence of during summer. lakes, and rivers present in the region. However, with the exception of Nouakchott and Toumbouctou whose ducting frequencies are not negligible, there are usually weaker amplitudes at 5. Impact of atmospheric refractivity night that are virtually duct free. Following our pre- on propagation of waves vious analysis, the probability of large variations owing to anomalous propagation is higher during rainy We showed that the AP has seasonal variations seasons. This is due to the presence of lakes and large with higher frequencies during wet months. wet surfaces, which are generally suitable for the However, concerning the eAect of atmospheric J. Earth Syst. Sci. (2020) 129:220 Page 13 of 16 220

Figure 13. Diurnal variation of mean distribution of modiBed refractivity at 3000 m altitude in Cotonou from 2005 to 2006.

Figure 14. As in Bgure 13, but for Ndjamena. refractivity, the damages evaluated during the frequency of an event as well as duct, which is same seasons should be considered as a time oper- caused by variation in air refractivity, is likely to ation. Changes in refractivity alongside the radio have low values in the level of wave propagation wave commonly characterize the damages. The unlike those in the normal atmosphere. These 220 Page 14 of 16 J. Earth Syst. Sci. (2020) 129:220 outcomes imply during certain seasons, modiBca- daytime (06:00–18:00) and in April for night-time tion in the level of radio wave propagation can be (18:00–06:00) (Bgures 13 and 14). The lowest noticed. Reinforcement of the strength is high values of the modiBed refractivity around Febru- during day time and this can lead to superposition ary and April could be seen as a consequence of of waves for long distant operators. Conversely, the the high temperatures which prevailed when the strength remains unchanged when the gradient of rainy season is about to start. Furthermore, modiBed refractivity is not equal to normal value higher value in the modiBed refractivity was seen dM À1 in the wet season with maximum in August (79\ dz 157 M km ). This anomalous propagation prevailing in the region may result to (September). damages (interferences) experienced in some localities in west-central Africa. From that view, super-refraction, for example, could be the source 6. Summary and conclusion of loss in signals from distant mobile phones, radio, or television stations in these cities during the dry For the Brst time, the AP conditions recorded over season. ModiBcation of signals is also observed west-central Africa involve high-resolution radio when they interfere with existing signals of same survey data. frequency, which are locally transmitted. The main outcomes of these studies based on the The quality of the signal, broadcasted by radio/ anomalous propagation of damages on atmospheric television stations or other telecommunication refractivity in the lower troposphere are: services, often depends on the meteorological con- 1. High diurnal variation in anomalous propaga- ditions that dominate the area. Broadcasting ser- tion in all seasons. vices require a good coverage area. It is also 2. The AP frequency from the ground surface to required to minimise interference with other 3000 m of altitude is preferably high in the wet telecommunication services having similar fre- season (MAM-SON). When the dry season quency bands. From this study, the expected duct occurs, around ending November, the total free time in percentage is dominant over other appearance of AP falls drastically to a minimum refractivity cases for the encountered stations. mainly in DJF. 3. We also noticed that maximum appearance of AP indicates the observation of maxima during 5.1 Case study: Cotonou and Ndjamena stations the day, especially in the mornings while low values occur generally from 1800 to 0600 LT for Monthly mean occurrence of the three character- the dry season and from 1200 to 1800 LT in wet istics of propagation conditions (super-refraction, months. ducting and sub-refraction) for Cotonou and 4. The occurrence of ducting characterizes Ndjamena are depicted in Bgures 11 and 12. These seasonal variation of AP in west-central Africa. statistics of occurrences of various propagation The later reaches its peak at daytime during the conditions were computed every month over the dry and rainy season. 2005–2006 period of study. These Bgures show that 5. Temperature inversion is largely responsible for these anomalous propagation conditions have dif- ducting observed during the rainy season. The ferent occurrence pattern. We noticed that, the variability in the gradient of modiBed refractivity is occurrence of super-refraction is always high in also high during rainy seasons. This could be both cities during all the months of the year. For because of humidity inversion during these periods. the two years of study, the ducting has the lowest 6. High negative values in the vertical gradient of occurrences and, sub-refractive events have values modiBed refractivity characterizes the lower in between. These stations are under ducting eAect atmosphere, around 3000 m above the sea level, all seasons, combined with super-refraction, the during the day (0600–1800 LT), while values are both conditions may have exposed Cotonou and low (or even positive) at night (1800–0600 LT). Ndjamena under the interference of radio signal That leads to an almost permanent case of super- from distant stations. refraction. By November, when the Harmattan wind season is dominant in the region, the values of The detection of AP patterns in telecommuni- modiBed refractivity fall sharply and reach their cation instruments like RADAR echoes is crucial minimum for the periods mainly in February for for their proper operation and application in real J. Earth Syst. Sci. (2020) 129:220 Page 15 of 16 220 life estimation of rainfall. If not taken into account, Dalmaz M 1977 Variation de la refraction atmospherique en it might be misinterpreted, thus lowering public fonction de la temperature aG oztepe;€ La Met eorologie . VI credibility on the Cood warning and forecasting 10 113–124. Faccani C, Rabier F, Fourrie N, Agusti-Panareda A, Karbou systems. F, Moll P, Lafore J P, Nuret M, Hdidou F and Bock O 2009 Despite the short-term study due to unavail- The impacts of AMMA radio survey data on the French ability of in-situ high-resolution data, these pre- global assimilation and forecast system; Wea. Forecasting, liminary statistical results can serve as a reference https://doi.org/10.1175/2009waf2222237.1. document for future investigation and comparative Falodun S E and Ajewole M O 2006 Radio refractive index in studies or research on anomalous propagation in the lowest 100-m layer of the troposphere in Akure, southwestern Nigeria; J. Atmos. Sol. Terr. Phys. 68(2) these regions. 236–243. Falodun S E and Okeke P O 2012 Radio wave propagation measurements in Nigeria (preliminary reports); Theor. Acknowledgement Appl. Climatol. 113 127–135. Fink A H, Agusti-Panareda A, Parker D J, Lafore J P, Our sincere appreciation goes to AMMA for pro- Ngamini J B, ABesimama E, Beljaars A, Bock O, Christoph M, Dide F, Faccani C, Fourrie N, Karbou F, Polcher J, viding the radio survey data used for the analysis in Mumba Z, Nuret M, Pohle S, Rabier F, Tompkins A M and our work (http://www.amma-international.org). Wilson G 2010 Operational meteorology in west-central Africa: Observational networks, weather analysis and forecasting; Atmos. Sci. Lett. 12 135–141. Author statement Freehafer J E 1988 Tropospheric refraction; In: Propagation of short radio waves (ed.) Kerr D E, Peninsula Publishing, pp. 9–22. Samuel Kaissassou developed code, designed, Guy N, Rutledge S and Cifelli R 2011 Radar characteristics of performed and analysed the simulation experi- continental, coastal, and maritime convection observed ments and prepared the manuscript with contri- during AMMA/NAMMA; Quart. J. Roy. Meteorol. Soc. butions from all co-authors. A Lenouo, A Vondou, 137 1241–1256. R Tanessong, and W Pokam supervised the Kaissassou S, Lenouo A, Nzeukou A, Tchawoua C and Vondou research, co-designed the experiments and D 2015a Seasonal variations of surface duct conditions in Ngaoundere, North Cameroon; Meteorol. Atmos. Phys. reviewed the manuscript. K Tamo and J Kanko 127, https://doi.org/10.1007/s00703-015-0387-z. performed Bgures and quality control of data. Kaissassou S, Lenouo A, Tchawoua C, Lopez P and Amadou T 2015b Climatology of anomalous propagation radar over west-central Africa; J. Atmos. Sol.-Terr. Phys. 123 1–12. Lavergnat J and Sylvain M 2000 Radio wave propagation; In: References Principles and Techniques, John Wiley and Sons, New York, ISBN:047149027x. Ao C O 2007 EAect of ducting on radio occultation measure- Lopez P 2009 A 5-year 40-km-resolution global climatology of ments: An assessment based on high-resolution radio super refraction for ground-based weather radars; J. Appl. survey; Radio Sci. 42 RS2008, https://doi.org/10.1029/ Meteorol. Climatol. 48 89–110. 2006rs003485. McGregor G R and Nieuwolt S 1998 Tropical climatology; 2nd Babin S M 1996 Surface duct height distributions for Wallops edn, Wiley, Chichester, 339p. Island, Virginia, 1985–1994; J. Appl. Meteorol. 35 86–93. Mentes S and Kaymaz Z 2007 Investigation of surface duct Barla M C 1986 Sur les indices de refraction atmospherique en conditions over Istanbul, Turkey; J. Appl. Meteorol. Turquie; La Met eorologie . VII 13 2–6. Climatol. 46 318–337. Bean B R and Dutton E J 1996 Radio-meteorology; Dover Ojo O 1977 The climates of west Africa; Heinneman, London, Publishers, New York, 435p. pp. 60–72. Bech J, Codina B and Lorente J 2007 Forecasting weather Parker D J, Burton R R, Diongue-Niang A, Ellis R J, Felton M, radar propagation conditions; Meteorol. Atmos. Phys. 96 Taylor C M, Thorncroft C D, BessemoulinPandTompkinsA 229–243. M 2005 The diurnal cycle of the west African monsoon Bech J, Codina B, Lorente J and Bebbington D 2002 Monthly circulation; Quart. J. Roy. Meteor. Soc. 131 2839–2860. and daily variations of radar anomalous propagation Parker D J, Fink A, Janicot S, Ngamini J B, Douglas M, conditions: How ‘normal’ is normal propagation? In: ABesimama E, Panareda A A, Beljaars A, Dide F, Diedhiou Proceedings of Second European Conference of Radar A, Lebel T, Polcher J, Redelsperger J L, Thorncroft C and Meteorology, Delft, Netherlands; ERAD Publ. Ser vol. 1, Wilson G A 2008 The AMMA radio survey program and its Copernicus GmbH, pp. 35–39. implications for the future of atmospheric monitoring over Cheng Y, Shengqi Z, Dongxiao W, Yuanzheng L and Jinglong Africa; Bull. Am. Meteorol. Soc., https://doi.org/10.1175/ Y 2015 Statistical characteristics of the surface ducts over 2008bams2436.1. the South China Sea from GPS radio survey data; Acta Steiner M and Smith J A 2002 Use of three-dimensional Oceanol. Sin. 34 63–70. reCectivity structure for automated detection and removal 220 Page 16 of 16 J. Earth Syst. Sci. (2020) 129:220

of non-precipitating echoes in radar data; J. Atmos. Willoughby A A, Aro T and Owolabi I E 2002 Seasonal Oceanic Technol. 19 673–686. variations of radio refractivity gradients in Nigeria; J. Tapio S, Tobias B and Gerald H 2014 Migrations and Atmos. Sol.-Terr. Phys. 64 417–425. dynamics of the inter-tropical convergence zone; Nature Zribi M, Sahnoun M, Baghdadi N, Le Toan T and Hamida A Geosci. 513, https://doi.org/10.1038/nature13636. 2016 Analysis of the relationship between backscattered Turton J D, Bennetts D A and Farmer S F 1988 An P-band radar signals and soil roughness; Remote Sens. introduction to radio ducting; Meteor. Mag. 117 245–254. Environ. 186 13–21.

Corresponding editor: SURESH BABU