Above-Ground Searches

1 Search Strategies

1.1 Introduction

This chapter is primarily concerned with searching, detecting, and locating sources on the ground or above it. In general, only rough guidelines can be given to handlers because weather, terrain, vegetation, and other factors can be highly variable over time and distance and can act independently and in concert to produce a myriad of confounding effects for the search dog team. The handler needs to be continually alert and flexible to changes in the weather, especially wind and physical settings, while searching to thoroughly cover their search area (i.e. to place their dogs downwind and within scenting range of all potential source locations). These skills need to be developed and maintained during training.

Scent is transported by gravity flow of air, convection, and wind, which are the primary processes of scent transport in the atmosphere. These pro­cesses create scent plumes that are thought to behave like smoke plumes from a chimney or camp fire. Scent plume movement from outdoor sources is pri­marily determined by weather, especially wind, and physical settings (ter­rain, vegetation) that create and influence air movement. Plumes move by gravity and buoyant flow of the scent molecules when there is no wind. The use of dogs to efficiently detect and locate explosives, drugs, people, cadav­ers, and other sources requires information on the characteristics of scent plumes and how their movement is influenced by weather, terrain, and vege­tation. However, there is a lack of scientific information on plumes from these sources, especially in outdoor settings. Consequently, SD handlers must use information derived from studies of wind and plumes of other sources (e.g. insects, smoke plumes) to infer the probable behavior of scent plumes from the sources they seek.

1.2 Searching and Detecting

Evolution has produced dogs with naturally efficient strategies for searching and locating scent sources. A moving dog obviously uses its nose to detect the presence of a scent plume. Moving directly upwind is not efficient since the dog can detect scent only in a narrow corridor close to its line of travel as shown in Figure 1. An exception to this occurs when air is channeled into a narrow valley, especially at night when the cool side slopes channel cold air with scent into a plume at the valley bottom. Trained hunting dogs and SDs, when not under the influence of their handlers, will search naturally by mov­ing directly crosswind (quartering the wind) with their heads canted slightly upwind. With this method, dogs can detect scent upwind from as far as con­ditions allow, which maximizes the area searched from a narrow corridor to a wide swath (Figure 1). This is the most efficient method since the greatest area is covered with each pass and fewer passes through the search area are needed. Quartering the wind minimizes the time and energy required to search an area for both the dog and handler, and when properly executed, it results in a high POD. Proper execution requires the handler to have a high level of spatial awareness and changing conditions which must be developed and honed during training.

Quartering the wind is a gridding method. It requires the team to move upwind at the end of a pass through the search area before turning to make another pass. The distance moved upwind (grid spacing or lane width) is left to the judgment of the handlers based on their experience in training and searching with their dogs in similar conditions. The grid/lane width depends on the nature and size of the source, whether it is above or below ground, terrain, and vegetation in the search area; weather (especially wind and air stability); and other factors. Searches for intense above-ground sources

(e.g. recently deceased subjects) may have grid lines separated by hundreds of yards. Searches for small above ground but covered sources (teeth, shell cases, cigarettes) may have grid lines separated by a few yards and may even require cross gridding to reliably detect and locate them.

Figure 1 Comparison of upwind and quartering the wind search strategies. Up arrows are the paths of dog teams A and B. Team A covers a narrow strip (area A) while moving upwind. Team B covers a strip as wide as the dog can detect the scent plume (area B) and extending across the search area. This maxi­mizes the area covered and minimizes the time and energy required by the team to search their area (Osterkamp 2001).

A few advanced handlers use a modified version of the gridding method for searching (quartering upwind, Figure 2) that is used by insects, fish, birds, and dogs when they have detected a source (DeBose and Nevitt 2008). The dogs are taught to move back and forth in front of them (left to right and right to left) using hand signals, voice and/or whistles, while they walk nearly straight upwind (Osterkamp 2001, 2002, 2003). It is a fast and effi­cient method when conditions allow it to be used. Gridding methods can­not always be used, and it is often necessary to modify them to ft existing weather, terrain, and vegetation.

Large area outdoor searches for live or deceased subjects have identi­fied variations and additions to the grid method that can be used for any large above ground sources. These include hasty searches and hilly terrain searches (along ridges and in valleys and drainages). Hasty searches are often the first method employed by the handler. These are used when limited time is available to conduct the search, the required POD is low, or to delineate the boundaries of the search area prior to conducting a more thorough search. They often follow natural or man-made travel paths such as trails, roads, for­est and field boundaries, fences, ridges, or streams. Large slopes that are not too steep can be searched along contour lines working from upslope down­ward, if possible, to conserve the energy of the team.

1.3 Locating Sources

Dogs use the same natural strategy that insects, fish, and birds use to locate a source (Figure 2). If the plume is not continuous, they quarter the plume upwind which involves moving across the plume at an angle upwind. When they pass through the edge of the plume, they turn into the wind and go back across it until they re-enter the plume and again quarter upwind. The process is repeated as they advance upwind toward the source. When the plume becomes continuous, they move directly upwind to the source. Continuous plumes usually occur close to the source and are also associated with laminar flow, turbulent fow over relatively smooth surfaces and chan­neling. Very low velocity winds may produce a meandering scent plume that varies in direction, confusing the direction to the scent source. Strong, swirl­ing, gusty winds may produce gusts of scent that also confuse the direction to the source.

Figure 2 When a source is detected, insects, fish, and birds quarter upwind to locate the source and dogs use the same strategy.

While all the details and implications of this method are not completely clear, it appears that dogs use their wet noses to determine the direction of the wind, then quarter across wind to detect the scent plume and quarter upwind to follow the plume to the source either instinctively or as a learned behavior. Many dogs with experience in hunting, foraging, and searching use the wind naturally in this way. Puppies and young dogs can be readily taught to quarter by the handler (Osterkamp 2001, 2002, 2003). Teaching a dog to quarter at the direction of the handler and, where possible, using this strategy in training and searching exploits the natural behavior of the dog when searching for scent and conserves the energy of the team.

A search dog that is trained to quarter out from the handler in a good search pattern, can be controlled at a distance, and can be directed from a distance to search in a grid/lane or specific areas will be of great value to the team effort (Osterkamp 2001). Some agencies (e.g. FEMA, US military) require dogs to perform these behaviors as part of their training and certifi­cations and bird hunters spend many hours teaching dogs these skills.

1.4 Calm and Extreme Turbulence

Calm and extremely turbulent conditions require special methods. Completely calm air is rare; there is usually a slight drift of air that may change direction randomly and repeatedly. This can move and spread scent in all directions from a source so when the dog encounters a meandering plume it may not be able to determine the direction to the source. Low wind speeds can also produce meandering scent plumes. When scent is encoun­tered under these conditions, the dog may start to search randomly, circling and looping while obviously in scent. The handler needs to recognize this behavior and develop a strategy to help the dog. One possible strategy is to search in a spiral out from the point where the scent was detected, and another is to grid and/or cross grid the area. If the search area is small, it may be helpful to place the dog on leash and conduct a precise grid search to find the source. If the source is not found, it may be necessary to wait for wind to develop or to return at a time when there is wind.

Extremely turbulent conditions have the air repeatedly swirling and gusting in different directions, especially when there are large objects in the flow (trees, rough terrain, buildings). These objects create turbulence that fragments and redirects the scent plume and sends scent bearing gusts in all directions. If the dog detects scent and determines the wind direction from a gust, it may not be the direction to the source. Under these conditions, the dog may detect scent from one direction and then another and become con­fused when trying to move in the direction of the source. One strategy is for the handler to stop, try to determine the prevailing wind direction and then quarter upwind. Spiral and grid searches are options as well as returning when the wind is less turbulent.

These considerations indicate that search dogs can be expected to have difficulties locating scent sources when the wind is nearly calm and when it is extremely turbulent. Therefore, it can be hypothesized that there may be a range of optimum wind conditions at some intermediate velocity where they will be most successful locating a source. It is likely that the size and intensity of the source and the environmental setting (e.g. field or forest) would also be important parameters. There is support for this hypothesis from a study of skunks, foxes, and raccoons (Ruzicka and Conover 2011) that showed these predators were most active when wind speeds were roughly 5 to 10 mph in a field-like setting. This indicates the optimum wind speed for predators searching in a field-like setting is 5 to 10 mph. Optimum wind speeds may be different for SDs but there is no information available.

Search methods can determine whether the dog will get access to the scent plume. Failure to access the plume is often a result of handler error and a common cause of failure for SDs. An efficient method for searching is to first perform a hasty, free search (of leash) around and through the search area and follow with gridding if nothing is found.

2 Special Effects

2.1 Slope, Valley, and Prevailing Winds

Slope, valley, and prevailing winds can interact to develop a myriad of wind conditions that differ depending on site conditions (Schroeder and Buck 1970). For example, upslope and up valley winds can occur in one part of a valley and downslope and down valley winds in another. Upslope winds usually dominate the ridges and saddles, while upslope and up valley winds combine to define wind speeds and directions at the lower elevations.

With strong heating, late afternoon upslope winds in mountain topogra­phy tend to hold weaker prevailing winds above the ridge tops so that surface winds are primarily those associated with thermal convection (Figure 3). Later in the day, as the upslope winds weaken, the onset of downslope flow lowers the prevailing wind level back onto the slopes and ridge tops. When inversions develop in the valleys, lower slope and valley winds are not dis­turbed by the prevailing winds. Below the inversion, winds are cool, gentle, downslope and down valley. Above the inversion, the prevailing wind on upper slopes and ridges is generally warmer, stronger, and more turbulent.

Figure 3 (a) Late afternoon heating can hold prevailing winds above the ridge tops. If the winds are cool and dense, they may blow up and down slopes when crossing wide valleys and from ridge to ridge when crossing narrow valleys. (b) When inversions form in the valleys, slope and valley winds below the inversion are undisturbed by prevailing winds above. (From Schroeder and Buck 1970.)

Morning upslope winds begin first on east facing slopes just after the sun strikes the slopes and tend to flow directly upslope and up minor side drainages (Schroeder and Buck 1970). As the up valley winds increase during the day, the upslope winds turn more up valley (Figure 4). Daily upslope wind speeds are maximum on south and southwest facing slopes because they receive more sunlight and are minimum on north facing slopes. On densely forested slopes, upslope winds may exist above the tree canopy with a downslope flow in the shaded cooler trunk space below. Where there is dense undergrowth in the trunk space, this downslope flow can be retarded, diverted, or halted.

Prevailing winds in the same direction as slope and valley winds rein­force them, and when opposite to them reduce, cancel, or reverse them. This is especially true for the usually weak downslope and down valley winds at night. Prevailing winds that set up recirculation zones in the valley can enhance, retard, destroy, and redirect slope and valley winds (Figure 5) and cause them to be periodic. Periodic winds can be especially difficult to inter­pret because the source may be in the recirculation zone or the scent may be carried from another direction by the prevailing wind.

Combined slope, valley, and prevailing winds create especially difficult conditions for SD teams because these winds vary with both location and time. Handlers must be continually alert to wind direction and how it influ­ences coverage of the search area. It may be necessary to modify their search plan or to return to parts of the search area to cover missed areas.

Figure 4 Upslope winds are directly upslope initially but are turned to a more up valley direction as the up valley wind increases during the day. (From Schroeder and Buck 1970.)

Figure 5 Prevailing wind moving into a valley divides at multiple points and creates eddies in a horizontal recirculation zone behind a low ridge (right of cen­ter). This wind sweeping over the top of the ridge and the two peaks can create eddies in vertical recirculation zones on their lee sides and channel wind into the valley from the right and left sides. (From Schroeder and Buck 1970.)

2.2 Scent Collectors

SDs often show interest in objects (scent collectors) that protrude into the air­flow downwind of scent sources and in recirculation zones. Scent collectors that consist of natural materials such as vegetation (bunches of grass, weeds, bushes, trees), soil, and rock have surfaces that are especially attractive to scent molecules. This allows scent from a plume to collect on them in quanti­ties that dogs can detect. An example is the rough bark on the downwind side of a tree where scent accumulates in the recirculating eddies from an upwind source (Figure 15 - Scent and Wind). Scent collects on the bark or is trapped in the dead air spaces of crevasses in the bark. Figure 6 shows two large trees separated by about 20 yards with explosives placed about 7 f high on the tree in the back­ground. With a light wind toward the tree in the foreground, a dozen EDs that worked this setting alerted or stood upright on the downwind side of this tree, indicating that scent was accumulating there. Some dogs were clearly in scent on the upwind side of the tree but moved around to the downwind side where they alerted. When a dog detects scent from a source in a forest, it may show interest or alert on multiple trees while working upwind toward the source, indicating that several trees in the scent plume may accumulate enough scent for them to detect. Rocks, bunches of grass, and bushes that project above surrounding vegetation and lie in the scent plume of a source are other common examples. Cave-like hollows formed by vegetation with the wind blowing into them can also collect enough scent to cause an alert.

Figure 6 Explosives placed 7 ft high on the large tree in the background with the wind toward the tree in the foreground caused EDs to alert upward on this tree on its downwind side (shown), which indicates scent was collecting there.

Scent trapped in dead air spaces may eventually collect on surfaces in these spaces. These could include the dead air space near the ground between stems of grass, in folded leaves, and in the duff on a forest floor. Dead air spaces in corners, in tiny recirculation zones (road curbs), and in the large recirculation zones associated with hills, ridges, rocks, forest edges, and in forest clearings may also trap scent molecules that collect on local vegetation and surfaces. Scent collectors and trapped scent may act as secondary sources as suggested in Figure 6. Young dogs in training and dogs consistently trained on faint scent may alert on these secondary sources, but should be encouraged to fol­low the scent plume to the primary source. If the wind changes from the time the scent collected to when the dog detects it, the dog cannot follow the scent plume to the primary source. Handlers must recognize the dog’s behavior and use their knowledge of weather, terrain, and vegetation and their effects on local scent movement to help the dog find the source. Grid searches and cir­cling or spiraling out from the alert are common strategies.

An article search study in Florida (Mesloh and Mesloh 2006), which used one dog to search for human scented bottle caps placed on the ground sur­face in tall grass, suggested that the best time to search was mid-morning after a few hours of sunlight. It was hypothesized that the scent was trapped and collected near the ground surface overnight and that some uplift (con­vection) was required to bring the scent up and out of the grass to the level of the dog’s nose. While plausible, other factors (wind, humidity, barometric pressure, the dog’s movement) may have been involved. Not enough infor­mation was available to evaluate the hypothesis. For short grass, a forested surface with leaves or needles, or a relatively bare soil surface, the optimum conditions for detection of small articles would be expected to be different, but there is no available information.

2.3 Ponding

Scent ponds are depressions in terrain or vegetation that collect scent. Examples (Figure 7) are ponds or puddles that have gone dry, depressions, a valley blocked by a dense stand of trees, a length of ditch that has relatively dense vegetation on the downslope end, and depressions over old graves. Syrotuck (1972) and Rebmann et al. (2000) also give examples of ponding (pooling) in these depressions. Ponding commonly occurs at night under relatively calm and clear conditions by gravity flow of cold air and scent into the pond or by gravity flow of heavy scent molecules into a pond under calm conditions. The pond may fill partially or completely with scent that remains trapped until there is enough turbulence to flush it out. This turbulence can be mechanical, produced by the wind interacting with objects in the flow, and/or thermal caused by solar heating in the depression. A cloudy day with little wind may allow the scent to remain in the pond all day.

Figure 7 Cold air and scent from a surface or buried source flow downslope at night and fill the depression (pond) with scent. Scent remains in the pond until daytime turbulence removes it.

When dogs enter a pond and detect scent, they usually show a CB while obviously searching for a source. After searching and not finding it in the pond, they may leave the pond and then return to it. This behavior may be repeated a few times, but the dog will eventually give up and leave the area especially if the handler appears to be moving away. It is critical for the han­dler to recognize the behavior and help the dog find the source. If the ground is sloped, even slightly, the source is likely upslope. If there is no discernible slope, spiraling out from the pond margin may be a useful strategy. For a large pond, scent may have ponded in only one part of it.

2.4 Channeling

Anything that channels wind channels the scent it carries. Channels are com­monly formed by terrain (valleys, canyons, streams), vegetation (paths, trails, roads in fields, and forests) and structures (berms, walls, houses, buildings). Channels can be one-sided like the edge of a forest with a field, a steep ridge, or the bank of a lake or river. When the wind direction is at an angle to the channel, wind bearing scent can be diverted into the channel and along it. Figure 8 is an example of channeled scent from a cadaver. Syrotuck (1972) and Rebmann et al. (2000) also show settings where channeling may occur.

Dogs that detect scent in the channel must follow the scent upwind in the channel to find the source. This process may be confounded by discon­tinuities in the scent plume, presence of scent collectors in the channel, flow along an outside curve, or sources that are not in the channel but rather some distance of to the side. In these cases, the handler must correctly interpret the dog’s actions to help the dog locate the source. For a source of to the side of the channel, the handler needs to use the dog to determine where the scent first enters the channel and then attempt to use their knowledge of local scent movement to locate the source. This is difficult, and it may not always be pos­sible to succeed, especially when the scent comes from a distance.

Scent moving with upslope flow on the side of a ridge may be prevented from entering a road on top of the ridge by a stronger wind that channels along the road. Dogs working along the road cannot detect scent from the upslope flow which makes it necessary to work the dog along the edge of the ridge rather than in the road.

Figure 8 CDs and people detected channeled human decomposition scent along a forest road on a wooded ridge with the wind from the northeast.

2.5 Chimney Effect

Schroeder and Buck (1970) noted that the sun shining on the dark surfaces of trees, power poles, vertical surfaces (cliffs, sides of structures), sloping sur­faces (hillsides facing the sun), and other surfaces that are dark compared to the surroundings (ploughed fields, dark crops, island of trees in a field) heats the air in contact with them. This causes the air to rise along the heated sur­face in a kind of “chimney” effect and facilitates the escape of warm air aloft. In the absence of a stronger prevailing wind, this upward convective move­ment draws air from the area near these features upward (Figures 1 and 12), which causes it to rise out of reach of a dog. Trees are a common exam­ple when scent is drawn to them and upward on their exteriors or interiors. Dogs may alert or show interest at the base of the trees or look upward where scent may have collected on the rough bark or the underside of branches. Isolated features on level ground must be searched by gridding with small spacing, possibly just a few times the size of the source, to detect a source within them.

2.6 Ridge to Ridge Scent Transport

Conditions that allow wind to transport scent from ridge top to ridge top (Figure 3) occur when prevailing winds are stronger than up valley con­vective winds, when inversions form in valleys and when downwind ridges are somewhat higher. Winds may or may not pass through the valley bottom in the first case but cannot pass through it when inversions are present. The ridges can range in size from those a thousand feet or higher in mountain­ous terrain and separated by a mile or more to those several yards high in relatively fat terrain and separated by several tens of yards. If the winds do not penetrate the valley, dogs working along a ridge typically show a CB, turn into the wind, start downslope, lose the scent, and return to the ridge top (Figure 3). They may repeat this behavior once or twice, but if the handler continues to move along the ridge, the dog will usually abandon the scent. The handler must recognize this behavior, determine, and correctly interpret local conditions that influence wind direction and devise a way to help the dog find the source. A likely location would be on the adjacent upwind ridge, although this may be influenced by local topography and processes that redi­rect the wind. Handlers can get experience with ridge to ridge flow by setting up this type of problem at short distances during training.

3 Gridding and Detection Distances

3.1 Introduction

SD handlers often conduct searches without the benefit of any formal search management and need simplified methods to select grid/lane widths to attain a desired POD. There is little published information that can be used by han­dlers in the field to choose grid/lane widths for above ground sources except for the empirical stability method of Graham (1994) derived for detecting live persons. For other sources, handlers must rely on anecdotal information and their personal experience.

Searches by SD teams rely on the combined handler and support persons (visual) and canine (scent) searches, and both are strongly dependent on dis­tance. Theoretical calculations beyond the scope of this work show that the distances at which non-detection (misses) occur as well as other quantities are also needed for calculating POD from search theory (Robe and Frost 2002; Chiacchia et al. 2015). The calculated POD depends on factors such as the level of effort, size of the search area, and how easy or hard it is to detect the object of the search. However, this calculated POD based on search theory differs from the POD usually reported (percent of sources in an area that were detected).

There are many factors that influence the calculated POD and detection distances that vary with time and location during a search. The primary fac­tors include the scent flux or intensity (amount of scent from the source), source location, weather, terrain, vegetation, and the amount of effort made by the team. The scent flux depends on the source size, temperature, location, and whether it is contained or not. Weather (wind direction, speed, presence of sunshine, clouds, air temperature, ground temperature, humidity, precipi­tation) influences the ability of SDs to work and detect sources. Interactions of these weather factors with terrain and vegetation control the movement of the scent plume. The exact characteristics (size, shape, location, pattern) of the terrain and vegetation elements are highly site specific in their effects, which makes it difficult to define detection distances and grid/lane spacing in general.

Experience obtained during training is necessary for developing search strategies and choosing grid/lane widths. The effects of conditions that increase scent concentrations (scent collectors, recirculation zones), decrease scent concentrations (turbulence), cause scent to rise (convection), and pre­vent scent from reaching the dog’s level (inversions) can only be learned by experience. Also, detection is not equivalent to success because the dog must still follow the scent plume to the source and examples of conditions that make this impossible are given herein. In these situations, the handler must learn how to help the dog.

Factors associated with individual dogs that influence detection dis­tances include physiological (health, diet, conditioning), psychological (han­dler dog interactions), training, motivation to search, threshold for detection, and presence of unfamiliar distractions (scents, sights, and sounds).

3.2 Examples

Table 3.3 shows the cumulative POD by air stability class and distance from the source for a dog/handler team searching for a live subject in areas typical of the eastern US (Graham 1994). Additional examples are given below to illustrate PODs for some types of searches and conditions.

Reed et al. (2010) conducted controlled search trials in northern California oak woodlands to assess how scat detection rates of two dogs were influenced by the distance of scats from a search line and by variation in six environmental factors. Both dogs detected >75% of scats (75% POD) located within 11 yds of the line and the dogs’ detection rates decreased with increasing distance of scats from the line. Detection rates for the two dogs varied significantly with distance and decreased to about 30% and 40%, respectively, at 27 yds. Among environmental factors, deterioration of scats by precipitation was the most important variable explaining variations in scat detection rates for both dogs.

A detailed study of detection distances for desert tortoises in open shrub desert habitat (Cablk et al. 2008; Nussear et al. 2008), which used two trained dogs, showed that the tortoises could be detected up to about 70 yards. Mean detection distance was about 15 yards. Detection rates were about 70% for relative humidity between 16% and 85%. Wind speeds were up to 20 mph and greater wind speeds resulted in greater detection distances. No rela­tionship was found between detection distances and tortoise characteristics (age, class, sex, or size), environmental conditions (temperature, wind speed, and relative humidity), or other study parameters. Detection distances for knapweed in short vegetative cover on relatively fat ground using three dogs (Goodwin 2010) were up to 68 yds, like the results for desert tortoises.

Cablk and Sagebiel (2011) studied the ability of three certified CDs to detect teeth in Nevada. Handlers were advised that the research trials would focus only on teeth as targets, could train for >2 months prior to the tri­als, and were required to keep training records. The site of the trials was fat with a pine overstory and a shrub/scrub understory consisting of mountain mahogany, sagebrush, and native grasses. Ten teeth were randomly placed in plots (10.9 × 10.9 yds) on the ground surface or partially buried and were not visible to the handler. The plots were cross gridded with six grid lines in each direction placed at 5 f 6 in spacing. The dogs were worked along the grid lines on leashes that allowed them to cover one half of the grid spacing, 2 f 9 in. The teeth recovery rates for the three dogs using cross gridding were 78%, 61%, and 20%. The results showed that dog teams could recover indi­vidual human teeth in the field setting with high precision, that the team’s capability varied significantly, and that training records supported a team’s expected field performance.

Two dogs were used in a study to find brown tree snakes in Guam in a dense tropical forest area about 44 × 44 yds in size during early morning hours only (Savidge et al. 2011). Detection rates for the two teams were 44% and 26% for an average of 35%. Detection distances as indicated by a CB or alert ranged up to 13 yds. About a third of these CBs/alerts were within a yard of the snake. The results were for snakes mainly in trees at an aver­age elevation of 3 yds. Temperatures ranged from 73 to 93°F and humidity ranged from 68 to 100% with a mean of 89%. Wind speeds ranged from 0 to 2 on the Beaufort scale with a mode <0.6 mph (i.e. calm conditions). Success increased with increasing average humidity and decreasing average wind speed. Low wind speed may have allowed scent to pool at ground level in dense vegetation near the snake which enhanced detection.

Relatively long detection distances were found for detecting feces from right whales in the open ocean (Rolland et al. 2006). The dogs were worked downwind of a pod of whales or areas where whales had been previously sighted. When a dog detected scent, the boat was steered in the direction indicated by it or in upwind transects perpendicular to the wind (gridding). Feces were detected at distances that ranged up to 1.2 miles. Finding the feces was not difficult because of their size, good weather, and the absence of any significant thermal or mechanical turbulence over open water. These condi­tions would be expected to produce relatively undisturbed scent plumes that could be followed upwind to the source.

This study also compared the abilities of humans and dogs to detect the feces. Humans detected seven samples at 61 to 393 yds while the dogs detected the same samples at 164 to 616 yds.

The above examples are not adequate to define grid/lane spacings for desired PODs. They do suggest grid/lane spacings that may result in detec­tion of similar sources under similar conditions and, in some examples, the potential PODs. They also suggest that grid/lane spacing should be much less than maximum detection distances. However, these studies were all limited, and the results are likely to change when more sources, dogs, and a wider range of conditions and study parameters are included.

4 Distant Alerts

4.1 The Problem

Bryson (1984) noted that their dogs consistently alerted on searchers up to 0.6 miles in mountainous terrain. A distant alert, while not precisely defined, is taken to be an alert or CB offered by a dog in response to the scent from a source that is far away. There is no widely accepted definition of the distance. It could range from a few tens of yards for small or buried sources to much more than a mile depending on the size and intensity of the source, weather, terrain, vegetation, and other factors. Distant alerts are often distinguished by the fact that it is often difficult or impossible for the dog to follow the scent plume to the source. Some of the conditions that favor long distance trans­port of scent and distant alerts include fanning, looping, large eddies and wind gusts, sweeps and openings in forest canopies, and ridge to ridge scent transport. Large areas of terrain that are fat, gently sloping, open, or water bodies also favor long distance transport. Humans can experience “distant alerts” from large intense sources such as above ground decomposing cadav­ers. Distant alerts by CDs of a mile or more have been reported in searches for these cadavers and for live subjects (Bryson 1984; Palman 2011; Irwin 2008; McMahon 2014). There does not appear to be any reports of distant alerts by other types of SDs. Wolves have been reported to detect live prey at 1.5 miles (Mech 1970) which is similar to other predators (jackals, hyenas).

With fanning, lack of vertical mixing allows the scent plumes to be transported without much change in concentration at the top of an inversion so that dogs could detect scent from a distance where it contacts elevated ter­rain by working them at this elevation. Eddies and wind gusts much larger than the scent plume can also transport the plume or large segments of it with little change in concentration. However, once these contact the ground it is common for weather, terrain, vegetation, and other intervening turbu­lent processes to modify the scent plume (dilute, fragment, redirect, elevate) and generally make it difficult or impossible for dogs to follow the plume to the source. This is also true for sweeps and openings in forests that bring wind bearing scent down below the canopy. When the dog cannot follow the plume to the source, it then becomes a problem for the handler/IC to solve. Knowledge of scent plume movement can help them locate the source or nar­row its location so that there is a greater chance of finding it.

Distant alerts are an opportunity for the team to find the source. Difficulties in locating a distant source depend on the distance, number, and characteristics of intervening factors that influence scent plume movement. For example, daytime upslope and up valley flow can redirect scent before it is picked up by prevailing winds. Channeling by valleys, canyons, ridges, rivers, forest edges, trails, and roads can move scent significant distances at angles to the prevailing winds. Forests with a continuous canopy may pre­vent scent from reaching the ground at the dog’s level except in openings and places where sweeps occur. These and other factors may make it difficult to find distant sources by searching upwind. Distant alerts are easier to resolve in sparsely vegetated terrain that often occurs in arid regions, at higher eleva­tions, and in relatively fat or gently rolling terrain.

Distant alerts may involve all the processes that influence scent plume movement such as scent collectors, ponding, channeling, chimney effects, up and downslope flows, thermals, gusting, looping, sweeps and ejections, forest openings, ridge to ridge flow, and recirculation zones. These processes may act independently or in concert to produce scent conditions and concentrations that dogs can detect at a distance. Scent may concentrate on scent collectors in recirculation zones, in the lee of steep changes in terrain, and at forest edges. Scent from a distant source can be brought down to ground level by looping, sweeps, in forest openings, and when wind bearing scent contacts elevated ground. A common and particularly difficult situation in hilly and forested terrain occurs when upslope winds bearing scent encounter stronger prevail­ing winds that move the scent at a significant angle to the upslope winds.

Scent in a forest can cause dogs to stand on their hind legs or on trees looking up, move around obviously in scent, self-reward (e.g. pick up a stick), whine, or to stand or sit looking upward or at the handler, and appear to be frustrated. A recirculating eddy on the upwind or lee side of a hill or ridge from a source that is far upwind may cause the dog to follow the scent up or down the slope until they lose the scent and then to return to the start. They may repeat this behavior or just move away if the handler moves. Some dogs just stop and stare upwind when detecting the faint scent of a distant source and then may look at the handler. In these cases, the handler should take a few steps toward the dog, an action that seems to give them “permission” to follow the scent plume. Other observed behaviors include chomping grass (nervously), head turns, and stopping to carefully examine common scent collectors (e.g. clumps of grass, bushes, rocks, lee side of trees).

Scent over water that is colder than the air (common during the day over water) is either on the surface or in a thin layer of air above it so that it is useful to get the dogs as close as possible to the water surface. When scent is present, dogs may have their noses almost in contact with the water or even have their lower jaws in it. If the water is flowing, scent collects on anything protruding from the water (logs, brush, sticks, leaves) and on vegetation on the outside of bends. As noted previously, a stream may also channel scent from a distance either on the water surface or just above it.

4.2 Locating the Source

CDs that detect scent from a faint source will give a CB/alert that the handler should recognize. However, handlers and dogs both require special training to locate a distant source and only a general outline is given here. Handlers must become more sensitive to often subtle behaviors that show the dog is in scent; to the effects of microscale weather, terrain, and vegetation on scenting conditions; and to the presence of special effects that influence the move­ment of scent. They must also develop the habit of continually analyzing dog behavior and scenting conditions and adjust their search tactics as necessary. Skill at recording and interpreting observations and data is required. Dogs must gain the experience and confidence they need to follow faint scent for long distances or search for isolated scent on scent collectors. One procedure is to train the dog in trailing before training any other detector skill, espe­cially one that requires nose to the ground searching. This procedure is used to train mine detector dogs (GICHD 2004). Dogs trained to trail on leash or of leash and then cross-trained in cadaver search can be trained to fol­low discontinuous scent from a distance upwind from both live subjects and cadavers (McMahon 2014). However, it is difficult to apply unless the terrain is relatively open which is characteristic of high altitudes and arid regions.

As in all dog training, simplify the initial conditions to where you know the dog will be successful. Start with short distances, simple terrain, and sparse vegetation. Increase the variables (distance, terrain, vegetation, time) one at a time in small steps that allow the dog to always be successful. Work each step with differing conditions characteristic of your search areas and where special effects may influence scent movement. The goal of this training is to improve the ability of the team to follow scent plumes, continuous and discontinuous, and to locate the paths of scent in scent collectors, when loop­ing conditions are present, and when quartering upwind to detect puffs of scent. When there is no scent or not enough scent available for the team to follow, handlers must use their experience to help their dog.

If a well-trained SD gives a CB/alert but cannot locate or follow a scent plume to the source, it should initially be considered a distant alert. The probability of locating a source that produces a distant alert increases when the handler records data and observations necessary to interpret it. Multiple CB/alerts from other dogs in the area or the same dog can also increase it, assuming the handlers are competent at recognizing them.

Current technology includes GPS and cell phone tracking capabilities that should be used by all SD teams to provide a recorded track of their path through the search area and a record of the position of any CB/alerts. If a dog indicates scent or alerts in a certain direction, the wind direction at the position of the dog or the direction they are pointing should be recorded immediately since wind can be highly variable, especially in forests. If the dog is clearly in scent but not indicating a particular direction, the local and prevailing wind directions should be recorded.

For distant alerts, the handler or their support should record the usual data for any CB or alert (date, time, map or GPS coordinates, and wind direction) supplemented by observations on weather conditions, terrain, and vegetation and their interpretation of the CB/alert. Weather conditions should include air temperature, wind, percent of cloud cover, precipitation, and length of the handler’s shadow. Wind data should include the wind direction at the level of the dog, at the tops of any trees near the alert, any obvious indicator of wind (gusts, dust, low cloud movement), and the prevailing wind speed and direction. Weather conditions with the prevailing wind should be recorded hourly by IC for the search area but may not always be done. Prevailing wind could be from the above observations or from a nearby weather station such as an airport or city. Cloud cover and the length of the handler’s shadow are needed to provide an estimate of atmospheric stability. Terrain can usually be determined from topo maps but small scale features such as trails, gullies, cliffs, depressions, small ponds or streams, and others may not be apparent. The presence of water bodies and especially flowing water in nearby stream and river channels should be noted. Vegetation characteristics include type, height and density, presence of openings and clearings, and nearby edges of fields and forests. The handler’s interpretation should include the presence of special effects (scent collectors, ponding, channeling, chimney effect, up and downslope flows, gusting, looping, sweeps and ejections, ridge to ridge flow) and any other observations of conditions that may influence scent movement.

There are several strategies that may be useful for interpreting dis­tant alerts. For complex terrain and/or vegetation, one strategy is to try to accumulate as many distant alerts as possible in a first attempt to define the general area from which these alerts are originating and then to search for the source in that area. Their nature of being “distant” often means that the usual search boundaries must be expanded. This can be done with daytime searches by taking advantage of upslope winds on sunny days and prevail­ing winds which can carry scent across ridges, saddles, passes, and hill and mountain tops. If the winds are light and variable in these areas or the areas are forested, the scent plumes may be intermittent. This means that teams moving rapidly may miss the scent plumes, so it is desirable to conduct these searches at a slower pace. Improving POD may also be done by using several teams separated by about 100 yds, moving slowly, one team mov­ing very slowly, or by teams spending more time in saddles, passes, and any place where scent may be channeled or near the tops of hills, mountains, and ridges that intercept the prevailing wind. Distant alerts during daytime searches, when conditions favor upslope flow, may indicate that the source is below the elevation of the alert.

In some searches, it may be possible to box the source (i.e. define the area where the source must be) by using wind from different directions and then concentrate resources in that area. The idea is that alerts with a prevailing north wind usually suggest the source is in that direction. If alerts are then experienced with a south wind but farther to the north than the first alerts, then the source is likely in the area between the two sets of alerts and simi­larly for east and west winds. This method relies on the presence of favor­able winds that may require weeks to occur during which the decomposition scent may become fainter. It is important to conduct as many searches as possible if the subject may still be alive or before decomposition scent of a deceased subject decreases.

Another strategy is to conduct late afternoon to early morning searches in clear, stable weather with relatively calm conditions at the bottom of gullies, valleys, drainages, and slopes, since gravity flow of cool air bearing scent drains downslope in shade and at night. These searches may be more successful than the usual daytime searches because upslope flow in daytime tends to be deep and turbulent and downslope drainage flow tends to be thin, less turbulent and concentrates scent closer to the ground. Downslope drainage flow may start and stop at random times, so teams should pay special attention to the flow, stopping when it stops and continuing after it starts. Alerts in shade during late afternoon to early morning searches in clear stable weather with little or no prevailing wind may indicate that the source is above the elevation of the alert.

4.3 Examples of Distant Alerts

It is impossible to give a definitive interpretation of distant alerts since enough information can never be obtained. The necessary information for the search area would include detailed terrain and vegetation maps, maps of wind, air temperature, cloud cover and air stability at times just before and during each alert, presence of special effects, and other information. The interpretations of distant alerts given below are based on written reports and discussions with dog handlers that conducted the searches or were present at them and on an incomplete knowledge of local scent movement. While inadequate, it is hoped that these examples of distant alerts will help improve the ability of handlers and search managers to find the lost and missing. Additional accounts of dis­tant alerts have been given by Irwin (2008) and McMahon (2014).

4.3.1 Machias River, Maine, Suicidal Subject

Figure 9 illustrates the effects of scent channeling by a river and the use of human and dog alerts to find the subject. The subject’s vehicle was found at the location shown on Figure 9 in November about 5 to 6 weeks after he was last seen, so there was a long time for scent to spread around the area. Teams from MESARD (Maine Search and Rescue Dogs) searched downstream along the river and the team on the west side of the river had several alerts on small hills. The wind was from the east and the weather cold, cloudy, and foggy. The team on the east side did not work far enough downstream to find the subject. At the same time, a team working the water in a canoe did not have any alerts even though they passed very close to the subject who was lying about 10 yds from the 6 to 8 f high banks. However, neutral air stabil­ity, a recirculation zone on the west slope of a ridge to the east, and other fac­tors may have prevented scent from reaching the canoe level where the dog could have detected it.

A few days later, game wardens reported smelling decomposition scent as they walked a high bank on a bend in the river near the subject’s vehicle on a sunny day with wind from the south blowing directly up a long straight portion of the river to the bend. The high bank may have created a recircula­tion zone or pool of scent that was defected to the west and north over the bank where the people could smell it.

Several ideas based on the above facts were developed to guide the search. Lines drawn upwind from the human alerts at the bend and east from the dog alerts and the location of the vehicle on the east side of the river suggested the subject was downriver on the east side somewhere near the intersection of the lines. Lack of an alert by the team in the canoe indicated that the body was not in the water. These considerations suggested that the east bank of the river should be searched to or beyond the intersection of the lines. The subject was found by a MESARD dog team on the east side of the river about two-thirds mile south of the vehicle and about a quarter mile north of the dog alerts. On the first day of the search, it appears that the hills on both sides of the river may have channeled the subject’s scent to the south where it was picked up by the east wind and transported across the river during neutral stability conditions to the dog team on the west side although other interpre­tations are possible.

Figure 9 Distant alerts by a CD and human decomposition scent detected by people illustrate the effects of scent channeling by the river, possible neutral sta­bility, and the use of the alerts to find the subject. (Map and search information courtesy of Deborah Palman, MESARD.)

4.3.2 Waterville Valley, New Hampshire, Alzheimer’s Subject

Figure 10 shows CD alerts (including indications and interests) and detec­tion of decomposition scent by people over a 12 week period after a woman with Alzheimer’s disease was reported missing. The woman’s body was found by a MESARD searcher traveling home on an ATV on a cross country ski trail after a day of searching with her dog. It was located 25 yds of the trail that led back to the woman’s house and over 170 yds uphill from Avalanche Brook.

A tremendous effort was made by dog teams and ground searchers over 12 weeks in the fall after the woman disappeared. They searched along every trail plus all the water courses within 1 mile of the place last seen (PLS). Unfortunately, she was over 1.4 miles from the PLS in a place where foli­age prevented her from being seen. Dog alerts and indications were up to 1.2 miles from the body as shown on Figure 10.

Winds were generally strong from the northwest and would sometimes continue at night. Waterville Valley is bordered to the north and west by peaks and ridges typically 3,000 to 4,000 f in elevation. Two major valleys, one from the west-northwest and the Mad River valley to the north, channel northwest winds into Waterville Valley. This complex setting created highly variable circulating winds and eddies in the search area which distributed scent throughout the valley and made it impossible to reliably determine the location of any scent detected. The setting was further complicated by Avalanche Brook because the stream has a steep gradient, is set well below the level of the surrounding woods, and is in a valley cut down into the ter­rain. This suggested that the valley was shielded from higher level circulating winds so that scent could flow downstream, which was verified by observa­tions of the movement of leaves just above the water surface.

The north facing slope where the body was found was partially protected from the wind by the terrain and shaded by the terrain and forest at a time of low sun angles. It appears that scent flowed downhill from the body to the stream channel, was carried by the air downstream, and accumulated in the bend at the junction with the Mad River, where there were three alerts and decomposition smell detected by people. Interest by a dog to the east of the body was likely a direct result of the northwest winds channeled east up Avalanche Brook. Alerts by two dogs to the north were possibly a result of high level circulation in the valley.

Figure 10 Distant alerts and CB by MESARD and New England SAR canine units and detection of decomposition scent by humans over a 12 week period after the woman went missing. (Map and search information courtesy of Deborah Palman, MESARD.)

Interpretation of the effects of the swirling variable winds in the valley was difficult. However, the three alerts near the junction of the river and brook and the interest of a dog farther to the east suggested that the body was in the drainage of the brook. Working out the details of this assumption eventually led to discovery of the body. There is little that can be done under these extremely complex conditions, but accumulating alerts, cataloging and recording observations (like the leaf movement over the water), and persis­tence in analyzing and searching can lead to finding the source. If it is an option, waiting for different wind conditions may sometimes help.

4.3.3 Clarks Fork River, Wyoming, Missing Person

A man was reported missing on October 1 and his vehicle was located near the Clarks Fork River on October 4. Canine teams from Park County Search & Rescue and Northwest K-9 Search & Recovery searched over the next 2 weeks and had multiple alerts (Figure 11) that eventually led to finding the man’s body on October 22 near the river at the bottom of cliffs where he fell.

Figure 11 Some of the distant alerts by PCSAR (Park County Search & Rescue) after the man went missing. The filled circles, •, are the locations of the alerts and the body. Lines are drawn upwind from the alerts. (Map and search information courtesy of Kris Brock, PCSAR.)

The alert in the upper left quadrant was about 1.4 miles from the body. It appears to be the result of channeling by the wind upstream from the body to where the river turned north and the scent was carried up and out of the canyon. The three alerts in the lower right quadrant are <½ mile from the body. The upper and lower ones may have been the result of channeling by the canyon and terrain near the cliffs. The middle alert is difficult to explain; however, time lapse photographs of clouds in mountains often show extreme variations in wind directions that may have caused it.

5 Summary

The primary methods of scent transport in the atmosphere are by gravity flow of air, convection, and wind. These processes create scent plumes that are thought to behave like smoke plumes from a chimney or camp fire. Scent molecules move by gravity and buoyant flow when there is no wind. The use of dogs to efficiently detect and locate explosives, drugs, people, cadavers, and other sources requires knowledge of the characteristics and movement of their scent plumes.

Dogs search naturally by moving directly crosswind (quartering the wind) which minimizes the time and energy needed to search an area and results in a high POD (Figure 1). When a source is detected, insects, fish, birds, and dogs quarter upwind to locate the source which involves moving across the plume at an angle upwind (Figure 2). When they pass through the edge of the plume, they turn into the wind and proceed back across it in the direction they came from until they re-enter the plume and again quarter upwind advancing them toward the source.

Calm and extremely turbulent conditions require special methods. Under calm conditions, the dog may start to search randomly, circling about. Under turbulent conditions, the dog may detect scent from one gust and then become confused by another when trying to move toward the source. Possible strategies to help the dog include spiral and grid searches and returning when there is wind or it is less turbulent.

Optimum wind speeds in a field are roughly 5 to 10 mph but may be dif­ferent for other terrain and vegetative conditions. An efficient method is to first perform a hasty search around the borders and through the search area and follow with a grid search if nothing is found.

Late afternoon upslope convective heating can hold prevailing winds above the ridge tops (Figure 3). If prevailing winds are cool and dense, they may fol­low slopes when crossing wide valleys but cross from ridge to ridge in narrow valleys. When inversions form in the valleys, slope and valley winds below them are undisturbed by prevailing winds above. On densely forested slopes, upslope winds may exist above the tree canopy with a downslope flow in the shaded cooler trunk space below. Upslope winds are directly upslope initially but turn to a more up valley direction as the up valley wind increases during the day (Figure 4).

Wind in hills and mountains can be channeled by terrain and create verti­cal and horizontal eddies which result in complex search conditions (Figure 5)

Scent collectors consisting of natural materials such as vegetation, soil, and rock have surfaces which are especially attractive to scent molecules. Scent collects on them in quantities that dogs can detect. Examples are the rough bark on the downwind side of a tree and cave-like hollows formed by vegetation with wind blowing into them (Figure 6). Scent collectors can act as secondary sources. Young dogs in training and dogs consistently trained on faint scent may alert on them but handlers should encourage and help them follow the scent plume to the primary source.

Scent ponds are depressions in terrain or vegetation that collect scent (Figure 7). When scent ponds are encountered, it is critical for the handler to recognize the behavior of the dog and help the dog to find the source.

Anything that channels wind channels the scent it carries (Figure 8). Channels are commonly formed by terrain, vegetation, and structures. Channels can be one-sided like the edge of a forest with a field, a ridge, or the bank of a lake or river. Dogs may need help following the scent in the channel and finding where scent entered the channel.

The dark surfaces of trees and any vertical or sloping surfaces in sun­light heat the air in contact with them, which causes the air to rise along the heated surface. This facilitates the escape of warm air aloft. The upward con­vective movement can draw air from near the base of these features upward. Dogs may alert or show interest at the base of trees or look upward where the scent collects on the rough bark or the underside of branches.

Wind can transport scent from ridge top to ridge top when prevailing winds are stronger than up valley convective winds, when inversions form in the valleys, and when downwind ridges are somewhat higher. Dogs working along a ridge show a CB, turn into the wind, start downslope, lose the scent, return to the ridge top, and may repeat the behavior. The handler must recog­nize this behavior and devise a way to help the dog find the source.

Distant alerts can occur more than a mile from a strong source. Locating the source of a distant alert is usually difficult because of the distance, num­ber, and characteristics of intervening factors that influence scent plume movement (Figures 9, 10, and 11). These factors often include scent col­lectors, channeling, up and downslope flows, thermals, gusting, looping, sweeps and ejections, forest openings, ridge to ridge flow, and recirculation zones.

Dog behaviors during a distant alert include looking upward, self-rewarding, whining, staring upwind, biting grass, and others. In dealing with distant alerts, handlers must become more sensitive to often subtle behaviors of their dogs when in scent; to the effects of microscale weather, terrain, and influence the movement of scent. The probability of locating a source from a distant alert increases when the team records data and observations neces­sary to interpret the alert.

Strategies for locating the source include trying to accumulate more alerts, working ridges, saddles and hilltops during the day, working drain­ages and valleys when side slopes are in shade and at night, and boxing the source (i.e. define a smaller area where the source must be) by using winds from different directions.

Several examples of distant alerts are described.

Tom Osterkamp