Modeling
Bison Jump Dynamics at Bonfire Shelter
Modeling
Bison Jump Dynamics at Bonfire Shelter
 |
Free downloadable National Elevation Dataset (NED) data was obtained from the United States Geological Survey (USGS) at http://seamless.usgs.gov/ for the Langtry, TX Quadrangle |
|
Raster data format, cell size 30m X 30m |
|
Data recorded in Geographic Coordinate System (GCS), NAD 83 datum |
|
GRID file contains elevation data at 30m resolution |
|
But project analysis requires higher resolution DEM! |
|
In Summer 2003, the QUEST Archaeological Research Program conducted fieldwork at Bonfire Shelter |
|
Efforts included topographic mapping of site, canyon bottom, and surrounding uplands |
|
Sokkia EDM/Total Station was used |
|
Mapped the X,Y,Z coordinate locations for all data points using the UTM coordinate system |
|
855 points were mapped in order to capture topography of the land |
|
The data cover approx. 1.15 km2 |
|
Data are patchy due to rugged landscape and uneven sampling |
|
The NED DEM was downloaded and relevant files extracted |
|
File opened and viewed using ArcView 8.3 and ArcScene |
|
Click here to view the regional DEM |
|
The DEM was created from 855 data points taken from within and around Bonfire Shelter | ||||
|
Kriging was the first method of interpolation attempted in order to create the DEM surface, though this method proved ineffective because it failed to capture the sharp edge of the cliff (see original DEM) | ||||
|
Additionally, points taken from within the rockshelter had to be removed from the dataset
|
|
The DEM was then created using the inverse distance weighted (IDW) method of spatial interpolation
| ||||||||
|
The coordinate system of the Bonfire DEM was then converted to GCS NAD83 to match with the regional DEM. The two DEMs were overlayed to create the final integrated DEM. | ||||||||
|
Once the accurate, high-resolution DEM was created, detailed analysis of the purported bison jump could be conducted |
Using the regional and Bonfire DEMs, the first goal of the project was to determine likely bison drive lanes given the topography of the landscape. Literature on bison drives and jump sites was referenced, including historic ethnographic accounts from the 18th and 19th centuries during which American bison were extensively hunted by both Native Americans and European Americans. From the literature, a series of topographic features common to most bison jumps emerged:
|
Bison herds were first rounded up, most often in areas with preexisting ideal grazing conditions or on paths that led to water sources (Frison 1978). |
|
Drive lanes, sometimes several kilometers in length, were used to navigate the bison. These drive lanes were broad and flat, providing conditions conducive to getting the animals to move quickly while providing few opportunities for escape. |
|
Drive lanes sometimes incorporated certain natural topographic features which functioned as a barricade along the sides of the drive lane. For example, steep banks may have been used to confine the herd movement. |
|
Bison jump sites sometimes involved the use of cliff edges which remained virtually invisible until the herd is directly on the edge. For example, at the Vore Buffalo Jump site, a large sinkhole was used to jump bison herds. The sinkhole would have been difficult to see until it was too late to divert the path of the herd (Frison 1978). |
|
A series of rock piles were often used to mark the edges of these drive lanes, forming archaeological features known as drive lines. Historically, brush or dried bison dung was also used for this purpose. Drive lines helped to direct the hunters herding the bison and also offered places for other members of the hunting party to hide. These individuals would jump out from their hiding place as the herd approached, keeping the animals from straying from the drive lane (Verbicky-Todd 1984). |
|
Drive lanes were broken into two parts. The first part was used to herd the bison and get them moving in the right direction. Often, at the last moment, a 90 degree turn was incorporated into the lane a short distance from the actual jump point. At this time, the goal was to incite the bison and get them moving at high speeds towards the cliff edge. The drive lines would narrow, forming a funnel which navigated and forced the bison directly over the cliff edge (Frison 1978). |
The first goal of the analysis was to look for the above features on the landscape and attempt to identify the most likely locations for prehistoric drive lanes. Assuming that drive lanes required relatively flat and featureless strips of land for effective bison herding, the regional DEM was analyzed for such features.
|
The slope of the regional DEM was first calculated using the Spatial Analyst extension for ArcView 8.x. Given the wide range of values, the resulting DEM was reclassified into 9 classes. (Click here for reclassified slope DEM) |
|
In order to separate the really flat areas from the less flat areas, the DEM was again reclassified into two categories: flat and non-flat. Classes 1-2 were grouped into the "Flat" category, and Classes 3-9 were grouped into the "Not Flat" category. [Click here for the two-class slope DEM) |
|
While this reclassification helped to eliminate some of the noise in the slope DEM, the layer still needed to be simplified in order to remove subtle landscape features. The goal was to simplify the DEM so that areas dominated by flat cells were visibly apparent. In other words, small areas of minor slope changes, which are arguably insignificant on the broader scale of the landscape, are removed. |
|
A low-pass filter was performed on the DEM in order to reduce the DEM resolution using the neighborhood statistics application in Spatial Analyst. The majority statistic was used, and a neighborhood of 4x4 was selected. This slope DEM was used in the analysis. [Click here for the resulting low-pass filter DEM) |
From the final slope DEM, areas of contiguous flat cells become more apparent. (Click here for map of flat areas). These areas would be the most likely locations for drive lanes.
When driving a herd of bison across the landscape, it can be assumed that people will choose the least rugged areas of the landscape in order to affectively move the bison without any major obstacles. It can therefore be assumed that the drive lanes would be located along least-cost pathways. For this study, least-cost pathways are the routes in which both distance (source to target) and degree of slope are minimized. While the starting point of the bison drive lane remains unknown, the ending location, Bonfire Shelter, is known.
|
A cost distance and cost direction raster were first created using the Cost Weighted function in Spatial Analyst. The notch was identified as the source. |
|
Arbitrary starting points were created (click here for map of locations). |
|
The Shortest Path function in Spatial Analyst was used to calculate the shortest path between the arbitrary bison starting points and the known end point, the notch. The cost distance raster and cost direction raster were used to calculate the least-cost pathways between the points. The resulting map shows a series of probable pathways connecting various locations across the region to Bonfire Shelter. (Click here for the least-cost pathway map). |
Based on the topography of the landscape, several pathways can immediately be ruled out. It can be assumed that no bison drive lane would cross a canyon or a river. Therefore, all southern and western routes can be eliminated. These paths must all cross a canyon to reach the site and are therefore unlikely routes. The remaining paths derive from the eastern or northern portions of the region. (Click here to see the possible paths).
This shows that if the bison herd started at any point in the northern area of the region, which covers approx. 25 km2, then the bison would have been funneled down the same path. Any herd starting in the area directly east of the site would have also been directed to the site along a converging path.
But, when the DEM is overlaid with the slope DEM, it reveals which areas of the region contain large, broad pathways suitable for bison drive lines. The red box surrounds the area containing the most ideal path for a bison drive. (Click here for the map showing the most likely path).
This shows that if bison originated from any point in the entire northern portion of the region (an area of over 25 km2), then they would have most likely been driven down this pathway. This suggests that the entry point into the upland area surrounding the site would have been from the north, rather than any other direction. (Click here for a close-up of the site area)
From this information, the path of the most probable drive lane was predicted. [Click here for the predicted drive lane)
Once the route into the site area had been predicted, it was necessary to predict the end portion of the drive lane. How was the notch approached? What path would have served most effectively as the tail-end of a drive lane?
|
In order to determine areas of natural topographic depression, the hydrology modeling extension for ArcView was used. This extension predicts that natural stream routes of the input DEM. |
|
The resulting layer shows the stream networks for the Bonfire DEM. (Click here for the Stream map) |
From this, several hypothetical path lines leading to the notch were created. These lines were created by connecting preexisting data points and forming polylines. Note that several of these lines originate from areas other than the proposed drive line. These were included to ensure that other possible entry points were considered. [Click here for a map showing the potential path lines).
In order to determine which of the proposed paths offered strategic advantage to the prehistoric hunters, visibility analyses were conducted on the Bonfire DEM. The question of greatest importance was whether or not the edge of the cliff was obscured when approaching the notch from various directions. If so, it may have been easier to drive the herd of bison over the edge. If they couldn't see the edge, then they would be less likely to run off course and avoid the plunge. Line-of-sight can be determined from a set points within a DEM in order to determine which features of the landscape are visible from that location. This is done in ArcView by using the Viewshed function within the Spatial Analyst extension. The resultant output is a binary raster layer containing cells for land that is visible and land that is not visible.
|
Points along polylines were used as locations from which to create the viewsheds. |
|
Individual points were selected from each line so that viewshed layers could be created from each point along the line. This method shows how the view changes as one moves along the line. [Click here to see the selected shot] |
|
The calculations create a viewshed layer showing which areas of the Bonfire DEM are visible from the selected point and which areas are not. [Click here for Viewshed example]. From this point, there is clearly limited visibility of the area. Importantly, it shows that the canyon is not visible. |
|
Viewshed layers were created for all points along the hypothetical lines, creating approx. 100 different viewshed layers. |
|
Each potential path line was evaluated to determine at what point the edge of the cliff became visible. This process determined what approaches may have been most deceptive and thus most useful to the prehistoric hunter. The results of this analysis are explained in the Results portion of this website. |
Another method of determining visibility is through the use of the camera flyby function in ArcScene animation. This simulates movement through the 3-D DEM surface. Flybys help to simulate the perspective of the bison as it moves across the landscape. Flybys were created along the different pathways to simulate bison movement towards the cliff edge. Below are links to the the selected movies created by the flybys. [Click here to show the path lines again.] Sorry--videos may be pixilated!
|
Movie showing flyby along Path 2 (approaching from the east) |
|
Movie showing flyby along Path 3 (approaching from the northeast) |
|
Movie showing flyby along Path 4 (approaching from the south) |
|
Movie showing flyby along Path 5 (approaching from the southeast) |
|
Movie showing flyby along Drive Lane and Path 7 (approaching from the north) |
For a successful bison kill, the orientation of the cliff face must be such that the bison are upwind of the cliff face. If the bison were downwind of the cliff face, then they might smell the hunters before they approached the edge and potentially escape the kill. In a survey of bison jump sites in the Northwestern Plains, it has been found that most bison jump sites are oriented in certain directions (Polk 1979). At jump sites, cliff faces tend to be oriented in a north, northeast, south, or east direction. How does the Bonfire Shelter cliff face compare to these other jump sites?
|
In order to see the cliff face orientation, a blue line parallel to the cliff edge was superimposed on the Bonfire DEM. A red line perpendicular to the blue line was added to the map. |
|
The DEM was then rotated so that the blue line was parallel with the page orientation. A north arrow was placed on the map. The cross was then superimposed on the north arrow to show the orientation of the cliff face. [Click here for cliff face orientation] |
|
The cliff face is facing a north-northwest direction |
Though slightly secondary to the main portion of this analysis, the notch was also explored as a feature in which the bison may have been killed. Given that the majority of the bone from both Bone Bed 2 and Bone Bed 3 are located in the talus cone directly beneath the notch, the possibility remains that the bone eroded down slope after being originally deposited. The talus cone deposit may be a secondary deposit. If this were the case, then the bison kill would have occurred on top of the cliff edge rather within the canyon. Arroyos are a feature used commonly in prehistoric bison hunting. They offer a means of confining the animals so that they can be easily speared by the hunters. Perhaps the notch functioned as an arroyo in which the animals were corralled and killed. (Although I feel this is a unlikely circumstance, I simply wanted to address the possibility).
If this were to have occurred, the notch would have to be large enough to fit good number of bison. Bone Bed 2 contains a minimum of approx. 27 animals, though it is estimated that the deposit contains upwards of 120 animals. Bone Bed 3 contains a minimum of approx. 197 animals, though it is estimated that the deposit contains some 800 animals. Could the notch fit that many animals? In order to figure this out, the volume of the notch had to be calculated.
|
A flat plane was placed over the Bonfire DEM surface. This plane had an elevation of the highest point of the notch (405). |
|
The plane was then clipped to match the exact shape of the notch. [Click here to view the plane above the notch] |
|
The Bonfire DEM was then subtracted from the notch DEM in Raster Calculator of the Spatial Analyst extension. The resulting DEM reflected the difference in elevation between the notch at the highest point and the notch at the lowest points. Essentially, this reflects the area contained within the notch. |
|
In order to calculate volume, the cell values were then multiplied by the cell size (length and width) to produce volumes. |
|
By adding up all the cell volumes, a total volume for the entire notch was calculated. The total notch volume is approx. 835 cubic meters. |
Next, it needed to be determined how many bison could fit into a space of that volume.
|
It was estimated that a bison is approx. 2m x 3m x 1.5 m, or 9 cubic meters. (These estimations are rough! ) |
|
Therefore, the notch could fit approx. 93 bison, very tightly packed. |
The notch could not have contained the maximum amount of bison estimated for Bone Bed 2 or Bone Bed 3, making it even less likely that the notch was used as a kill spot, unless the bone beds represent a series of smaller kills superimposed on one another. Logically, it would not have made sense to bother getting the bison that close to the edge but kill them in the notch (where they are still able to fight back)!
