Of the 83,000 square kilometers within the study area, 78,747 square kilometers were evaluated as part of the probability model; 4350 square kilometers of land in the Upper Snake analytic unit not under Bureau of Land Management control were excluded. (Table 5.1) Systematic inventories have been conducted over approximately 4% of the model, and inventories greater than 640 acres in extent comprise 80% of that area. A total of 5284 sites are reported within the model area, 1819 of them fall within the larger inventory blocks. The following chapter describes respective hydrologic units and presents results of the probability model for each analytic unit within the model area.

PILOT/THOUSAND SPRINGS VALLEY ANALYTIC UNIT

Analytic Unit Description

The Pilot Springs/Thousand Springs analytic unit covers approximately 1.1 million acres (1785 mi²)/.4 million hectares. (4623 km²) It lies in within the northeastern corner of Nevada with a small portion falling within western Utah. The analytic unit lies within the Great Basin region, but with drainage eastward into the Great Salt Lake Desert and Bonneville Basin is considered a sub-unit of the Great Salt Lake hydrographic unit. The upland characterization of this analytic unit drove the decision to analyze it separately from the larger Great Salt Lake analytical unit. (Figure 5.1)

Several small valleys and basins comprise the Pilot/Thousand Springs analytic unit. Thousand Springs Valley and Pilot Springs Valley are the most predominate, covering a major portion of the analytic unit. Toano Draw slopes northward into Thousand Springs Valley and Tecoma Valley extends north from Pilot Springs Valley. A number of relatively low ranges and mountains define the limits and interior of the analytic unit. The Toano Range, Pequop Mountains, Windemere Hills and the Snake Mountains define the southwestern extent of the Pilot/Thousand Springs area. Knoll Mountain and Cedar Mountain mark the hydrographic units northern extent, while the Delano Mountains, Pilot Range, and Leppy Hills form an eastern boundary. Ninemile Mountain and Murdock Mountain separate Toano Draw and Thousand Springs Valley from the eastern valleys. Elevations of the surrounding mountains are relatively low, extending between 2200 and 2700 meters amsl.

All valleys within the Pilot/Thousand Springs analytic unit are externally drained. Thousand Springs Creek Flows north and eastward from Toano Draw and the Snake Range around Ninemile Mountain, then southeasterly through Tecoma Valley into the northwestern uplands of the Great Salt Lake Desert. Pilot Springs Creek drains southward through Pilot Springs Valley then terminates in an extensive sand sheet and dry flat between the southern extent of the Pilot Range and the Leppy Hills. The southern extent of Pilot Springs Valley lies at 1340 meters amsl, just above the Gilbert Shoreline of Lake Bonneville. Toano Draw and Thousand Springs Valley lie at elevations between 1800 and 1600 meters. As it drains through Tecoma Valley, Thousand Springs Creek attains an elevation of 1400 meters amsl.

Vegetation within Pilot/Thousand Springs analytic unit is primarily sagebrush with juniper and juniper/pinyon forest on mountain slopes. Barren areas occur in the southern dunes and flats while desert shrub communities in the lower portions of Pilot Springs Valley and Tecoma Valley.

Analytic Results

Prehistoric Evidential Themes

Of the 4622 square kilometers within the Pilot/Thousand Springs Valley analytic unit approximately 3.5% (164 km²) have been assessed by inventories larger than 640 acres. Four hundred sixty prehistoric sites are reported within those inventories, while 697 sites are reported within the analytic unit as a whole. (Table 5.2) (Figure 5.2)

Sampling within each of the evidential classes is relatively consistent. The juniper steppe vegetation zone is less than 1 square kilometer in extent and has not been sampled. Less than 2% of the juniper/pinyon zone has been inventoried. Areas lying more than 1000 meters from streams and between 3000 and 5000 meters from potential wetlands have also been poorly sampled.

Calculated weights for each evidential theme suggests that a predictive pattern for sites occurs within the desert shrub vegetative community, within 1000 meters of potential wetlands and within piedmont slopes. (Table 5.3) Positive contrasts for slope and distance to springs or streams are inconsistent across analytic runs, and calculated chi-squares suggest a normal distribution of sites within high contrast classes.

Within vegetation evidential themes, desert shrub is the only class with a high contrast. While 10% of the sites lie within the juniper/pinyon zone, distribution of sites is less than anticipated for a positive pattern association. (Figure 5.3)

Potential wetland areas within the Pilot/Thousand Springs analytic unit are relatively few, as reflected by the cumulative extent of those areas lying outside of the 5000 meters buffered area. Areal extents of the three buffered zones are approximately equal, and contrast is uniformly high and strongly predictive for the 0-1000 meter buffer. (Figure 5.4)

When only inventoried areas are considered, the piedmont is the most predictive class for sites. An analytic run using all sites identifies flats as the most predictive class, but by controlling for inventoried space, the number of sites within that area is reduced by almost 42%. By contrast, 78% of all sites within the piedmont landform are accounted for by inventories greater than 640 acres in extent. (Figure 5.5)

Prehistoric Predictive Response

Posterior probabilities generated within the response theme for the Pilot/Thousand Springs Valley analytic unit cluster within three groups with breaks at 0.072 and at 0.044. The prior probability for a normal distributional pattern is 0.029, well below the (Table 5.4) (Figure 5.6) discernable break for low probability. With breaks at those points, well over one-half of the sites within the analytic unit falls within the area of lowest probability. In an attempt to capture additional sites and balance the distribution of sites within each probability zone, breaks were re-drawn at 0.044 and 0.029, just above the intersection with prior probability. (Table 5.5) (Figure 5.7) Results of the model derived from those parameters were tallied and show an almost even distribution of sites within high to moderate and low probability areas. (Table 5.6) The results appear to be biased by a relatively high frequency of sites (training points) lying within the large area of flats and sagebrush, with a corresponding low weight and contrast relating to a normalized distribution. (Figure 5.8)

The response table presents a tally of presence or absence of predictive evidential classes, then recalculates probabilities based upon the area and number of training points within each row of tabulated intersections. If large areas contain a proportional number of sites, probabilities will by definition remain near the prior probability. Likewise, negative weights and negative contrasts will still retain their lower probabilities in the response theme.

The logic behind the response theme is that as intersecting predictive themes overlap, corresponding probabilities validate the predictive relationship within each defined class. Probability based correlations fail when a significant number of the evidential classes exhibit negative contrasts as a result of lower than expected frequencies within disproportionately large areas.

In order to derive a version of the response theme based upon the overlap of predictive evidential classes, evidential themes were reclassified using the binary values assigned to inside or outside pattern within each theme. Using Spatial Analyst^Ò, a new class was calculated by combining each of the predictive layers into a single class. Rows containing 1, for presence within the predictive pattern, were totaled, with results ranging from 0, no overlap present to 3, all three themes intersect. Those results were then re-classified into low (0 overlap), medium (1 class present) and high (2 or 3 classes present) sensitivity zones.

The resulting response presents a better fit of probability layers to the actual site area. Total area varies between the two different response runs due to grid variation within the vegetation evidential theme. (Table 5.7) (Figure 5.9) The distribution of sites within high and medium probability zones comprises more than 70% of the total site area within 55% of the total model area. Ratios of site area to model and inventory area exhibit the same trend, with highest ratios descending significantly from high to low sensitivity zones.

Historic Evidential Themes

Sixty-nine historic sites are recorded within the Pilot/Thousand Springs Valley analytic unit. Of those, only 19 (28%) fall within inventories greater than 640 acres in extent. (Table 5.2) (Figure 5.10) Weights tables for the historic evidential themes, indicate varying positive contrasts within buffered distances to roads and water. Chi-square for roads is significant at the 400 meter buffer of inventoried sites, but is not significant for distance to water. When buffer areas for roads between 0 and 400 meters are combined, chi-square remains significant.

Historic Predictive Response

Since a response theme cannot be created with only one class inside the pattern, the 0-400 meter buffer from water was used as a predictive theme. The resulting grid classifies probability as high or low; area within 400 meters is high, greater that 400 meters is low. (Figure 5.11) Summary tables reflect the expected site distribution with 64% of the analytic unit comprising the low probability zone, while slightly more than 8% of all sites fall within that area. More than 90% of all sites and 85% of inventoried sites fall within the high probability zone. (Table 5.8) (Figure 5.12)

RUBY/LONG VALLEY ANALYTIC UNIT

Analytic Unit Description

The Ruby/Long Valley analytic unit is the hydrographic unit within the GBRI study area. It shares its eastern boundary with the Spring/Steptoe Valley Analytic unit, and its northern extent with the Pilot/Thousand Springs Valley analytic unit. (Figure 5.13) In addition to Ruby and Long valleys, the analytic unit includes Clover Valley and Independence Valley in the north along with Butte Valley and Jakes Valley in southeast. The analytic unit covers approximately 2.6 million acres (4095 mi²)/1.0 million hectares (1060 km²). Bounding ranges of the hydrographic unit include the White Pine Range, Ruby Mountains, and Humboldt Range to the west, Wood Hills and Windemere Hills to the north and the Pequop Mountains, Cherry Creek Range and Egan Mountains to the east. The Maverick Springs Range, Butte Mountains and Medicine Range provide a barrier between Ruby/Long Valley in the western portion of the analytic unit and Butte/Jakes Valley in the east.

Elevations of the Ruby Mountains and Humboldt Range exceed 3000 meters amsl. Northern ranges are lower, averaging 2700 meters amsl while southern bounding ranges and interior ranges extend to 2800 meters amsl. Likewise, valley floor are relatively high averaging 2000 meters in elevation with valley floors between 1850 and 1800 meters.

Hydrologically, each of the valleys within the analytic unit is internally drained. The Franklin River and the Ruby Marshes, consisting of Ruby Lake and Franklin Lake are the major hydrographic features within Ruby Valley. Snow Water Lake serves as a major hydrologic collection point for Clover Valley. Bounding mountains of the remaining valleys provide ample perennial flow, but all terminate in dry flats at the valley bottom. Faulting has produced numerous springs along the steeper eastern escarpment of the bounding and interior mountain ranges.

Vegetation is similar to that in Spring/Steptoe Valley. Limber pine and alpine vegetation occurs on the highest slopes, with juniper/pinyon woodlands on lower more protected slopes. Riparian meadows and wetland habitat dominates the area of perennial lakes and marshes, while sagebrush is the dominant vegetation on the piedmont and upper valley slopes. Lowest portions of the valley floor consist of desert shrub communities while dry flats and valley bottomland is sparsely vegetated.

Analytic Results

Prehistoric Evidential Themes

Of 10,606 square kilometers in the Ruby/Long Valley analytic unit, approximately 927 square kilometers, 8.5% of the total area, have been inventoried. (Figure 5.14) Six hundred thirty-eight sites are reported as part of inventories greater than 640 acres, 973 sites are identified within the entire analytic unit. (Table 5.9)

All analytic classes, were sampled during previous inventories. The Ruby Marshes and surrounding marsh and wetland habitat have been extensively investigated, and unlike most of the analytic units, steeper slopes have been more intensively examined.

Weights of evidence tables identify classes within each evidential theme that lie “inside” the predictive pattern. (Table 5.10) Normalized contrast for meadows are highest in all runs of prehistoric sites. Other vegetation classes display negative or very low positive contrasts. Sagebrush and water have relatively high contrasts when all sites are considered. Meadows within the analytic unit cover less than 2% of the entire area and are considered part of the marsh environment. (Figure 5.15)

Contrasts for distance to springs and streams are consistently high for inventoried areas between 1000 and 2000 meters from that class of water. Buffered areas from 200 to 1000 meters from a water course are consistently sampled, but reveal lower than expected or marginal site frequencies. The lowest contrasts are evident at distances more than 2000 meters from springs and streams. (Figure 5.16)

Proximity to wetlands is highly predictive within the Ruby/Long Valley analytic unit. Highest contrasts are evident within 1000 meters of potential wetland habitat, while areas lying more than 3000 meters from that zone show a negative correlation with normal site distribution. Proximity to wetlands correlates well with vegetation contrasts. (Figure 5.17)

Slopes between 0 and 5 degrees are highly predictive for sites within this analytic unit. Nearly two-thirds of the inventoried area occurs on flat slopes, and positive contrasts are evident on slopes up to 15 degrees. Slopes above 15 degrees uniformly exhibit a negative contrast. (Figure 5.18)

Landform strengthens the relationship of slope as a predictive theme in the Ruby/Long Valley analytic unit. When all sites are considered, both flats and piedmont have a high predictive contrast, while inventoried areas show highest contrasts within the piedmont. Chi-square statistics confirm a non-random distribution of sites on the piedmont. (Figure 5.19)

Prehistoric Predictive Response

Normalized posterior probabilities were used as a means to evaluate tabular results from the response theme generated for the Ruby/Long Valley analytic unit. (Figure 5.20) Prior probability for the response theme was set at 0.0181 and observed breaks within normalized posterior probability were set at 0.182 and 0.0033. (Table 5.11) (Figure 5.21) Highest probabilities for encountering sites occur when evidential classes identified as inside the predictive pattern intersect. Combinations of proximity to wetlands, springs and streams, vegetation, and slope have the highest posterior probabilities, but combinations of three or more evidential themes are also common within the range of moderate probabilities and two or more combinations of evidential themes within the low probability area.

Summary tables for the Ruby/Long Valley analytic unit response theme shows that while 33% of the area is classified as low probability, over 57% of the sites occur within that zone. A similar pattern occurs for inventoried sites. (Table 5.12) The higher frequency of training points within the low probability area creates a normal distribution and with it, a posterior probability lower than the prior distribution.

To realign response themes so that they better correspond with predictive patterns within evidential themes, new probability areas were calculated by totaling binary theme values of each theme. Since all five themes contained predictive classes, additive scores ranged from 0 to 5. Probability classes were grouped into three classes, 0-1 low, 2 medium, and 3-5 high. (Table 5.13) Site area distributions within newly defined probability zones provide a better fit of the data. Site densities are highest in the high probability zones and lowest in low probability areas. Slightly more than 11% of the site area occurs within the low probability zone, which accounts for slightly less than 20% of the analytic unit. (Figure 5.22)

Historic Evidential Themes

One hundred fifty-seven historic sites are reported within the Ruby/Long Valley analytic unit and 81 of these are located within 640 acre or larger inventory units. (Table 5.9) (Figure 5.23) The area within 200 meters of roads, and within 200 meters of streams or springs, revealed the highest contrast within historic evidential themes. (Figure 5.24) (Figure 5.25) Nearly 75% of the inventoried sites (61) lie within 200 meters of roads, while 32% of inventoried sites (26) lie within 200 meters of potential water sources. Distances greater than 200 meters are uniformly less predictive. (Table 5.14)

Historic Predictive Response

Historic response themes generated for the Ruby/Long Valley analytic unit show three possible breaks in the posterior probabilities; 0.014 to 0.008, 0.008 to 0.003 and 0.003 to 0.0009, with a prior probability set at 0.003. (Table 5.15) (Figure 5.26) Lower relative contrasts for proximity of sites to potential water sources create a cluster of training points with posterior probabilities below the prior expected value. Summary tables show that the resulting probability map meets expectations for site density in the low probability area. Seventy-five percent of the analytic unit comprises the low sensitivity zone, and 25% of the sites fall within this area. (Table 5.16) Fourteen training points associated with proximity to water are associated with the low probability zone. The medium probability zone is relatively small and contains a single set of training points associated only with proximity to roads. (Figure 5.27)

Using similar methods employed to clarify the prehistoric predictive response, the historic response theme was re-classified using only the intersection of predictive evidential classes. A more balanced distribution is evident within the reclassification. (Table 5.16) Forty three percent of the analytic unit is classified as high or medium probability, with 84% of the inventoried site area falling into that classification. Sites falling within 200 meters of water previously associated with the low probability zone are now included within the medium classification. (Figure 5.28)

SPRING/STEPTOE VALLEY ANALYTIC UNIT

Analytic Unit Description

The Spring/Steptoe Valley analytic unit lies to the east of the Ruby/Long Valley analytic unit and includes Spring Valley and Steptoe Valley to the south, and Goshute Valley and Antelope Valley to the north. The analytic unit covers approximately 3.4 million acres (5323 mi²)/1.3 million hectares (13787 km²). (Figure 5.29) Topography is typical of north-trending grabens within the Great Basin. High bounding ranges create an orthographic effect on precipitation patterns depositing more moisture along west-facing slopes. Steeply faulted bounding ranges produce numerous springs along eastern pediment slopes.

Steptoe Creek and Duck Creek are the major hydrologic features along the western side of the analytic unit. Both drain northward into Goshute Lake at the southern end of Goshute Valley. Spring Valley Creek is the major drainage in the eastern portion of the analytic unit. It flows north through Spring Valley, then terminates in a large depression and dune field south of Spring Creek Flat. Antelope Valley is relatively dry. Drainages flow from the surrounding mountains to the valley floor. Numerous spring complexes occur within Spring Valley, especially along the toe of western piedmont slopes. Marshes and ponds are present in Steptoe Valley along Steptoe Creek southeast of Ely and west of McGill.

Elevations of the valley floors within the Spring/Steptoe analytic unit are relatively high, ranging from 1900 meters in the south to 1750 meters in the north. The Pequop Mountains and Toano Range bound the hydrographic unit in the north, Cherry Creek and Egan Range on the west and the Snake Range and Ferber Hills to the east. The Schell Creek Range separates Steptoe and Spring Valleys. Wheeler Peak (3952 meters) in the Snake Range is the highest peak within the analytic unit. Mountain elevations are highest in the southern portion of the analytic unit, averaging 3500 meters. Northern ranges average approximately 2500 meters in elevation.

Vegetation is typical of the Great Basin. Highest elevations are dominated by alpine vegetation including limber and bristlecone pine; juniper/pinyon forest covers more temperate lower slopes. The sagebrush zone dominates open pediment slopes and is replaced by desert shrub communities on the lower flats. Depressions and valley bottoms are sparsely vegetated.

Analytic Results

Prehistoric Evidential Themes

Approximately 387 square kilometers (2.8%) of 13,787 square kilometers within the entire Spring/Steptoe Valley analytic unit were inventoried. (Figure 5.30) Eight hundred twenty three sites were identified within the analytic unit; 410 (50%) of those were recorded in inventories larger than 640 acres in extent. (Table 5.17)

Larger surveys have occurred within a sample of all analytic classes within the analytic unit. However, vegetation zones within Great Basin pine, and barren areas along with mountain landforms and slopes greater than 15 degrees are under sampled. Great Basin pine and barren vegetation zones respectively comprises 0.9% and 1.2% of the of the entire model area. Approximately 28% of the analytic unit comprises the mountain landform. Thirty-nine percent of the analytic unit consists of slope greater than 15 degrees.

Evidential theme classes with highest contrasts and correspondingly significant chi-square results were identified as lying “inside” the predictive pattern. (Table 5.18) Highest contrast for vegetation within inventoried sites was associated with sagebrush, and the associated chi-square was also considered significant. Sagebrush retains its high contrast when all sites are considered. A high negative contrast within the juniper/pinyon class along with a high positive contrast for non-sites within that class strongly suggests a lower than expected relationship between sites and the juniper/pinyon zone. (Figure 5.31)

Highest contrast for distance to water is within the 1000 to 2000 meter buffer band. While greater numbers of sites are located between 0 and 1000 meters of streams and springs, their numbers reflect a normal distribution in relation to area. The band between 200 and 400 meters shows the strongest negative contrast with fewer sites than would normally be expected within that region. (Figure 5.32)

Very few potential wetland environments are present within the Spring/Steptoe Valley analytic unit and less than 20% of the analytic unit lies within 5000 meters of this evidential theme. Not surprisingly, highest site contrasts within inventoried areas are within the buffered class that lies more than 5000 meters from potential wetlands. When all sites are considered, areas more than 5000 meters from wetlands have a negative contrast, indicating a weak correlation with sites. Since the results of the weights calculations are inconclusive, no classes within potential wetlands were selected as most predictive for analysis.

Weight calculations for slope in all runs identifies gradient between 0 and 5 degrees as having the highest contrast. Likewise, chi-square for slopes between 0 and 5 degrees meets the critical value for non-random distribution. While sampling discrepancy may marginally effect contrast values, the high frequency of sites both within inventoried samples and when all sites are considered suggests that slopes of 0 to 5 degrees are “inside” the predictive pattern. (Figure 5.33)

Flats exhibit the highest contrast for landforms in all analytical runs. Frequency of sites is highest in this class and suggests a greater than normal distribution of sites. In contrast, the negative contrast for piedmont with adequate sampling suggests that fewer sites than expected occur within that class. (Figure 5.34)

Prehistoric Predictive Response

Normalized posterior probabilities were used as a means to evaluate tabular results from the response theme generated for the Spring/Steptoe Valley analytic Unit. (Figure 5.35) Prior probability for the response theme was set at 0.0122 and observed breaks within normalized posterior probability were set at 0.016 and 0.0095. (Table 5.19) (Figure 5.36) Highest probabilities for encountering sites occur as evidential classes identified as inside the predictive pattern intersect. Three or more combinations of flats and 0 to 5 degree slope, with sagebrush or 1000 to 2000 meters from water have the highest probability scores and the highest frequency of training points. As intersecting conditions decrease, probability becomes moderate and where only a single evidential condition is met, probabilities fall below the prior value of 0.012 and are low. Analysis of the response themes show that by area, the highest proportion of sites fall within high and medium probability zones. (Table 5.20) High probability zones have been more intensively surveyed than zones of medium or low probability and the lowest proportion of inventoried site by area occurs within the medium probability zone. When high and medium probability zones are combined, however, over 80% of the inventoried site area falls within that area. About 66% of all site areas fall within the high and moderate probability zones.

When probabilities are recalculated by composite predictive class, the extent of high and medium probability areas are increased and of low probability area is decreased. (Table 5.20) Twenty-nine percent of the analytic unit lies within the high probability zone and 49% of all sites fall within that area. Forty-four percent of the analytic unit is classified as medium probability and 44% of the sites occur there, while 19% of all site areas fall within the remaining 27% of the analytic unit. Within inventoried site areas, almost 73% of the sites fall within high probability areas, 19% in medium probability zones and 8% in the low probability zone. (Figure 5.37) Over 56% of the inventories have been conducted within areas identified as high probability and most of the site area lies within that zone. Correlations between site density and high probability areas may be biased by sampling within the Spring/Steptoe Valley analytic unit

Historic Evidential Themes

One hundred forty-four historic sites are included in the analysis of the Spring/Steptoe Valley analytic unit. Seventy-nine of these have been identified within inventories greater than 640 acres in extent. (Table 5.17) (Figure 5.38)

Contrasts within buffered classes of roads and water sources are easily discernable. Highest contrasts are evident within 200 meters of existing roads, and weights for non-sites exhibit a negative contrast within the same buffer. (Table 5.21). (Figure 5.39) Other contrasts for buffered distances to roads are either negative or lightly positive. Highest contrast and corresponding chi-square value for water is for the buffered area greater than 1000 meters from water. (Figure 5.40) Areas within 200 meters of roads and more than 1000 meters from water were selected as most predictive for the Spring/Steptoe Valley historic response theme.

Historic Predictive Response

Three apparent breaks are evident in the normalized posterior response for Spring/Steptoe Valley analytic unit historic themes. Breaks lie at 0.006 and 0.0025, with prior probabilities set at 0.0026. (Table 5.22) (Figure 5.41) Highest probabilities occur within 200 meters of roads, or where proximity to roads and distance to water is greater than 1000 meters intersect. No training points fall within a very small zone defined as medium probability. The resulting probability map reflects high and low probabilities. (Figure 5.42) The summary table (Table 5.23) indicates that 26% of the analytic unit is characterized by high probability and 78% of all sites fall within that area. Conversely, the low probability zone covers 75% of the area and contains less than 22% of the site area.

Reclassification of the response theme creates a medium probability zone for the analytic unit. It consists of the area defined by the area within 200 meters of roads, or any area more than 1000 meters from streams and springs. (Figure 5.43) Forty-three percent of the analytic unit falls within the medium probability zone and 66% of all sites lie within that area. (Table 5.23) The extent of low probability area decreases to 51% of the analytic unit and contains 16% of the all historic sites, while 6% of the area and 19% of the sites fall within the high probability zone. Distribution of inventoried sites is slightly higher in high and low probability zones but 80% of the sites still fall within combined high and moderate probability areas.

GREAT SALT LAKE ANALYTIC UNIT

Analytic Unit Description

The Great Salt Lake sub-region covers approximately 10.2 million acres (16,079 mi²)/ 4,164,611 hectares (41,646 km²) within southern Idaho, extreme eastern Nevada, and north central Utah. (Figure 5.44) Six hydrographic basins comprise the Great Salt Lake sub-region within the study area. The majority of hydrographic units contain lakebed deposits derived from the relatively recent Lake Gilbert high stand (10,500 B.P.) and current Great Salt Lake shorelines. Slightly more than 10 meters separate the modern and prehistoric shoreline. That area comprises 18% of the sub-region (Figure 5.45). Periodic fluctuations of the Great Salt Lake create changing environments along lake shorelines. At elevations between 1290 and 1310 meters shorelines encroach upon steeper alluvial slopes of surrounding mountain ranges, effectively eliminating potential river fed marsh areas. (Madsen 1982:208) Six hydrographic sub-regions fall within the Great Salt Lake analytic unit. (Table 5.24)

Curlew Valley

The Curlew Valley hydrographic unit lies in the northeastern portion of the Great Salt Lake analytic unit. The northern half of the sub-region lies within Idaho; the southern half within Utah. Several semi-bolsons comprise this hydrographic unit, all of which slope to the southwest and drain into the Great Salt Lake (Figure 5.46). The hydrographic unit is relatively mountainous and is bounded by the Pleasantville Hills, Samaria Mountains, and the West Hills to the east. The Promontory Mountains and North Promontory Mountains define the southern extent of the Hydrographic unit. Portions of the Raft River Mountains, Black Pine Mountains extend into the Curlew Valley hydrographic unit along its western extent; the Sublett Range and Deep Creek Mountains extend into the hydrographic unit from the north. The Hansel Mountains and North Hansel Mountains extend south through the center of the hydrographic unit. Curlew Valley and Hansel Valley are the predominant lowland features of this hydrographic unit. Deep Creek, flowing through the upper portion of Curlew Valley is the dominant hydrographic feature. The Curlew National Grasslands lie in the upper portion of Curlew Valley, where the Sublett Range, Deep Creek Mountains, Pleasantville Hills, and North Hansel Mountains merge to form a narrow, well-watered basin.

Elevations within the Curlew Valley hydrographic unit range from 2429 meters in the Deep Creek Mountains to 1285 meters at the Great Salt Lake. Curlew Valley averages 1400 meters across its broad southern extent. The upper narrower portion lies at approximately 1580 meters. Scattered pinyon/juniper woodlands with a sagebrush understory occur on the upper slopes of the surrounding and interior mountains. Lowlands range from barren to sparse shadscale communities.

Wetlands are common along the southern periphery of the Curlew hydrographic unit. The Bear River National Wetlands extends along the eastern side of the Promontory Mountains and the Great Salt Lake. Rozel Flat lies west of the Promontory Mountains and the Locomotive Springs State Wildlife Management Area occurs at the delta of Deep Creek and the Great Salt Lake.

Northern Great Salt Lake Desert

The Northern Great Salt Lake Desert hydrographic unit encompasses the northern half of the Great Salt Lake Desert. (Figure 5.47) Only the extreme western edge of the hydrographic unit lies within Nevada. Its eastern edge borders the Great Salt Lake while Interstate 80 arbitrarily bound the southern boundary. The Pilot Range, Goose Creek Mountains and Raft River Mountains define the western and northern periphery, respectively. The Leppy Hills lie in the southwest corner of the hydrographic unit. Elevations range from 1285 meters on the desert floor to 2600 meters in the Pilot Range and 2598 meters at Ingham Peak in the Grouse Creek Mountains. Several small ranges lie scattered about the northern Great Salt Lake Desert rising as high as 2300 meters. The 1295 meter shoreline of Lake Gilbert (10,500 B.P.) roughly defines the edge of the Great Salt Lake Desert sand sheet. The Gilbert shoreline and others marking the Lake Bonneville recession are visible along the western ranges and mountain “islands” throughout the hydrographic basin.

Surrounding mountain slopes drain south and east into the Great Salt Lake Desert. Grouse Creek Valley and Tecoma Valley provide most consistent drainage systems but both terminate at the edge of the desert. Lowlands along the western edge of the Grassy Mountains sustain a viable marsh environment.

Scattered pinyon/juniper woodlands occur in uplands of the highest interior mountains and along the bordering western ranges. As elevation decreases, sagebrush gives way to saltbrush communities while most of the bottomlands are barren.

Great Salt Lake

The Great Salt Lake hydrographic unit lies wholly within the current extent of the Great Salt Lake. (Figure 5.48) Mean elevation for the lake during September 1984 was 1282 meters. Land along the periphery of the Great Salt Lake hydrographic unit, if present, consist of sandy beach or salt flat. The Bear River and Farmington wetlands border the hydrographic unit, but lie outside of its boundaries. Three islands Firemans Island, Antelope Island, and Carrington Island, are prominent topographic features in the southern part of the lake. The Promontory Mountains form a peninsula within the north-central portion of the hydrographic unit and vegetation is sparse to barren.

Rush-Tooele Valleys

Rush and Tooele valleys are typical of the north-south trending valleys commonly associated with the Great Basin. This hydrographic unit lies south of the Great Salt Lake (Figure 5.49) and consists of the Tooele Valley, a broad open flat sloping northward into the Great Salt Lake, and Rush Valley, a larger enclosed basin to the south. The hydrographic unit is bounded by the Stansbury and Onaou mountains to the west, the Sheeprock Mountains and West Tintic Mountains in the south and the Oquirra Mountains to the east. South Mountain (2011 meters) divides Tooele and Rush Valleys. Deseret Peak (3362 meters) in the Stansbury Mountains and Flat Top Mountain (3237 meters) in the Quirra Mountains provide the highest relief along the hydrographic unit boundary. The Tooele Valley continues sloping northward from South Mountain with elevations ranging from 1600 meters to 1285 meters at the Great Salt Lake. Mud flats and sand sheets dominate the northern portion of the Tooele Valley as it juts into the Great Salt Lake. Stansbury Island is a prominent peninsula at the extreme northern end of the valley.

Hydrologically, Rush-Tooele Valley is characterized by steep, well-watered canyons draining into the valley floor from the surrounding ranges. Small wetlands and ponds lie at the 1520 meter elevation below South Mountain in the northern part of Rush Valley. Wetlands also lie at the north end of the Stansbury Mountains and several sloughs grade into the mud flats at the north end of Tooele Valley. Vegetation ranges from limber pine at highest elevations, pinyon/juniper woodland on slopes above mountain pediments, to barren mud flats at lowest elevations.

Skull Valley

The Skull Valley hydrographic unit lies west of Rush-Tooele Valleys, with the crest of the Stansbury Range as a common boundary. (Figure 5.50) The Cedar Mountains rising to 2300 meters, define the hydrographic units western extent, while the Sheeprock Mountains and Davis Mountain trend southeasterly to form a southern boundary. The Lakeside Mountains form a partial northern boundary. Elevation of the valley floor ranges from 1525 meters in the south to 1285 meters in the north where it enters the Great Salt Lake. Extensive mud flats dominate the valley floor below 1300 meters in the northern half of Skull Valley.

Deep canyons along the west slope of the Stansbury Range provide substantial hydrologic inflow, sustaining drainages that eventually flow through the mud flats to the Great Salt Lake. Less competent drainages in the Cedar Mountains characterize the hydrologic regime along the valleys drier west side. Vegetation ranges from barren on the mud flats to desert shrub on the valley floor with pinyon/juniper and sagebrush on the mountain slopes.

Southern Great Salt Lake Desert

This hydrographic unit is the southern extension of the Northern Great Salt Lake Desert hydrographic unit. (Figure 5.51) It shares similar characteristics; dry mud flats and a sand sheet comprise most of the unit, but fewer “islands” occur within its interior. The Goshute Mountains, Ferber Hills, and Deep Creek Range form the western boundary of the Southern Great Salt Lake Desert. The Leppy Hills and Danger Cave mark the extreme northwest corner of the hydrographic unit. To the east, the hydrographic unit boundary is shared with the ranges bordering Skull Valley. Several low, north-trending ranges extend into the Great Salt Lake Desert, creating the hydrographic unit’s southern boundary. Deep Creek Valley, Snake Valley, Fish Springs Flat, and Dugway Valley lie between these southern ranges and drain northward into the desert. White Horse Flat and related badlands lie between the Goshute Mountains and Ferber Hills. Highest elevations occur within the Goshute Mountains, with Goshute Peak rising to 2929 meters. Southern valleys slope northward with highest elevations between 1600 and 1550 meters. The desert floor where at the boundary with Northern Great Salt Lake Desert is 1285 meters. Wildcat Mountain and Granite Peak (2154 meters) are “island” features within the hydrographic unit.

Intermittent streams originating in the surrounding mountains provide water flow into the hydrographic basin. Sustainable wetlands occur at the north end of White Horse Flat where Felt Wash, originating in the Goshute Mountains, enters the Great Salt Lake Desert mud flats. That marsh lies at 1290 meters. Numerous springs feed the lowlands of Fish Springs Flat at an elevation of 1309 meters along the eastern terminus of the Fish Springs Range and the Fish Springs Wash delta.

Vegetation is typical of the Great Salt Lake Desert. Lowest elevations are barren mud flats and sand sheets, grading to desert shrub communities as elevation rises from the desert floor. Pinyon/juniper uplands grade sagebrush communities along alluvial fans.

Analytic Results

Prehistoric Evidential Themes

Approximately 1062 square kilometers, 2.5% of the total area, were inventoried within the Great Salt Lake analytic unit. (Table 5.25) (Figure 5.52) Three hundred eleven sites from those inventories were considered in the weighted analysis. One thousand one hundred sixteen sites are reported for the entire analytic unit.

Within all analytic classes, only Great Basin pine, Juniper steppe and Wet grassland classes within the vegetation evidential theme have not been inventoried. (Table 5.26) The total extent of the missing class area is 202 square kilometers (0.4% of the total analytic area). Inventoried space in several zones, including chaparral and water in the vegetation theme, mountain areas in landform and slopes greater than 15 degrees are under-represented within the inventoried sample. When evaluating weighted contrasts, sampling inconsistencies and site densities relative to the model area data set as well as inventoried data sets were considered.

Weights of evidence tables identify classes within each evidential theme that lie “inside” of the predictive pattern. (Table 5.26) Normalized contrast for juniper/pinyon vegetation class is highest when evaluated with all categories of prehistoric sites. Chi-square is also significantly high for the class. Barren areas, or those with sparse vegetation, have a correspondingly high negative contrast value, indicating a lower than expected probability for sites. By contrast, barren areas have the highest contrast for non-sites. (Figure 5.53)

Contrasts for distance to springs and streams is variable across each different analytic run. When all sites are considered, the 200 meter buffer has the highest contrast. When inventoried sites are weeded, the 400 meter buffer distance has the highest contrast, while the sites within 250 meter cells have highest contrasts in areas greater then 2000 meters. The large expanse of desert within the Great Salt Lake analytic unit and peculiarities of the weeding process appear to be driving the contrast results. Since the weighted results are inconsistent across all analytic runs, distance to water was not included as a predictive theme. (Figure 5.54)

Proximity of sites to wetlands, on the other hand, uniformly identifies the 0 to 1000 meter buffer as a reliably predictive class. (Figure 5.55) In the three analytic runs with sites, the 1000 meter buffer exhibits the highest contrast values. Corresponding negative values are present in the non-site analysis. Site location more than 5000 meters from potential wetlands is inconsistently identified in the weights tables. Relatively high contrasts in weeded all site and inventoried site analysis, likely reflect an upland adaptation within the analytic unit.

Like distance to streams and springs, analytic runs for slope are less than conclusive. Contrast values for slopes 15 to 30 degrees are based upon a relatively high frequency of sites within a slope class that accounts for less than 1% of the analytic unit. Those sites are most likely rockshelters. Contrast for slope between 5 and 15 degrees is also relatively high, but chi-square calculations suggest that the distribution of sites within that class is normal. Most of the sites within the analytic unit lie on slopes between 0 and 5 degrees, but proportionally, that frequency lies outside of the expected distribution. While one would be inclined to accept the 15 to 30 degree slopes as predictive for the theme, non-site analysis suggests that it is likewise predictive for non-sites. As a result, slope was not selected as a predictive theme in the Salt Lake analytic unit.

Landform as an evidential theme provides stronger relational contrasts than incremental slope. While most of the sites occur within flats along the valley floors, the piedmont is consistently characterized by higher than expected site frequencies. Chi-square also confirms a non-random distribution. The piedmont as a landform class subsumes a portion of the slopes within a 5 to 30 degrees range and was selected as a predictive evidential class. (Figure 5.56)

Prehistoric Predictive Response

After identifying evidential theme classes that lie inside the predictive pattern, a response theme was calculated in order to compile a probability map based upon the likelihood of encountering a site within the aggregated evidential themes. (Figure 5.57) Normalized posterior probabilities were used as a means to evaluate tabular results. Observed breaks are apparent at posterior probabilities of 0.016 and 0.004. (Table 5.27) (Figure 5.58) Highest probabilities for encountering sites occur in areas within 1000 meters of potential wetlands and on piedmont slopes, or a within a combination of proximity to potential wetland, piedmont slopes, and juniper/pinyon vegetation zones. Probabilities decrease in areas characterized by the presence of a predictive single evidential theme. When no predictive classes are present, posterior probabilities fall below the0 .005 critical prior probability value, and predictive probabilities are lowest.

Results of the probability model were analyzed in a spatial context in order to validate model results. (Table 5.28) Extent of the sensitivity areas, those within both the entire analytic unit and inventoried portions of the analytic unit, were contrasted with actual areal extent of the sites within each sensitivity zone. Highest ratios of site area to sensitivity area should fall within zones of highest probability if the model is accurate. Summary tables show that the areal density of all sites and inventoried sites are indeed highest within areas of high to medium sensitivity. While areas of low sensitivity comprise two-thirds of the analytic unit they consistently maintain the lowest values of site to total area. (Figure 5.59)

The response theme calculated with Spatial Data Modeler accurately grouped the intersection of predictive classes into probability zones. As a result, recalculating the response using Spatial Analyst^Ò produce similar results. Variation in probability and site areas change less than 0.5%.

Historic Evidential Themes

Two hundred three historic sites are reported within the Great Salt Lake analytic unit. Within the 1062 square kilometers subset of inventories greater than 640 acres in extent, 61 sites are considered for analysis. (Table 5.25) (Figure 5.60) Distance to existing roads and water sources were considered as predictive evidential themes for historic resources. (Table 5.29) Roads were buffered at 200 meter intervals to 1000 meters, and water sources at 200, 400 and 1000 meter intervals. Like prehistoric sites, weights were calculated for using all sites, inventoried sites, and inventoried site areas within a 250 meter inventoried grid. Within the 250 meter grid, non-site weights were also calculated for comparison with weighted site results and calculation of chi-square.

Within the road evidential theme, a buffered distance of 200 meters consistently revealed the highest positive contrast, with a corresponding negative contrast for non-sites. (Figure 5.61) Chi-square results for this class fail to meet the desired threshold for non-random distribution. Highest contrasts are reflected within the 600-800 meter buffer for site areas within the 250 meter grid, but a high positive contrast in non-site weights within the same class suggests that the occurrence of non-sites is highly probable for that buffered area. With the conflicting results, no class within roads was clearly predictive for a positive contrast.

Within inventoried areas, contrasts for distance to water showed increasingly high probability as distance from the water source increased from 0 to 1000 meters. Chi-square results show a correspondingly positive relationship. Negative contrasts for non-site areas corroborate the probability of encountering sites within an ascending radius of water sources. (Figure 5.62)

Historic Predictive Response

Response themes cannot be built without a minimum of at least two evidential themes. The water buffer alone could have been used as a predictive mask, but further analysis of the weights tables showed a good correlation for sites not occurring more than 1000 meters from roads, or more than 1000 meters from water sources. The highest negative contrasts for each theme was selected as “inside” the pattern and used to calculate a response theme. Since the resulting response table calculates weights and probabilities for combined classes, the only difference between selecting positive or negative contrasts is that the intersection of evidential theme classes appears at the bottom of the table, associated with the lowest posterior probabilities.

Three observed breaks are evident within posterior probabilities generated for historic evidential themes in the Great Salt Lake analytic unit. (Table 5.30) (Figure 5.63) Breaks occur at 0.002 and 0.0007, with prior probabilities set at 0.0003. No areas within high probability fall within buffers lying further than 1000 meters from water or roads. Moderate probability areas lie within 1000 meters of roads, but occasionally more than 1000 meters from water, and low probability zones always occur more than 1000 meters distant from roads, or more than 1000 meters from roads and water.

The corresponding sensitivity map (Figure 5.64) and summary table (Table 5.31) shows that areas of highest and medium probability include well over 80% of all sites and inventoried sites by area. Slightly less than 50% of the analytic unit falls within the low probability area, yet less than 10% of the historic site areas occur within that zone.

Recalculating the historic probability by intersection of predictive classes redefines medium and low probability zones. The medium probability area is expanded to 34% of the analytic unit with 29% of all sites present, while the low probability zone decreased to 28% of the analytic unit and includes slightly more than 1% of all sites. Within inventoried areas, less than 1% of the sites fall within the low probability zone. (Figure 5.65)

UPPER SNAKE ANALYTIC UNIT

Analytic Unit Description

The Upper Snake sub-region covers approximately 3.0 million acres (4801 mi²)/1.2 million hectares (12,435 km²) within southern Idaho, northeastern Nevada, and northwestern Utah (Figure 5.66) or 15% of the GBRI study area. Three hydrographic units, Salmon Falls, Goose, and Raft comprise the analytical portion of the sub-region. (Table 5.32) All three drain in a northeasterly direction towards the Snake River. Complex, dendritic drainage patterns dominate the Upper Snake sub-region.

Salmon Falls

The Salmon Falls hydrographic unit lies in the westernmost portion of the Upper Snake analytic unit. Salmon Falls Creek and its tributaries is the dominant hydrologic feature of the hydrographic unit. (Figure 5.67) Elevations range from 2631 meters at Ellen D Mountain, and 2410 meters at Middle Stack Mountain near Contact, Nevada, in the southern portion of the hydrographic unit to 900 meters at the confluence of Salmon Falls Creek and the Snake River in the northern portion of the hydrographic unit. Major physiographic feature include the O’Neil and Shoshone Basins, Antelope Pocket and Browns Bench, a major obsidian source, all within the southern half of the unit. Vegetation is primarily sagebrush with some pinyon/juniper woodland. Topography becomes more subdued progressing northward through the Hydrographic unit. Higher mountains give way to low ridges and dissected basalt plateaus.

Goose

The Goose hydrographic unit lies in the central portion of the Upper Snake analytic unit, covering portions of Idaho, Nevada, and Utah. (Figure 5.68) Goose Creek and its tributaries dominate the hydrology of this hydrographic unit. Like Salmon Falls Creek, it drains northward towards the Snake River. Lowest elevations (1290 meters) occur in agricultural lands near the Snake River. Monument Peak (2454 meters) lies in the uplands within the mountainous, west central portion of the hydrographic unit. The Sawtooth National Forest administers most of this area. To the south, low hills and ridges characterize the hydrographic unit, while the northern one-third is relatively flat agricultural lands. Deadman Ridge and Middle Mountain flank respective western and eastern edges of the unit, while Big Draw and Cedar Mountain Draw lie in the south. Sagebrush dominates the landscape outside of agricultural areas, pinyon/juniper woodlands are found in the steeper uplands.

Raft

Along the eastern side of the Upper Snake analytical unit, the Raft River and its tributaries create a major hydrologic feature. (Figure 5.69) The hydrographic unit lies within Idaho and Utah. Several ranges including the Jim Sage Mountains, Alison Mountains, Black Pine Mountains, and Middle Mountain bound this horseshoe-shaped basin. The Raft River Mountains provide a topographic divide between the Upper Snake sub-region and the Great Salt Lake sub-region to the south. Highest elevations occur at Cache Peak (3151 meters) and Mount Independence (3033 meters) along the western edge of the hydrographic unit. The Upper Raft River Valley, Junction Valley and the Holt Basin lie in the southern portion of the hydrographic unit. Basins in the south average approximately 1700 meters, while agricultural lands in the Raft River Valley lie at 1285 meters near the confluence of the Raft and Snake Rivers. Juniper and pinyon dominate higher elevations across the hydrographic unit. Sagebrush is the dominant non-cultivated plant community.

Analytic Results

Prehistoric Evidential Themes

Since a major portion of the Upper Snake analytic unit contained no spatial inventory data, statistical relationships between site and non-site components could not be evaluated. Site location data was derived primarily from Bureau of Land Management data sets. Forest Service and other agency lands within the analytic unit were not included in the analysis.

One thousand six hundred seventy-five prehistoric sites have been recorded on 8085 square kilometers of BLM land within the Upper Snake analytic unit. (Table 5.33) Weights tables for evidential themes were compiled using all weeded sites so that only one training point would occur within each analytic cell. Weeding reduced the total number of training points by approximately 20%. In each of the evidential themes the class with the highest positive contrast was selected as most predictive, with the remainder falling outside of the pattern regardless of whether contrast were positive or negative. (Table 5.34)

Within the vegetation evidential theme, the juniper steppe class and sagebrush zone have a positive contrast, while juniper/pinyon is least predictive for sites. Juniper steppe was considered inside the pattern since its contrast and positive weights were highest. The area of juniper is relatively small, 5.6% of the total analytic unit and weeded sites account for 10.6% of the total. (Figure 5.70)

The only positive contrast for distance from streams and springs is within the 0 to 200 meter buffer. Areas more than 1000 meters from water courses exhibit the highest negative contrast. Over 56% of the weeded sites are located within the 0 to 200 meter buffer. (Figure 5.71)

Areas within 1000 meters of potential wetlands, while relatively small (10.3% of the analytic unit) also have the highest contrast. Areas lying more than 5000 meters from potential wetlands cover the greatest proportion of the analytic unit. They are also moderately predictive, but considered outside of the probability pattern. (Figure 5.72)

Slope and landform exhibit contrasting predictive results. Two percent of the analytic unit lies on slopes between 15 and 30 degrees, but almost 6% of the weeded sites occur within that slope class. Most of the area and most of the sites fall within the 0 to 5 degree slope class. It has a high negative contrast. (Figure 5.73) Within landform, however, areas along the basin or valley floors have the highest contrast. Landform classes are evenly distributed across the analytic unit, but piedmont and mountain both have negative predictive contrasts. (Figure 5.74)

Prehistoric Predictive Response

Snake response themes were run using most predictive classes for each evidential theme. Three distinct breaks occur within the normalized posterior probability values. Highest probabilities range between 0.24 and 0.17, medium probability ranges between 0.17 and 0.058, while lowest probabilities fall between 0.058 and 0.021. Prior probability was set at 0.045. (Table5.35) (Figure 5.75)

The resulting table shows that less than 1% of the analytic unit falls within the high probability zone and a similar percentage of sites are associated with that area while more than one-half of the sites fall within the low probability zone which extends over 66% of the analytic unit. (Table 5.36) Site densities within relatively large cells within the low probability zone (values 8 and 7) (Table 5.35) likely cause the skewed response. (Figure 5.76)

By calculating an intersection of predictive themes, a better correlation between site density and probability is gained. The total area of high probability is expanded to almost 19% of the analytic unit and it contains nearly 24% of the sites, while the low probability zone is decreased to 12% of the analytic unit and includes 8% of the sites. Most of the analytic unit lies within areas of medium probability (70%) and most of the site areal falls within that zone. (Table 5.36) (Figure 5.77)

Historic Evidential Themes

One hundred nineteen historic sites were recorded within the 8085 square kilometers of land managed by BLM. (Table 5.32) Within both the road and water evidential themes considered for historic resources, the buffered class between 0 and 200 meters is most predictive. Contrasts within roads decline consistently with each increasing buffer. (Figure 5.78) A similar, but not as striking decline occurs with buffered distance to water courses. (Figure 5.79) (Table 5.37)

Historic Predictive Response

Three breaks are evident in the historic response theme when the area within 200 meters of roads and water are selected as predictive classes. Breaks occur at normalized posterior probabilities of 0.006 and 0.0027, the prior probability was set at 0.0033. (Table 5.38) (Figure 5.80) Summary tables show that the high probability area has the smallest extent (less than 10% of the analytic unit) while 77% of the analytic unit falls within the low probability zone and 13% falls within medium probability. (Figure 5.81) Sites are evenly distributed across probability zones with approximately one-third of the sites within each zone. (Table 5.39) Proximity to water but not roads, and distances greater than 200 meters for either evidential theme, fall within the low probability zone. Again, the number of training points relative to low probability area biases the response pattern.

Summary calculation of the predictive classes creates a modified response with better site to area ratios between probability zones. Training points associated with distance to water but not roads are included in the medium probability zone with the summary calculation. The composite summary produces a 25% reduction to the areas of the low probability zone and corresponding increase in the extent of the medium zone. (Table 5.39) One-third of the sites still remain in the high probability zone, while almost one-half fall within zones of moderate probability. The remainder of sites lie within the low probability zone. (Figure 5.82)

V. ANALYSIS

Analytic Unit Description

Prehistoric Evidential Themes

Prehistoric Predictive Response

Historic Evidential Themes

Historic Predictive Response

Prehistoric Evidential Themes

Prehistoric Predictive Response

Historic Evidential Themes

Historic Predictive Response

Prehistoric Evidential Themes

Prehistoric Predictive Response

Historic Evidential Themes

Historic Predictive Response

Northern Great Salt Lake Desert

Great Salt Lake

Rush-Tooele Valleys

Southern Great Salt Lake Desert

Prehistoric Evidential Themes

Prehistoric Predictive Response

Historic Evidential Themes

Historic Predictive Response

Salmon Falls

Goose

Raft

Prehistoric Evidential Themes

Prehistoric Predictive Response

Historic Evidential Themes

Historic Predictive Response

Return to Top