Geology topics

Research about United States

Source-linked reports with geographic coverage including United States.

At least 1,729 records · Page 96Linked to original sources

USGS2025

Reservoir and riverine sources of cyanotoxins in Oregon’s Cascade Range rivers tapped for municipal drinking water supply

Reservoirs and downstream rivers draining Oregon’s Cascade Range provide critical water supplies for over 1.5 million residents in dozens of communities. These waters also support planktonic and benthic cyanobacteria that produce cyanotoxins that may degrade water quality for drinking, recreation, aquatic life, and other beneficial uses. This 2016–2020 survey examined the sources and transport of four cyanotoxins—microcystins, cylindrospermopsins, anatoxins, and saxitoxins—in six river systems feeding 18 drinking water treatment plants (DWTPs) in northwestern Oregon. Benthic cyanobacteria, plankton net tows, and (or) Solid-Phase Adsorption Toxin Tracking (SPATT) samples were collected from 65 sites, including tributaries, reservoirs, main stems, and sites at or upstream from DWTPs. Concentrated extracts (320 samples) were analyzed with enzyme-linked immuno-sorbent assays (ELISA), resulting in >90% detection. Benthic cyanobacteria ( n = 80) mostly Nostoc, Phormidium, Microcoleus , and Oscillatoria, yielded microcystins (76% detection), cylindrospermopsins (41%), anatoxins (45%), and saxitoxins (39%). Plankton net tow samples from tributaries and main stems ( n = 94) contained saxitoxins (84%), microcystins (77%), anatoxins (25%), and cylindrospermopsins (22%), revealing their transport in seston. SPATT sampler extracts ( n = 146) yielded anatoxins (81%), microcystins (66%), saxitoxins (37%), and cylindrospermopsins (32%), indicating their presence dissolved in the water. Reservoir plankton net tow samples ( n = 15), most often containing Dolichospermum , yielded microcystins (87%), cylindrospermopsins (73%), and anatoxins (47%), but no saxitoxins. The high detection frequencies of cyanotoxins at sites upstream from DWTP intakes, and at sites popular for recreation, where salmon and steelhead continue to exist, highlight the need for additional study on these cyanobacteria and the factors that promote production of cyanotoxins to minimize effects on humans, aquatic ecosystems, and economies.

Oregon↗

USGS2025

Experimental evaluation of Eastern box turtle (Terrapene carolina carolina) detectability in visual search surveys

Understanding how detection probability varies over time, space, or in response to measurable covariates is important to inform the monitoring and assessment of many species. A standard model to understand detectability, the availability/perception model, admits that detection probability is the composite of two components: availability and ability of surveyors to detect the target. Availability is largely affected by environmental and behavioral factors, whereas perception is primarily affected by attributes of individual observers and survey protocols, and thus can potentially be partially controlled by survey design. We designed and implemented a field study to understand the perception component of detection for Eastern Box Turtles ( Terrapene carolina carolina ) using visual encounter surveys. We obtained and deployed museum specimens of Eastern Box Turtle shells and subjected them to visual search surveys by observers in realistic field situations. Overall, about 50% of the box turtle shells were detected by observers, including 41.5% in what we categorized as partially visible and 63.0% as fully visible. There were significant differences among observers, which may be due to observer-specific variation in search technique; the observers varied in how well they achieved the protocol guidance. Therefore, in visual search surveys, care in study design and analysis should be taken to account for variation in perception to determine detectability, as our study suggests 37% of perceptible targets are missed by surveyors.

Maryland↗

USGS2025

The tortoise and the antilocaprid: Adapting GPS tracking and terrain data to model wildlife walking functions

Context The relationship between slope and terrestrial animal locomotion is key to landscape ecology but underexplored across species. This is partly due to a lack of scalable methodology that applies to a diversity of wildlife. Objectives This study investigates the slope-speed relationship for two species, Texas tortoise ( Gopherus berlandieri ) and pronghorn ( Antilocapra americana ), through the combined application of remote sensing, GPS tracking, behavior models, and parametric distribution. While using readily available Digital Elevation Models (DEM) for pronghorn, we explore the use of very high-resolution lidar Digital Terrain Models (DTM) from Unoccupied Aerial Systems (UAS) to characterize tortoise movements at micro-scales. Methods After classifying animal behavior with GPS tracking data and Hidden Markov Models (HMMs), we analyzed the relationship between the speed of the animals and the slope of the terrain using a 30-m DEM for pronghorn, and a fine-scale UAS DTM for Texas tortoise, and three nonlinear models: Laplace, Gauss, and Lorentz. Results High-resolution DTM, coupled with GPS tracking, accurately models the relationship of speed and slope at a micro-scale, while a DEM is suitable for a larger scale. Laplace models best predicted the speed of both the Texas tortoise and pronghorn. Models showed tortoises, which are not known for rapid and agile movement like the pronghorn, have a broader tolerance for varying slopes at a fine scale. Conclusions These findings enhance understanding of species-specific movement offering valuable insights for habitat management and conservation tailored to species’ behaviors and capabilities.

Colorado, Texas, Wyoming↗

USGS2025

Exposure to ultraviolet radiation induces escape hatching of Cisco (Coregonus artedi) embryos

Cisco (Otoonapii in Ojibwe; Coregonus artedi Lesueur, 1818), is a widely distributed stenothermic freshwater fish whose embryos typically incubate under ice and in the dark. We used Cisco as a model organism for testing the potential of UV-induced escape hatching behaviour. Owing to reduced ice cover and increased water transparency in north temperate lakes, these experiments provide insights into the resilience of coregonine embryos if exposed to ultraviolet radiation (UV-B; 280–320 nm). Eyed Cisco embryos were exposed to artificially sourced UV-B through a series of experiments that measured the hatching rate and fitness (heart rate and pigmentation pattern) 2 days after hatching and under cold [6.6°C] and warm [8.6°C] conditions. These experiments supported an extension of the escape hatching behaviour hypothesis, whereby UV-B exposure induced earlier (ca 30 days) and more punctuated hatching of Cisco embryos, independent of an increase in water temperature. UV-B exposure produced more larvae with irregular pigmentation patterns and reduced heart rates (by about 20%) – both of which could be indicative of reduced fitness. UV-induced escape hatching adusts the fundamental framework in which we characterise fish embryo resilience to increased UV-B exposure and the potential consequences of reduced ice cover. Earlier hatching from UV-B exposure could increase the recruitment bottleneck of these fish by reducing survivorship of the post-hatched larvae.

Minnesota↗

USGS2025

Mapping predicted ecological states at landscape scales using remote sensing data and machine learning

Dryland ecosystems, covering 45% of the Earth's land and supporting over one-third of the global population, face significant threats from land degradation and ecological state change. Managing these ecosystems is complex, and science-based frameworks like Ecological Site Descriptions and state-and-transition models are essential tools for guiding decisions to support ecological health while maintaining stakeholder values such as grazing, wildlife, and recreation. However, alignment of these frameworks with smaller scale soil survey maps limits their applicability to broader ecological processes. Here, we extend these frameworks to larger landscapes with a machine learning approach that integrates large-scale, high-resolution vegetation data with identified ecological states from a data-driven state-and-transition model developed for a landscape-scale Ecological Site Group. A “global” model, which used combined inputs from multiple remotely sensed datasets, outperformed individual dataset models based on evaluation with independent data. Ecological state maps generated through this approach broaden the utility of state-and-transition models across Ecological Site Groups, providing a more spatially robust tool for land management at watershed and larger landscape scales. These methods, and the associated ecological state maps, can help meet critical needs for improved land condition assessments that support development of resource management plans and help identify priority areas for restoration and conservation.

Arizona, Colorado, New Mexico, Utah, Wyoming↗

USGS2025

Slow rupture, long rise times, and multi-fault geometry: The 2020 M6.4 southwestern Puerto Rico mainshock

The M 6.4 mainshock of the southwestern Puerto Rico seismic sequence on 7 January 2020, was one of the most impactful modern earthquakes in the northeastern Caribbean. Due to its offshore location and complex aftershock distribution, its source kinematics remain poorly constrained. This active sequence illuminated a complex set of previously unrecognized structures that indicate multiple causative faults may have slipped during its rupture. Here, we utilize seismic and geodetic observations to enhance model resolution, estimate the finite slip of the mainshock, and test a multi-segment, geologically realistic fault geometry. Our refined model finds a lower rupture velocity and longer rise times than typical for an event of this magnitude. This indicates a slow-evolving rupture process that resembles characteristics of a tsunami earthquake. Although this normal/strike-slip faulting event was not tsunamigenic, these qualities, if pervasive for this region, have important implications for future seismic monitoring and hazards in southwestern Puerto Rico.

Puerto Rico↗

USGS2025

Increased flood exposure in the Pacific Northwest following earthquake-driven subsidence and sea-level rise

Climate-driven sea-level rise is increasing the frequency of coastal flooding worldwide, exacerbated locally by factors like land subsidence from groundwater and resource extraction. However, a process rarely considered in future sea-level rise scenarios is sudden (over minutes) land subsidence associated with great (>M8) earthquakes, which can exceed 1 m. Along the Washington, Oregon, and northern California coasts, the next great Cascadia subduction zone earthquake could cause up to 2 m of sudden coastal subsidence, dramatically raising sea level, expanding floodplains, and increasing the flood risk to local communities. Here, we quantify the potential expansion of the 1% floodplain (i.e., the area with an annual flood risk of 1%) under low (~0.5 m), medium (~1 m), and high (~2 m) earthquake-driven subsidence scenarios at 24 Cascadia estuaries. If a great earthquake occurred today, floodplains could expand by 90 km 2 (low), 160 km 2 (medium), or 300 km 2 (high subsidence), more than doubling the flooding exposure of residents, structures, and roads under the high subsidence scenario. By 2100, when climate-driven sea-level rise will compound the hazard, a great earthquake could expand floodplains by 170 km 2 (low), 240 km 2 (medium), or 370 km 2 (high subsidence), more than tripling the flooding exposure of residents, structures, and roads under the high subsidence scenario compared to the 2023 floodplain. Our findings can support decision-makers and coastal communities along the Cascadia subduction zone as they prepare for compound hazards from the earthquake cycle and climate-driven sea-level rise and provide critical insights for tectonically active coastlines globally.

California, Oregon, Washington↗

USGS2025

Comparison of two benthic assemblage sampling gears for use on intertidal oyster reefs in Louisiana

Background Estuarine biodiversity plays a vital role in supporting ecosystem functions yet remains threatened by climate change and anthropogenic activity. Tracking and identifying estuarine biodiversity trends helps management ensure long-term provisions of human and environmental benefits by contributing to the estimation of habitat loss and the monitoring of restoration and conservation progress. However, the sampling gear and biodiversity metric used may indicate different conclusions, which can lead to uncertainty in the actual state of the ecosystem-level biodiversity. Sampling benthic biodiversity in complex estuarine habitats, such as oyster reefs, is particularly challenging because no one gear type captures entire target assemblages, and differences in gear efficiency on these complex habitats make comparisons across gear types challenging. Methods We investigated how estimates of oyster reef-associated benthic taxa abundance, richness, Pielou’s evenness, and Shannon-Wiener diversity differed across three Crassostrea virginica reefs in Louisiana between suction sampler and substrate tray sampling gears ( n = 6), and how gear influenced comparisons across reefs (3 reefs × 6 replicates × 2 gears). Results Abundance and richness were higher, and Pielou’s evenness was lower, in trays compared to suction samples at all reefs. Shannon-Wiener diversity was similar in suction samples and trays at two out of three reefs. Amphipod taxa were numerically dominant in trays, skewing the distribution of abundances and driving the reef assemblage differences between gears. Abundance and Shannon-Wiener diversity were similar across reefs within each gear. However, there were significant differences in richness across reefs in tray samples only, while evenness differed across reefs only in suction samples. Our results highlight that gear choices, along with biodiversity metrics tracked, can result in different conclusions in biodiversity trends, ultimately affecting conservation decisions and management.

Louisiana↗

USGS2025

Evaluation of SARS-CoV-2 antibody detection methods for wild Cervidae

Wildlife surveillance programs often use serological data to monitor exposure to pathogens. Diagnostic sensitivity and specificity of a serological assay quantify the true positive and negative rates of the diagnostic assay, respectively. However, an assay’s accuracy can be affected by wild animals’ pathogen exposure history and quality of the sample collected, requiring separate estimates of an assay’s detection ability for wild-sampled animals where an animal’s true disease status is unknown (referred to hereafter as sampling sensitivity and specificity). We assessed the sampling sensitivity and specificity of a Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) surrogate virus neutralization test (sVNT) and conventional virus neutralization tests (cVNT) to detect antibodies for ancestral and Omicron B.1.1.529 variants of SARS-CoV-2 in wild white-tailed deer ( Odocoileus virginianus ) and mule deer ( Odocoileus hemionus ). We studied the influence of sample collection method using paired blood samples collected in serum separator tubes and on Nobuto strips from the same animal. Mean estimates of sampling sensitivity and specificity ranged from 0.21–0.95 and 0.94–1.00, respectively, varying by sample collection method, host species, and SARS-CoV-2 variant targeted by the assay. Broadly, sampling sensitivity was estimated to be higher for 1) sera collected in tubes, 2) detecting pre-Omicron SARS-CoV-2 variants, and 3) sVNT relative to cVNT assays. Sampling specificity tended to be high for all tests. We augmented our study with SARS-CoV-2 spike protein sequences derived from sampling locations and times coincident with white-tailed deer captures, finding common amino acid mutations relative to the sVNT Omicron antigen variant. The mutations may indicate that the SARS-CoV-2 variants circulating in cervids from 2021 through 2024 may be better adapted to cervid hosts and more closely related to variants that circulated in humans prior to Omicron variants. We conclude our study with an inter-test comparison of sVNT results, revealing that 40 % inhibition is an optimal threshold for test positivity when testing deer sera for responses to Omicron variant B.1.1.529, compared to the 30 % inhibition recommended for ancestral variants.

Colorado, Illinois, Minnesota, New York, Pennsylva↗

USGS2025

Detection of landslide-generated tsunami by shipborne GNSS precise point positioning

Precise point positioning (PPP) of ships using Global Navigation Satellite System (GNSS) data reveals the precise movements of marine vessels. This method may quantify anomalies in sea surface height with implications for oceanographic monitoring, exploration, and tsunami warning. The GNSS PPP data from the R/V Sikuliaq , a research ship of the University of Alaska Fairbanks, were processed to detect a small local tsunami generated by the Lowell Point landslide, which occurred near Seward, Alaska, on 8 May 2022 (UTC). The GNSS receiver aboard the R/V Sikuliaq recorded the waves generated by the landslide, with a maximum wave amplitude of 6 cm and wave periods between 40 and 50 s. These results are consistent with simulations of the landslide event.

Alaska↗

USGS2025

Footprints of past mining in Alaska (USA) derived from high-resolution satellite imagery

Mapping the land area used for mining in the past is essential for guiding the remediation of affected landscapes and assessing the resource potential of related waste products. Despite significant recent progress delineating footprints of active and inactive mining globally, the known inventory of such mine lands remains incomplete. Here, I describe a new map dataset of footprints of land surface disturbance and waste at sites of past mining in Alaska (USA) based on visual interpretation of satellite imagery. This dataset maps 6–14 times the area of previous regional and global mine footprint maps in Alaska and is the first in the region to explicitly delineate mine waste landforms (e.g., tailings piles). The data are publicly available from the U.S. Geological Survey under a “no rights reserved” Creative Commons (CC0) license agreement.

Alaska↗

USGS2025

Satellite imagery can predict bird species occupancy and inform multispecies management in pine savannas

Multispecies management can contribute to meeting growing challenges of preserving biodiversity, yet current game and threatened species management often focuses on individual species. Satellite imagery available at high spatial and temporal resolution provides a potential tool to overcome the challenge posed by multispecies management of linking patterns of habitat use among species. We sought to determine whether satellite imagery could be used to describe patterns of species occupancy and inform multispecies management in pine savannas in Georgia, USA. We conducted point-count surveys at 7 sites in 2022 for 3 bird species: Colinus virginianus (Northern Bobwhite), Dryobates borealis (Red-cockaded Woodpecker), and Peucaea aestivalis (Bachman’s Sparrow). We built single-season occupancy models comparing a set of models using covariates collected from field vegetation surveys and another set using covariates extracted from Sentinel-2 satellite imagery. We then used a multi-objective optimization algorithm to identify quasi-optimal management solutions (i.e., sets of covariate values from top satellite imagery metric models). We found that models created using satellite imagery performed well at predicting occupancy of all 3 species as measured by the area under the receiver operating characteristic curve (AUC > 0.8) and had higher AUC scores than field-derived habitat covariate-based models. We found combinations of metrics that could result in high rates of predicted probability of occupancy for all species (within 86% of highest possible occupancy probability), but these combinations did not exist at any of the sites. Our results demonstrate that (1) satellite imagery can allow users to build reliable occupancy models without intensive field-based vegetation surveys; and (2) C. virginianus , D. borealis, and P. aestivalis in pine savanna ecosystems could be simultaneously managed through more frequent burning, changes in canopy cover or by producing suitable heterogeneity of habitats after identifying an appropriate scale of management.

Georgia↗

USGS2025

Genetic structure and diversity in wild populations of the Light-footed Ridgway’s Rail reflect 20 years of augmentation through captive breeding and release

Captive breeding and release programs aimed at recovery of rare species can be informed by genetic data to help select high-diversity source populations, make pairing decisions to minimize inbreeding, and manage release strategies. We developed a set of 54 microsatellite loci to assess genetic structure and diversity across the United States range of the Light-footed Ridgway’s Rail ( Rallus obsoletus levipes ), a federally endangered marsh bird for which populations have been augmented by a captive breeding program annually since 2001. We identified three regional genetic clusters, with the highest genetic diversity reported in the central cluster, which included all sampled wetlands in north San Diego County. Recent (2019–24) captive-breeding adults all clustered within the northernmost cluster (Orange and Ventura Counties), which was expected given that this cluster included the source wetland for the captive breeding program. Gene flow rates, which approximate the proportions of individuals in a population originating from other populations, were relatively high among clusters (4–24 percent) and may have been enhanced through the release of captive-bred rails. Based on the genetic data analyzed in a genetic rescue decision framework, sourcing new breeding birds from the north San Diego County cluster could provide the greatest genetic diversity benefits. The northernmost cluster, which included Mugu Lagoon and all sampled Orange County wetlands, was considered the most in need of genetic rescue. Recent breeding pairs in the captive breeding program have comparatively low diversity and high interrelatedness. Sourcing birds from wetlands with high genetic diversity and population sizes, assessing genetic relatedness before pairing, and focusing releases in areas that have low estimates of genetic diversity could improve the distribution of genetic diversity across wild populations in the future.

California↗

USGS2025

Linking age and social status of cooperative breeders to vulnerability throughout the harvest season

Individual behaviors are influenced by environmental, genetic, and demographic factors. Some animals choose to live in groups and cooperatively breed, and their behaviors can change depending on dynamic factors such as group size and composition that affect group persistence. In Idaho, USA, gray wolves ( Canis lupus ) are harvested annually, providing an opportunity to investigate the effects of harvest and seasonal behaviors on a population of cooperative breeders. These annual hunting and trapping seasons overlap with the dispersal and breeding periods for wolves and we do not know how harvest affects the vulnerability of different sex and age classes during these important biological periods. We applied 9 years (2009–2018) of genetic, age, and harvest data from harvested wolves to investigate how behaviors (dispersal and breeding) and biological drivers might influence the vulnerability of wolves to harvest. We created pedigrees from genotypes of non-invasively collected scats to estimate the expected proportion of the wolf population composed of 3 different age classes (pup, yearling, and sexually mature or ≥2 years old) and compared them to the observed number of each age class harvested during biologically significant periods (i.e., dispersal and breeding). We found that pups were more vulnerable to harvest in December when wolf harvest transitioned largely to trapping (accounts for 66% of harvest), and found evidence that adults were more vulnerable to harvest during their breeding season in January and February. In contrast, we found no difference in the expected versus observed number of wolves ≥2 years old in the harvest during peak dispersal season (December), or in the expected versus observed number of yearlings in the harvest during September and October when pups are mobile and groups of wolves abandon the use of pup-rearing sites. Some age classes were disproportionally harvested during certain periods for specific years, but this was not consistent across all years, suggesting there is more to learn about the vulnerability of different age classes to harvest. We found harvest can disproportionally affect some demographic classes of individuals depending on year, biological period, and harvest type. With wolves continuing to recolonize historical ranges, our approach can benefit managers and future studies with the goal of identifying how interannual harvest affects groups of wolves.

Idaho↗

USGS2025

Movement patterns of invasive red swamp crayfish vary with sex and environmental factors

Invasive species disproportionately invade freshwater ecosystems, threatening biodiversity. Defining when, where, and why aquatic invasive species move can help inform management strategies, yet the movement ecology of some of the most pervasive invasive species remains unknown. Red swamp crayfish ( Procambarus clarkii ; RSC) are the most widespread invasive crayfish and negatively affect ecosystems worldwide. We employed high-dimensional acoustic telemetry to investigate the movement patterns of 24 individual RSC across three months in an invaded water body. We assessed the effects of various extrinsic factors, such as time of day, temperature, precipitation, and proximity to the water’s edge, along with intrinsic factors, such as sex, reproductive form, and size, at different scales, including movement steps, range distribution, and behavioral states. We found that movement patterns across all scales were overwhelmingly driven by sex and reproductive form. Furthermore, RSC showed increased overall activity at night and near the water’s edge. By establishing baseline movement patterns and identifying key contributing factors, these findings provide a foundation for the development of adaptive management strategies for controlling invasive RSC populations.

Michigan↗

USGS2025

Fisheries dependent and independent data inform a capture technique for an emerging invasive fish species in the mainstem Mississippi River; Black Carp Mylopharyngodon piceus

Black Carp Mylopharyngodon piceus were imported into the United States in the 1970s and 1980s for use in aquaculture; escape occurred and reported wild captures increased. Lacking species-specific capture methods, we assessed fisheries dependent incidental Black Carp catches for a common method, hoop nets, by kernel density analysis to identify an area of increased reporting and compare frequency of reports for water temperature, river stage, and capture date to identify seasonality. We then used fisheries independent effort to identify co-occurrence of species via non-metric multi-dimensional scaling and fit Black Carp catch and environmental covariates by generalized linear models to assess site-specific environmental covariates facilitating capture. The best approximating distribution was refitted for predictions and inference. The greatest density of fisheries dependent hoop net captures (39 %) was near the confluence of the Missouri and Mississippi rivers, primarily from July-September. Captures were characterized by median water temperature 26.7°C, river stage 5.02 m, and 223 day-of-year (DOY; mid-August). Ordination of fisheries independent catch identified similarity in environmental covariates of Smallmouth Buffalo Ictiobus bubalus and Black Carp. The probability of capturing ≥ 1 Black Carp increased with DOY, decreased with increasing current velocity, and increased with depth. Most captures occurred in outside bends (87 %) or side channels (12 %). Probability of Black Carp capture was low but increased in summer and early fall when stage is lower, facilitating reduced current velocity and access to deeper areas. Results may be validated beyond this river segment to test if site-specific hydrology or habitat characteristics facilitated increased commercial and biologist capture and for replication.

Illinois, Missouri↗

USGS2025

Biodiversity on public lands: How community science can help

Species inventories are crucial for conservation but are difficult to assemble and maintain. Bioblitzes, which encourage the public to document biodiversity in a particular area and timeframe, may offer useful information but their integration with other datasets poses challenges. We investigated the potential contribution of bioblitzes to natural resource management using observations from the United States National Park Service (NPS) 2016 Centennial Bioblitz. Through automated cross-referencing over 19,000 iNaturalist “research-grade” observations from 107 national parks with existing park inventory lists, we matched 86% of species documented in the Bioblitz to NPS species lists, based on current taxonomy, and matched another 6% of species using alternative scientific names using our matching process in R. Of the remaining 13.5% that did not match the NPS species lists, we manually found that 84% of the unmatched species were found within the lists or were outside the boundaries of the park, identifying 141 native species that were unrecorded in NPS species lists. Many introduced species were recorded; they were more likely to be documented in parks closer to cities. Parks near cities also drew more participants. Our study shows how public participation, through iNaturalist and bioblitzes, can facilitate biodiversity monitoring across large spatial scales.

Conservation Science and Practice↗

USGS2025

Spatial stream network modeling of water temperature within the White River Basin, Mount Rainier National Park, Washington

Water temperature is a primary control on the occurrence and distribution of fish and other ectothermic aquatic species. In the Pacific Northwest, cold-water species such as Pacific salmon ( Oncorhynchus spp.) and bull trout ( Salvelinus confluentus ) have specific temperature requirements during different life stages that must be met to ensure the viability of their populations. Rivers draining Mount Rainier in western Washington, including the White River along its northern flank, support a number of cold-water fish populations, but the spatial distribution of water temperatures, particularly during late-summer baseflow during August and September, and the climatic, hydrologic, and physical processes regulating it are not well constrained. Spatial stream network (SSN) models, which are generalized linear models that incorporate streamwise spatial autocovariance structures, were fit to mean and 7-day average daily maximum water temperature for August and September for the White River Basin. The SSN models were calibrated using water temperature measurements collected in 2010 through 2020. The extent of the models included the White River and its tributaries upstream from its confluence with Silver Creek in Mount Rainier National Park, Washington. SSN models incorporated covariates hypothesized to represent the climatic, hydrologic, and physical processes that influence water temperature. SSN models were fit to the measured data and compared to generalized linear models that lacked spatial autocovariance structures. Statistically significant covariates within the best-fit models included the proportion of ice cover and forest cover within the basin, mean August air temperature, the proportion of consolidated geologic units, and snow-water equivalent. Statistical models that included spatial autocovariance structures had better predictive performance than those that did not. Additionally, models of mean August and September water temperature had better predictive performance than those of 7-day average daily maximum temperature in August and September. Predictions of the spatial distribution of water temperature were similar between August and September with a general warming in the downstream part of the mainstem White River compared to cooler water temperatures in the high-elevation headwater streams. The proportion of ice cover emerged as an inversely related significant covariate to both mean August and September water temperature because streams that receive glacial meltwater are colder than non-glaciated streams. Water temperatures of the upper White River increased downstream and are attributed to warming of water temperature from accumulated solar radiation and inflow of non-glaciated tributaries. Estimated water temperatures for the upper White River model are 3–4 degrees Celsius (°C) warmer for tributaries, but 1–2 °C cooler for the mainstem compared to the regional-scale model. Differences between the upper White River SSN model and the regional-scale NorWeST model are attributed to the fact that the upper White River SSN included water temperature observations specific to the upper White River, whereas water temperature observations from lower elevation streams and downstream from the Mount Rainer National Park boundary were used in the regional scale model.

Washington↗