Introduction

Due to climate change and climate variability, there has been an increase in the intensity, extend and frequency of natural disasters, that has led to the occurrences of compound natural disasters which usually results in serious damage and loss of lives1. These natural disasters include earthquakes, landslides, hurricanes, volcanoes, droughts, and floods2. More so, natural disasters have resulted in serious catastrophes, for instance, the destruction of infrastructure, framework disturbance of livelihoods, stressing of government’s monetary spending plans, disease outbreaks and loss of both animal and human lives. Globally, there were over 300 annual occurrences of these natural hazards over the past decade3. In fact natural disaters claims tens of thousands of lives each year4. Over the past decade, approximately 45,000 people globally lost their lives from the natural disasters recorded, annually. The spatial distribution of the natural disasters spans different geographic locations including storms and floods affecting areas along the northern Pacific, the Caribbean Sea and Madagascar5. Africa is mainly affected by occurances of epidemics, droughts and floods5. Furthermore, the main disasters affecting the Alpine-Himalayan belt and the western Andes include earthquakes and floods with America, Europe, Oceania, and Asia being predominantly affected by floods5. Tsunamis mainly cause high fatality rates in countries like Thailand, Indonesia, and Japan5.

Besides the high fatality rates associated with these natural disasters, there is normally wide spread destruction of infrastructure, farm land, animals, ecosystems and livelihoods that follows. The 2011 Tohoku earthquake and tsunami in Japan, caused an estimated $210 billion in infrastructure damage, with Hurricane Katrina, the 2010 Haiti earthquake, the 2015 Nepal earthquake and the 2013 Typhoon Haiyan in the Philippines being responsible for infrastructural losses of up to $20 billion, $8 billion, $5.1 billion, and $2 billion in estimated damages respectively. The February 2023 Cyclone Gabrielle was reported to have contributed to an estimated $1 billion losses in farming infrastructure6, and destroyed electricity, transport and telecommunication infrastructure6,7. These losses in infrastructure further make difficult any post disaster recovery activities.

Infrastructure including road, health and communication platforms are critical in both disaster risk reduction and post-disaster recovery. The resilient infrastructure, which is designed to withstand potential hazards, plays a preventative role by minimising the impact of natural disasters8. This includes such infrastructure for instance earthquake-resistant buildings, flood-proof roads and resilient communication systems. Furthermore, infrastructure especially communication systems facilitate early warning systems, enabling timely evacuations and preparedness. Post-disaster, infrastructure becomes even more crucial as these provide essential services such as water, electricity, communication, and transportation, which are vital for immediate relief and long-term recovery. Considering the critical role played by communication infrastructure in pre and post-disaster reduction and recovery efforts, various studies have characterised the vulnerability of telecommunication infrastructure with a study in Australia highlighting telecommunications infrastructure as essential in rural communities for responding to natural disasters, like bushfires, cyclones and floods9. The effect of these infrastructural losses is further amplified in developing countries like Zimbabwe, that have limited financial capabilities.

In Zimbabwe, the most common natural disasters are tropical cyclones, droughts, landslides, cyclones and floods10. Landslides, often triggered by heavy rainfall associated with cyclones, pose a localised but significant threat, particularly in areas with steep slopes and unstable soil that characterise the eastern-areas of Zimbabwe8,11. While Zimbabwe isn’t directly in the main cyclone belt, the country experiences effects of cyclones that affect neighbouring countries11. These events can bring strong winds, heavy rainfall, and subsequent flooding, causing significant damage to infrastructure like roads, bridges, and power lines, as well as loss of life and displacement. Flooding and cyclones are a recurring hazard, often caused by heavy rains and overflowing rivers, impacting low-lying areas and those near waterways. Cyclones, landslides and the subsequent cyclone induced floods normally affect the eastern-areas of Zimbabwe notably Chipinge and Chimanimani. These include Cyclone Eline (2000), Cyclone Idai (2019) and Cyclone Ana (2021). Cyclone Idai was responsible for the death of 172 people, left more than 327 people missing, and displaced approximately 4500 people12. More so, in Chimanimani and Chipinge, some mobile network operators’ base stations were flooded and communication was cut off completely while grid electricity was also cut off for almost one month, affecting communication and social media platforms such as WhatsApp which are widely used at all disaster risk recovery (DRR) cycle stages9. Radio was also affected which further impacted DRR activities considering that radio provides the widest reach9. To this end, various studies have been conducted to show the importance of resilient communication infrastructure11 and other DRR activities in Chipinge, but little has been done to actually propose methodologies to implement resilient communication infrastructure that include radio platforms. In fact, little has been done to guarantee the resilience of communication and early warning systems disposable to the community in these disaster-prone areas like Chipinge in Zimbabwe. Considering that these communication platforms including radio antennas are located in geographic space, Geographic Information Systems (GIS) and geospatial technologies provide promising and proven techniques in the siting of different phenomena including resilient radio communication systems at optimum locations through Remote Sensing (RS), GIS aided Analytical Hierarchical Process (AHP) and Multi-Criteria Decision Analysis (MCDA).

Both AHP and MCDA have been used in the estimation of optimum locations for different phenomena including solar farms13, hospital site selection14, municipal solid waste landfill site selection15 and optimum base station sites. However, little has been done to use the same techniques in disaster risk management in particular for the optimum location of radio transmitters to be used as an early warning system and a communication tool before, during and after a disaster. This paper therefore implements a GIS aided AHP and MCDA analysis to determine the optimum locations of radio transmitters in Chipinge district.

Methods and materials

Study area

The research was carried out in Chipinge district (Fig. 1), Manicaland province, in southeastern Zimbabwe near the Mozambique border. There are 31 rural administrative wards and 8 urban wards in Chipinge district16. The district is divided into assemblies namely Chipinge South, Chipinge West, Chipinge East, Chipinge Central and Musikavanhu. Chipinge has a warm and temperate climate. Furthermore, Chipinge sits at about 1132 m above sea level and receives up to 1097.5 mm of rainfall annually17. The area is vulnerable to tropical cyclones since it falls in the overland path of cyclones originating in Mozambique and the Indian Ocean11.

Fig. 1
figure 1

Study area (a) Zimbabwe (b) Chipinge. Map generated in ArcMap version 10.8.

Full size image

Methods and materials

Various criteria (data sets) were considered for the siting of the resilient radio transmitters. The criteria were derived from literature and also from expert opinion. The data includes roads, rivers, settlements, power lines and elevation. The criteria’s information layers were created in a GIS (Arc Map 10.8). The layers created for every dataset which was in vector data format was converted to raster data format (rasterization), then used to create the distance maps with the use of the Euclidean distance and a digital elevation model (DEM) which was used to come up with the elevation map. The layers were then standardized where reclassification is used, using the multi-binary classification. The Analytic Hierarchy Process (AHP) was used for assigning weights to the different criteria. At this stage there is the assignment of weights and the weight assessments where these are utilised to complete the pairwise comparison having the un-normalized matrix (Table 3) and normalized matrix (Table 4) and finally assessing the weights using the constant ratio. The criteria aggregation was performed using the raster calculator which was applied to limit (constraints) to each dataset map.  Weights, calculated in Microsoft excel, were then applied to produce the final suitable sites. The Fig. 2 below summarises the methodology followed in the research.

Fig. 2
figure 2

Methodology flow chart diagram.

Full size image

Criteria selection and data acquisition

The data for roads (line data), rivers (line data), settlements (point data), and power lines (line data) were collected in vector format from the respective authorities responsible for the data in Zimbabwe. However, the DEM was downloaded from the Google Earth Engine (GEE) catalogue of the Shuttle Radar Topography Mission (SRTM) digital elevation dataset with a 30 m resolution. These variables were developed using expert opinions, previous research studies, and relevant legislation and regulations.

Methods of site selection

The Multi-Criteria Decision-Making Analysis (MCDA) approach is used in a Geography Information System (GIS) for the optimum placement of resilient radio transmitters in Chipinge district. The following criterions should be met:

  1. 1.

    Proximity to community settlements-should be close to people (C1).

  2. 2.

    Higher areas to avoid disaster risk/elevation (C2).

  3. 3.

    Proximity to power source/electricity (C3).

  4. 4.

    Proximity to water bodies/rivers—the radio transmitters must not be close to water bodies (C4).

  5. 5.

    Proximity to major roads (C5).

  6. 6.

    Proximity to district boundaries (C6).

The research used MCDA theories to estimate the suitable sites for the siting of the community station radio transmitters using the following:

  1. 1.

    Binary overlay (Boolean operators) method to determine the potential site.

  2. 2.

    Weighted Linear Combination (WLC) method for selecting alternatives from potential site areas.

  3. 3.

    Analytic Hierarchy Process method (AHP) to determine the preferred candidate sites from potential sites.

Multi-criteria decision analysis and GIS

MCDA, assists analysts and decision-makers in scenarios requiring the identification of priorities based on numerous criteria. This is common in settings when competing interests coexist18. MCDA can incorporate geographical data and stakeholder preferences into quantifiable values for assessment and subsequent decisions19. One of the most powerful components of a decision support system is the visualization of the context, structure of the problem, and alternative solutions. Thus, the combination of GIS with MCDA has the goal of assisting decision-makers by providing them with methods to evaluate many possibilities based on different, conflicting criteria18. The MCDA procedures are decision rules that connect the input and output maps. Both raster and vector GIS systems can be used to build MCDA approaches. Some GIS, for instance QGIS and ArcMap have built-in routines for the Weighted Linear Combination (WLC) and Ordered Weighted Averaging (OWA) processes, and they are also available as open-source modules or scripts, such as in QGIS20, which can be used to perform a specific type of MCDA. The multi-criteria analysis method was utilized to classify and weigh criteria. Quantitative analysis, such as scoring, ranking, and weighting, is required for multi-criteria analysis21, therefore the WLC was used due to the low interaction in-between the criterions.

Euclidean distance

The Euclidean distance was computed to convert vector data into raster data known as rasterization in Arc Map 10.8. The Euclidean distance tool was used to identify the land parcels that were within a certain distance from the edge of the settlements, rivers, roads, and power lines.

Standardization of the criterion

Since the criteria are evaluated on different scales; it is critical to normalize factors before merging them and, if necessary, to adjust them so that all criteria maps are positively associated with suitability. This was done by setting the suitability values of the factors to a common scale.

Analytic hierarchy process (AHP) method

The Analytic Hierarchy Process (AHP) uses a matrix, where each criterion is compared with the other criteria, relative to its importance22, in this case on a scale from 1 to 9 (Table 1). Where 1 = equal preference between two factors; 9 = a particular factor is extremely favoured over the other, furthermore, a weight estimate is calculated and used to derive a consistency ratio (CR) of the pairwise comparisons. If CR > 0.10, then some pairwise values need to be reconsidered & the process is repeated until the desired value of CR < 0.10 is reached.

Table 1 Fundamental scale of absolute numbers.
Full size table

Table 2 below shows the different weights assigned to the criteria derived from literature and expert opinions.

Table 2 Discretionary criteria and weights.
Full size table

Table 3 shows the weights assigned to each criterion and the inverse, and in ranking and rating.

Table 3 Un-normalized pairwise matrix.
Full size table

AHP weights were also expressed in numerical weights (Table 4) that sum up to 1 where each criterion has a weight.

Table 4 Normalized pairwise matrix.
Full size table

Weighted linear combination (WLC)

WLC method is a decision rule used to create composite maps in a GIS21. The weighting of each criterion by the decision maker is the foundation of the WLC approach. For each option, a total score is derived by multiplying the assigned weight value by a scaled value multiplied by the criterion25. If there are m alternatives and n criteria, each alternative is evaluated independently for each criterion. Each criterion’s weights represent its importance in comparison to the other criteria26. WLC is based on the weighted average concept, in which continuous criteria are standardized to a common numeric range and then combined using the weighted average. The decision maker directly assigns relative relevance weights to each attribute map layer. The overall score for each alternative is computed by multiplying the significance weight assigned to each attribute by the scaled value assigned to that attribute in the alternative, then combining the products across all attributes. The scores for each choice are added up, and the one with the highest overall score is chosen. The method may be used with any GIS system that supports overlays, and it allows the evaluation criterion map layers to be blended to form the final composite map layer. The WLC approach combines attributes by assigning each one a numerical value. The findings are then combined to create a suitability map. The criteria chosen (Table 5) for this study are settlements, roads, rivers, power lines and elevation.

Table 5 Chosen criteria and its respective buffers (km) for the suitable sites’ selections.
Full size table

Results

Settlements

Figure 3 shows the suitability map for siting the community radio transmitters when considering settlements. The results show that the local community radio transmitters can be sited in the eastern parts and the central parts of the district. Approximately 81% of the study area is most suitable for the siting, approximately 7% moderately suitable, and approximately 12% is unsuitable for siting the community radio transmitters. Areas, which are suitable, include ward 1, 2, 7, 9, 13, 23 and 25. The unsuitable, areas are found in ward 5, 27 and 30.

Fig. 3
figure 3

Radio transmitter site suitability map considering settlements. Map generated in ArcMap version 10.8.

Full size image

Elevation

Figure 4 shows the suitability map for siting the community radio transmitters when considering elevation. The most suitable areas to site the local community radio transmitters in the district are in the north-eastern parts and cover approximately 40% of the study area. The unsuitable areas for the siting cover approximately 50% of the study area  (southern region of the district). Areas which are suitable include ward 2, 6, 7, 10, 11, 12, 17 and 19.

Fig. 4
figure 4

Radio transmitter site suitability map considering elevation. Map generated in ArcMap version 10.8.

Full size image

Rivers

Figure 5 shows the suitability map for siting the community radio transmitters when considering rivers. The results show that approximately 90% of the study area is most suitable for the siting of the local community radio stations and approximately 10% of the area is not suitable to site the community radio transmitters. The suitable area is mostly concentrated in the central parts of the study area. Areas which are suitable include ward 2, 7, 12, 16 and 19. There are unsuitable areas which include ward 11, 15, 18, 23, 25 and 28.

Fig. 5
figure 5

River’s suitability map. Map generated in ArcMap version 10.8.

Full size image

Roads

Figure 6 shows the suitability map for siting the community radio transmitters when considering roads. The results show that approximately 80% of the district is most suitable for the siting of the transmitters for the local community radio stations and approximately 20% of the area is not suitable to site the community radio transmitters. Areas which are suitable for the siting of community radio station transmitters include ward 7, 8, 9, 11, 12, 13, 15 and 26.

Fig. 6
figure 6

Radio transmitter site suitability map considering roads. Map generated in ArcMap version 10.8.

Full size image

Power lines

Figure 7 shows the suitability map for siting the community radio transmitters when considering power lines. The results show that an estimated 45% of the district is most suitable to site the local community radio transmitters and approximately 55% of the area is not suitable to site the community radio transmitters. Areas which are suitable for the siting of the community radio transmitters include ward 2, 7, 8, 10, 11, 12, 17, 19, 24 and 26.

Fig. 7
figure 7

Radio transmitter site suitability map considering power lines. Map generated in ArcMap version 10.8.

Full size image

Final suitability map

Figure 8 shows the final suitability map for siting the community radio transmitters when considering all the criterions which are settlements, elevation, power lines, roads, and rivers. The results shows that the local community radio transmitters can be sited in the north-eastern parts of Chipinge district covering approximately 40% of the study area in wards 2, 6, 7, 8, 10, 12, and 13 (most suitable sites). Furthermore, approximately 25% of the study area is highly suitable for the siting of the transmitters, and the area includes ward 22, 23, 24, and 29. More so, approximately 20% of the study area is suitable (ward 1, 3, 4, 20), and 25, approximately 5% is moderately suitable , and approximately 10% of the study area is unsuitable (ward 5, 27, and 30) to site the community radio station transmitters.

Fig. 8
figure 8

Final suitability map for the siting of community radio station transmitters. Map generated in ArcMap version 10.8.

Full size image

Discussion

The results in Fig. 3 show that almost the entire study area is suitable for siting the local community radio transmitters when considering settlements. This is due to the fact the local radio transmitters should be sited, close to the settlements27. In fact, the frequency modulation (FM) signals used in such communication transmitters travel following the line of sight, hence the settlements and the transmitters must have a line of site28. The antenna and radio receiver(used by the villagers) must be able to “see” each other, with no obstacles in the way such as hills or tall buildings, therefore these community radio stations must be sited close to settlements29. Furthermore, the developed early warning system tool must be sited close to the settlements in such a way that when the risk disaster information is being broadcasted30, people in the community can receive and respond to the information and practise risk disaster preparedness for instance evacuation of people in low lying areas who are at a higher risk of disasters.

The results for the site selection considering just the elevation data show that the northern parts of the district are most suitable due to the regions higher elevation31. This has the effect of lowering the probability of the transmitters being destroyed or flooded during the cyclone induced floods that normally characterise Chipinge district due to its proximity with the Indian ocean32. Hence the most suitable areas to site the local community radio station transmitters are the northern parts of the district, with the southern parts of the study area being unsuitable as this is a valley33. However, FM has limits in mountainous places because even if the antenna is situated on top of the peak, there may be signal shadows in the valleys. The only way to use FM in hilly terrain may be to install one or more relay transmitters to cover the shadowed sections29.

Considering, that the local radio transmitters should be sited away from the rivers, as areas close to rivers are more prone to flooding34, the majority of the study area is suitable for the transmitter siting as there are few rivers in the district. Furthermore, considering roads, results show that almost every part of the area of interest is suitable for the siting the local community radio transmitter. This is due to the fact that the transmitter should be sited close to settlements which are naturally close to roads, allowing easy access to the transmitters. More so, considering the electricity network, the community radio transmitters need electricity power to function and broadcast the information however the community radio is mostly dependent on low-power FM transmitters with an output of between 20 and 500 watts29.

The results in Fig. 8 show the final suitability map for siting the community radio station transmitter when considering all the criterions which are settlements, elevation, power lines, roads, and rivers These criteria were overlayed using the weighted linear combination. The map was categorised into 5 classes to determine the suitability level to site the radio transmitters which are most suitable, highly suitable, suitable, moderately suitable, and unsuitable. The most suitable area to site the local community radio transmitter in the district (northeastern part) is attributed to the fact that in weighted linear combination the criterions with more weights such as the settlements dominated the suitability area which is the most favourable and few areas were unsuitable considering the weights of the settlements, proximity to power lines, proximity to the roads, and the proximity to the rivers.

The methodology, utilising AHP, GIS, and remote sensing, presents several inherent limitations. AHP’s reliance on subjective expert judgment can introduce biases and inconsistencies, potentially leading to rank reversals and difficulties with large numbers of criteria. GIS analysis is constrained by data availability, quality, and integration challenges, requiring specialized expertise and potentially oversimplifying real-world complexities. Remote sensing is limited by data acquisition costs, atmospheric effects, spectral and spatial resolution constraints, and the need for specialized processing and ground truthing. Furthermore, combining these methods can compound individual limitations, as inaccuracies in one component can propagate through the entire process. These limitations necessitate careful consideration during research design and analysis, including efforts to mitigate their impact and transparently acknowledge their potential influence on the results.

Conclusions

This research, focusing on enhancing disaster preparedness and response in Zimbabwe, offers potential benefits to several key stakeholders. Local communities stand to gain from improved access to timely and critical disaster-related information through the strategically located community radio transmitters. Furthermore, disaster management agencies and government officials can utilise the site suitability map to inform decisions regarding communication infrastructure development and the implementation of more effective early warning systems. Moreso, researchers and practitioners in disaster risk reduction, GIS, and remote sensing will benefit from the study’s methodological approach, potentially adapting and applying it to other contexts. In conclusion, this study has demonstrated the application of a multi-criteria decision analysis (MCDA) approach, integrating AHP, GIS, and remote sensing, to identify suitable locations for a community radio station transmitters in Zimbabwe. The resulting site suitability map, considering criteria such as settlements, elevation, power lines, roads, and rivers, provides valuable information for local communities and disaster management agencies. The identification of optimal locations for this vital communication tool contributes to improved dissemination of early warnings and disaster-related information, ultimately strengthening community resilience. This methodology can be adapted and applied in other regions and for different types of infrastructure critical for disaster risk reduction, offering a valuable contribution to the field. Future research could focus on validating the effectiveness of the chosen location during simulated or real disaster events and explore the integration of the radio station within a more comprehensive early warning system.