Abstract
Every year, hundreds of people go missing in the wilderness of Saudi Arabia. There is an urgent need to examine modern geographic techniques for finding such people. Geographical information systems, for example, play a crucial role in wilderness search and rescue (WiSAR), not only in mapping probability areas but also in applying further analysis and modeling methods to reduce time and effort and to guide life-saving task forces in the right direction. In this study of a hypothetical missing-person case in Saudi Arabia, two standard WiSAR models are compared: ring and mobility. In the presented study situation, both models can be used. However, the new approach used in the mobility model drastically reduces the extent of the possible search area, from 101,787 km2 in the ring model to 335.34 km of likely trails and unpaved roads, and also provides exact directions to where the missing person may be found.
1 Introduction
Search and rescue (SAR) is the operation for locating missing individuals as soon as possible [1]. “Search” is the term used for the first part of the operation, which entails locating the missing item or subject. “Rescue” refers to the second phase, which entails saving the missing item or subject after determining its location and returning it to a safer area [2]. Cases of missing persons (excluding kidnappings) may occur in urban areas; these are usually associated with children, elderly people, or mentally ill individuals [3]. Cases also occur outside urban areas in the wilderness (wilderness search and rescue [WiSAR] cases); these typically involve hikers, riders, tourists, etc. [4,5].
When a person is missing outside of a city, or in a wilderness area, SAR personnel face the task of locating them. This challenge involves limited resources and the possibility of a long search period, which can reduce the survival chances of the missing person [6]. The most common method for locating a missing individual is to create probability maps depicting potential locations. Numerous researchers have employed the ring model to create such probability maps [2,7].
Since World War I, SAR strategies and methods have evolved by utilizing dogs to locate wounded soldiers on battlefields (National Search and Rescue Dogs Association). Following World War II, WiSAR techniques evolved to include mathematical frameworks to increase the likelihood of locating missing objects [2]. Over time, these techniques evolved into different approaches, including the use of airplanes and helicopters, which continue to be employed in many WiSAR missions worldwide [8,9]. In addition, World War II witnessed a rise in the usage of probability maps, created by mathematical calculations or based on lost-person behavior, which resulted in the creation of probability areas identifying the most probable locations where an object might be located [7].
SAR approaches have evolved to incorporate remote sensing techniques, such as drones and unmanned technologies [1,3,10,11,12,13,14], as well as various geographical information systems (GIS) and mapping techniques, such as geostatistical and network analysis [7,15,16,17]. However, a variety of WiSAR-related factors, such as terrain, land cover, climate, and weather conditions, as well as a lack of data, play a crucial role in selecting the most appropriate approach. One factor of particular importance is the behavior of the missing person, as awareness of this can lead to a more subjective approach [18].
In this study, we aim to provide an overview of WiSAR in the Kingdom of Saudi Arabia regarding the methods used by local task forces and volunteering groups. We also adopt a new approach that is more suitable for people in arid and semi-arid regions, which share the same environmental and anthropogenic characteristics as those in a hypothetical-case study in the Al-Quwayiyah governorate in the middle of Saudi Arabia. Many areas of the world share similar environmental characteristics to those of Saudi Arabia, including Australia, where 70% of the country is classified as semi-arid or arid, with higher proportions still in central and western regions [19].
1.1 WiSAR missions in Saudi Arabia
Between 2009 and 2020, over 2,497 people went missing in the Saudi Arabian wilderness; 112 of these were found deceased, representing roughly 4.5% of the total number [20]. However, according to official reports from volunteer groups [21], up to 25% of missing-person cases in Saudi Arabian deserts each year end with the death of the individual involved. In addition, since 2014, WiSAR missions in Saudi Arabia have been led by the police instead of the civil defense force, with the help of volunteers who are typically untrained and unqualified.
The WiSAR process in Saudi Arabia usually begins when relatives of a missing person contact the local police in the province concerned. Usually, the police in Saudi Arabia start the investigation after 24 h by calling the local telecommunication company to inquire about the last time the missing person’s phone was connected to the network (cell sites); this process involves a further period of between 1 and 5 h. Meanwhile, the relatives of the missing person contact the volunteering groups. This also takes up several hours until they meet at the last known point (LKP)/last point seen, which is generally around the cell site area where the initial planning point (IPP) is determined. In other circumstances, if the missing person does not have a phone, the IPP starts at the point last seen (PLS), i.e., the location where the missing person was seen last [2].
According to the AOUN Association for SAR [21], the police and volunteer groups in Saudi Arabia use methods and tactics in WiSAR that rely on the rescuers’ experience with the surrounding terrain and land cover accumulated from previous tasks, with limited use of spatial technologies and mapping techniques, such as the Google Earth Engine, to explore the spatial nature of the area. Task forces are then sent to different destinations based on this accumulated information. At this point, the ground force (cars) and the air force (gliders and autogiros) start searching in different directions and linking up with each other using wireless communication and GPS to pinpoint their exact location.
What distinguishes WiSAR cases in Saudi Arabia is that about 90% of the SAR tasks are not about searching for the missing person but searching for the missing person’s vehicle, which is usually an easy object to find, compared with searching for a person [21]. This is primarily because the people who go missing in Saudi Arabia are often local tourists, campers, and herders who have cars to move around on dirty roads. These vehicles can sometimes get stuck in the sand or break down, leading to the individuals becoming lost or missing. Furthermore, due to severe climate and terrain conditions, missing persons are usually found close to their cars due to the difficulty in walking and moving around in such an environment [21].
The time a person can spend without water and food is contingent on age, health, and weather conditions, and may range from a few hours to several days. Theoretical search areas exceeding 5,026 km2 for every 40 km walked per day are possible if the missing person was driving a car. In contrast, the distance that a person can travel on foot varies according to the factors mentioned above and the terrain of the area, from a few hundred meters to up to 40 km per day in the case of a nominal walking speed of 5 km/h for 8 h a day [7]. Moreover, in many cases of missing people in the wilderness of Saudi Arabia, especially those during summer, the persons involved have died within 24 h of them being reported missing. In some of these cases, the direct distance between the location of the deceased person and the nearest safe place was not more than 2–3 km [21], suggesting that WiSAR operations have often been carried out without proper geospatial planning in advance.
1.2 GIS methods in WiSAR
In recent years, GIS has become highly significant in the planning of WiSAR missions [2,7,15,16,17,22]. In some cases, GIS technologies are too complex to be used in a short-term operations search because the many elements involved, including cost, training, data, and map sharing, can impact operations [5]. However, the ring model, the mobility model, and network analysis are the three most common methods for implementing GIS in WiSAR missions. In this study, because of a shortage of data sources (such as unpaved roads and hikers’ trail information), only the ring and mobility models were considered.
1.2.1 Ring model
The ring model is the most popular technique for creating probability of area (POA) maps. It is often known as the Euclidean distance model [7]. It is a primary geostatistical analysis method based on Koopman’s mathematical search theory [1,23]. Euclidean distance is used in the ring model, starting at the ring radius representing the IPP, where PLS and LKP generally exist [2]. From a 5% possibility of finding the missing person in the first ring to a 95% chance in the final ring, each successive ring signifies an increasing likelihood of finding the person [22]. The approach is novel because it prioritizes the search area, giving greater weight to closer rings. In addition, the ring model can be implemented with little effort using any GIS software or even a paper map [7]. However, the ring model is also limited because it relies on projected data to run in GIS software, which may be easily misunderstood when working with GPS data in the field. It ignores additional inputs such as geography and land use distribution.
1.2.2 Mobility model
The International Search and Rescue Incident Database defines mobility as “the amount of time the subject was moving.” This is the time a person spends walking away from the IPP [16], which is an essential factor in WiSAR missions. The mobility model is a method to make POA maps. It is similar to the cost–distance analysis often used for building infrastructure, wildlife habitats, and studying anthropology [7,24]. The mobility model uses raster data, such as the digital elevation model (DEM), land use, and land cover data, which are calculated using map algebra to make speed and resistance raster layers. These layers show where a missing person could be by leaving out areas where they cannot go because of a steep slope or the way the land is covered [25]. Unlike the ring model, this method considers different factors when determining the location of a missing object or person. However, this method is very subjective because it mostly depends on human decisions when calculating inputs [7].
1.3 Study area
The Al-Quwayiyah Governorate (Figure 1) is a large region in the central Saudi province of Arriyadh. It covers an area of 51,725 km2. The Al-Quwayiyah Governorate alone occupies about 14% of Arriyadh Province, which makes it the sixth-largest governorate in Saudi Arabia. Despite the vast area of the Al-Quwayiyah Governorate, the total population is less than 89,544. Topographically, the Al-Quwayiyah Governorate is a desert area with only a few urban regions along the Arriyadh–Makkah Highway, which passes through the area at an elevation of around 606–1,505 m. In the summer, temperatures exceed 48°C, and annual rainfall is less than 100 mm. Such severe climate and topographic conditions are also found in other countries around the world, and it is almost impossible for people to survive in such conditions anywhere without adequate water, food, and shelter. It is therefore especially important to address the problems associated with SAR in such environments, as missing persons in such places often fail to survive long enough before they are found, unlike in other areas, where missing individuals can survive for longer due to an abundance of resources necessary for life.
Study area. Al-Quwayiyah Governorate.
2 Materials and method
Though a missing person’s chances of survival drop precipitously over time [7], there are still two crucial considerations to bear in mind:
the POA
the probability of detection (POD).
Either of these may be modified to increase the probability of success (POS), which is calculated as follows [26]:
For instance, GIS methods can aid in increasing POS by utilizing models such as ring and mobility to improve the estimation of POD [16]. Alternatively, the area of search may be reduced to increase POS [7]. However, we sought in this study to contribute more than just lowering the POA. Instead, we used the mobility model and cost–distance analysis to obtain a better understanding of where the missing person might be going.
To this end, we carried out a hypothetical-case study of a missing person in his late fifties driving a 4 × 4 car. This matches the profile of tens of missing persons in Saudi Arabia each year. The PLS of the missing person was at the coordinates of 23°11′38″N, 44°41′53″E in the central area of the Al-Quwayiyah Governorate; at this location, the IPP was determined to be at the same point.
2.1 Ring model
We used multiple ring buffer (analysis) in ArcMap 10.8 to design six separate buffer zones around the IPP, assuming that the missing individual was traveling away from the PLS in his 4 × 4 vehicle at an average speed of roughly 30 km/h before the vehicle broke down. Using Euclidean distance, we were able to divide the search area into concentric rings, with a 15% chance of finding the missing person within a 30 km distance from the IPP (an area of 2,827 km2) and a 90% chance within a 180 km distance from the IPP (a total of 101,787 km2) (Figure 2).
Ring model at the IPP.
2.2 Mobility model
The mobility model used in this study serves two primary purposes:
decreasing the POA
estimating the exact POD that represents the possible final resting place of the missing person.
This approach is built on the assumption that a missing individual will head straight for the next hamlet bypassing any complex terrain such as steep inclines. Using the United States Geological Survey (USGS) 30m DEM (Table 1), we divided the research region into nine categories (Table 2), ranging from easily traversable locations with surfaces of 0–2° to completely impassable areas, such as mountains and cliffs, with slopes of 30–70°.
Used data characteristics
| Data | Source | Type | Resolution |
|---|---|---|---|
| DEM | USGS | Raster datasets | 30 m |
| Saudi Arabia province/governorate boundaries | General Authority for Statistics (SA) | Vector datasets | — |
Slope classes used in the mobility model
| Slope | New class | Accessibility |
|---|---|---|
| 0–2.1 | 1 | Easy |
| 2.1–3.8 | 2 | |
| 3.8–5.8 | 3 | |
| 5.8–8.4 | 4 | |
| 8.4–11.9 | 5 | |
| 11.9–16.3 | 6 | Moderate |
| 16.3–22.0 | 7 | |
| 22.0–30.0 | 8 | Almost impossible |
| 30.0–70.6 | 9 |
After categorizing the slopes of the study area into nine distinct classes, cost–distance analysis was used to determine the least-cumulative-cost distance for each cell, as well as backlink raster analysis to determine the direction and identify the next neighboring cell (the succeeding cell) along the least-cumulative-cost path from a cell to its least-cost source. Then, cost pathways were computed from the IPP to the five closest villages in different directions where the missing individual might be found (Figure 3). The five villages are Halban in the northwest, Ruwaydah in the north, Umm Jadder in the east, Al Hafirah in the southeast, Moses in the south, and Qiran in the southwest; these are the places where the missing person might have gone.
Mobility model calculations and findings: (a) slope classes, (b) cost distance, (c) directions-backlink, and (d) POD findings.
3 Results and discussion
Unlike the ring model, the mobility model is dependent on information gathered from relatives about a missing person’s possible location, so POS is boosted by lowering POA and by determining the exact POD, thereby saving time and energy (a similar result was found in a previous study [7]). In this study, the mobility model narrowed the POA down to 335.34 km of unpaved roads where the missing person’s automobile could have broken down, from 101,787 km2 in the ring model. In addition, the mobility model reduced the time it would take the missing individual to drive the car from 21 h to 9 h 51 min, assuming a speed of 30 km per hour (Table 3). Therefore, the mobility model managed to reduce the POA and the possible mobility time of the missing person by almost 50% and determined their exact direction.
Comparison of search areas and mobility times in ring and mobility models
| POA % | POA in ring model (area in km2) | Mobility time in ring model* (h) | POA in mobility model (distance in km) | Mobility time in mobility model* (h:min) |
|---|---|---|---|---|
| 15 | 2827.42 | 1 | 35.39 | 1:10 |
| 30 | 8482.28 | 2 | 52.56 | 1:45 |
| 45 | 14137.15 | 3 | 53.40 | 1:46 |
| 60 | 19792.02 | 4 | 59.63 | 1:58 |
| 75 | 25446.88 | 5 | 62.62 | 2:04 |
| 90 | 31101.75 | 6 | 71.72 | 2:23 |
| Total | 101787.52 | 21 | 335.34 | 9:51 |
*At an average speed of 30 km/h.
Although the ring model has several benefits, including speed and simplicity, it is evident that its results cannot go beyond the job of organizing prospective search areas, as other researchers have pointed out [7,22]. However, as noted above, the mobility model can provide helpful information about the missing person’s last known location and activity, which is an essential factor in WiSAR missions [16]. The advantages and disadvantages of the two models are summarized in Table 4.
Advantages and disadvantages of ring and mobility models*
| Ring model | Mobility model | |
|---|---|---|
| Advantages | Easy analysis | Considers terrain |
| Cheap and fast | Driving speed included | |
| No additional information is necessary | Detects POD | |
| Reduces time and effort in SAR | ||
| Disadvantages | No additional information (terrain, vegetation) included | Many information layers are necessary |
| No linear features (street/trails) included | Equal walking speed regardless of driving up or downhill | |
| Takes a too long time to search each zone | Resolution depends on input data | |
| Expertise and software are needed |
*Adapted from the study of Drexel et al. [16].
4 Conclusion and future research
This research introduces a novel method for determining search areas and trails in Saudi Arabian WiSAR operations. This approach considers the physical and anthropogenic environment of Saudi Arabia and the behavior of missing persons there. In this study, we used a hypothetical-case study set in Saudi Arabia to compare and contrast the strengths of the two most popular models for WiSAR: the ring model and the mobility model. Compared to the ring model’s 101,787 km2, the mobility model’s 335.34 km of probable routes via which the missing person may have traveled significantly reduces the POA, and the time the missing person has to travel decreases from 21 h to 9 h, 51 min. As a result, either model can be used successfully in WiSAR. The mobility model, on the other hand, can identify the POD of a missing person to save time and improve the POA.
Overall, additional research is needed on the patterns of missing-person behavior in Saudi Arabia because these represent a factor in WiSAR operations and in the use of cutting-edge technology such as unmanned aerial vehicles as a primary instrument for locating such individuals. Moreover, research efforts should be directed toward georeferencing missing cases in geodatabase systems and employing various analyses to increase our geographical understanding of these issues.
Acknowledgments
We want to thank the General Directorate of Civil Defense in Saudi Arabia for sharing information. As well as the AOUN Association for Search and Rescue in Saudi Arabia for sharing their knowledge and experience about WiSAR cases in the country.
-
Funding information: The authors extend their appreciation to the Deputyship for Research and Innovation, Ministry of Education in Saudi Arabia, for funding this research work via project no. IFKSUOR3-006-1.
-
Author contributions: F.S.A. and H.A.A. contributed to the study conception and data collection. F.S.A. prepared the material, carried out the analysis, and wrote the first draft of the manuscript. H.A.A. added some text in the section on the literature review and results. F.S.A. and H.A.A. read and approved the final manuscript.
-
Conflict of interest: The authors declare no conflicts of interest.
-
Ethical approval: Not applicable.
-
Data availability statement: Not applicable.
References
[1] Jurecka M, Niedzielski T. A procedure for delineating a search region in the UAV-based SAR activities. Geomatics Nat Hazards Risk. 2017 Jan 1;8(1):53–72.10.1080/19475705.2016.1238853Search in Google Scholar
[2] Ferguson D. GIS for wilderness search and rescue. Washington, D.C. 2008. https://proceedings.esri.com/library/userconf/feduc08/papers/gis_for_wilderness_search_and_rescue.pdf.Search in Google Scholar
[3] Niedzielski T, Jurecka M, Miziński B, Pawul W, Motyl T. First successful rescue of a lost person using the human detection system: A case study from Beskid Niski (SE Poland). Remote Sens. 2021 Jan;13(23):4903.10.3390/rs13234903Search in Google Scholar
[4] Heth CD, Cornell EH. Characteristics of travel by persons lost in albertan wilderness areas. J Environ Psychol. 1998 Sep;18(3):223–35.10.1006/jevp.1998.0093Search in Google Scholar
[5] Koester R. Lost person behavior: A search and rescue guide on where to look - for land, air and water. Charlottesville, Virginia: dbS Productions LLC; 2008.Search in Google Scholar
[6] Adams AL, Schmidt TA, Newgard CD, Federiuk CS, Christie M, Scorvo S, et al. Search is a time-critical event: when search and rescue missions may become futile. Wilderness Environ Med. 2007 Jun;18(2):95–101.10.1580/06-WEME-OR-035R1.1Search in Google Scholar PubMed
[7] Doherty PJ, Guo Q, Doke J, Ferguson D. An analysis of probability of area techniques for missing persons in Yosemite National Park. Appl Geogr. 2014 Feb;47:99–110.10.1016/j.apgeog.2013.11.001Search in Google Scholar
[8] Grissom CK, Thomas F, James B. Medical helicopters in wilderness search and rescue operations. Air Med J. 2006 Jan 1;25(1):18–25.10.1016/j.amj.2005.10.002Search in Google Scholar PubMed
[9] Ausserer J, Moritz E, Stroehle M, Brugger H, Strapazzon G, Rauch S, et al. Physician staffed helicopter emergency medical systems can provide advanced trauma life support in mountainous and remote areas. Injury. 2017 Jan;48(1):20–5.10.1016/j.injury.2016.09.005Search in Google Scholar PubMed
[10] Wing MG, Burnett J, Brungardt J, Dobler D, Cordell V, Sessions J. Search and rescue operations with an unmanned helicopter. Int J Remote Sens Appl. 2016 Jan;6:65.10.14355/ijrsa.2016.06.007Search in Google Scholar
[11] Tilburg CV. First report of using portable unmanned aircraft systems (Drones) for search and rescue. Wilderness Environ Med. 2017 Jun;28(2):116–8.10.1016/j.wem.2016.12.010Search in Google Scholar PubMed
[12] Karaca Y, Cicek M, Tatli O, Sahin A, Pasli S, Beser MF, et al. The potential use of unmanned aircraft systems (drones) in mountain search and rescue operations. Am J Emerg Med. 2018 Apr;36(4):583–8.10.1016/j.ajem.2017.09.025Search in Google Scholar PubMed
[13] Martinez-Alpiste I, Golcarenarenji G, Wang Q, Alcaraz-Calero JM. Search and rescue operation using UAVs: A case study. Expert Syst Appl. 2021 Sep;178:114937.10.1016/j.eswa.2021.114937Search in Google Scholar
[14] van Veelen MJ, Roveri G, Voegele A, Cappello TD, Masè M, Falla M, et al. Drones reduce the treatment-free interval in search and rescue operations with telemedical support – A randomized controlled trial. Am J Emerg Med. 2023 Apr;66:40–4.10.1016/j.ajem.2023.01.020Search in Google Scholar PubMed
[15] Pfau L, Blanford JI. Use of geospatial data and technology for wilderness search and rescue by nonprofit organizations. Prof Geogr. 2018 Jul;70(3):434–42.10.1080/00330124.2018.1432367Search in Google Scholar
[16] Drexel S, Zimmermann-Janschitz S, Koester RJ. Network analysis for search areas in WiSAR operations. Int J Emerg Serv. 2018 Jan;7(3):192–202.10.1108/IJES-02-2017-0005Search in Google Scholar
[17] Kroh P. Identification of landing sites for rescue helicopters in mountains with use of geographic information systems. J Mt Sci. 2020 Feb;17(2):261–70.10.1007/s11629-019-5805-0Search in Google Scholar
[18] Lin L, Goodrich MA. A Bayesian approach to modeling lost person behaviors based on terrain features in Wilderness Search and Rescue. Comput Math Organ Theory. 2010 Sep;16(3):300–23.10.1007/s10588-010-9066-2Search in Google Scholar
[19] Australian Maritime Safety Authority. National search and rescue manual. [Canberra]: Published by AMSA on behalf of the Australian National Search and Rescue Conference; 1997.Search in Google Scholar
[20] General Directorate of Civil Defense in Saudi Arabia. Annual statistical report [Internet]; 2021. https://998.gov.sa/Ar/Pages/default.aspx.Search in Google Scholar
[21] AOUN Association for Search and Rescue. Yearly Report [Internet]; 2021. https://aoun.sa.Search in Google Scholar
[22] Tambassi T, editor. The Philosophy of GIS [Internet]. Cham: Springer International Publishing; 2019 [cited 2023 May 14]. p. 246. (Springer Geography). http://link.springer.com/10.1007/978-3-030-16829-2.Search in Google Scholar
[23] Koopman B. Search and screening: General principles with historical applications. New York: Pergamon; 1980.Search in Google Scholar
[24] Kobler A, Adamic M. Identifying brown bear habitat by a combined GIS and machine learning method. Ecol Model. 2000 Dec;135(2):291–300.10.1016/S0304-3800(00)00384-7Search in Google Scholar
[25] Johnson M. Evaluating the utility of a geographic information systems-based mobility model in search and rescue operations [Unpublished Master Dissertation]. [USA]: University of Southern California; 2016.Search in Google Scholar
[26] Koopman B. Search and screening: General principles with historical applications. Arlington VA, editor. Alexandria, Virginia: Military Operations Research Society; 1999.Search in Google Scholar
© 2023 the author(s), published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.
