Search USGS⌕ Search

SEARCH · Search USGS

Results for “Statistical Methodology”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2Linked to original sources
USGS2021

Decision-support framework for linking regional-scale management actions to continental-scale conservation of wide-ranging species

Anas acuta (Northern pintail; hereafter pintail) was selected as a model species on which to base a decision-support framework linking regional actions to continental-scale population and harvest objectives. This framework was then used to engage stakeholders, such as Landscape Conservation Cooperatives’ (LCCs’) habitat management partners within areas of importance to pintails, while maximizing cross-taxa effects from the framework. The mathematical framework for the model had been previously developed for pintails. A key assumption incorporated into the model is that density dependence in survival occurs during the post-hunting (winter) period, where resources are hypothesized to be limiting. Because few data are available to directly inform this process, the approach used was to build a hierarchical Bayesian integrated population model (IPM) that simultaneously uses data from bird-band recoveries, breeding population counts, and harvest surveys to estimate values of parameters of an annual population projection model, including population size, survival rate, reproductive rate, and process and observation error variances, that are logically consistent with each other, given the mathematical structure imposed through the IPM. The main accomplishments of this study are (1) development of an IPM for pintail to guide harvest and habitat management, (2) development of a Prairie Parkland Region breeding submodel to predict pintail productivity, (3) development of statistical methodology to estimate pintail productivity (as measured by the ratio of juvenile to adults in hunter-collected wing samples) and winter survival and to relate these estimates to covariates, and (4) illustration of how to use a model and estimated parameter values to predict pintail population size and sustainable harvest as a function of habitat. Estimation of pintail survival from bird-banding data shows that there has been relatively little variation in survival over the period 1960–2013. A productivity model showed strong effects of breeding ground conditions, wintering-ground precipitation, and density dependence on pintail productivity. Thus, most temporal variation in pintail demographic rates has been due to effects on reproduction and not survival, including effects of breeding or wintering-ground habitat. These results indicate that habitat conservation efforts may be most effective if they focus on maintaining or increasing breeding and wintering-ground habitat to increase pintail productivity rather than pintail survival. Environmental perturbations in excess of historical experience, such as what could occur under climate change, might have meaningful effects on survival but cannot be estimated with current data. Direct effects of climate, land use, or management are likely to be greater on productivity than survival, but substantial uncertainty remains about predictions of equilibrium population size and sustainable yield.

Open-File Report↗
USGS2008

Estimating total population size for adult female sea turtles: Accounting for non-nesters

Assessment of population size and changes therein is important to sea turtle management and population or life history research. Investigators might be interested in testing hypotheses about the effect of current population size or density (number of animals per unit resource) on future population processes. Decision makers might want to determine a level of allowable take of individual turtles of specified life stage. Nevertheless, monitoring most stages of sea turtle life histories is difficult, because obtaining access to individuals is difficult. Although in-water assessments are becoming more common, nesting females and their hatchlings remain the most accessible life stages. In some cases adult females of a given nesting population are sufficiently philopatric that the population itself can be well defined. If a well designed tagging study is conducted on this population, survival, breeding probability, and the size of the nesting population in a given year can be estimated. However, with published statistical methodology the size of the entire breeding population (including those females skipping nesting in that year) cannot be estimated without assuming that each adult female in this population has the same probability of nesting in a given year (even those that had just nested in the previous year). We present a method for estimating the total size of a breeding population (including nesters those skipping nesting) from a tagging study limited to the nesting population, allowing for the probability of nesting in a given year to depend on an individual's nesting status in the previous year (i.e., a Markov process). From this we further develop estimators for rate of growth from year to year in both nesting population and total breeding population, and the proportion of the breeding population that is breeding in a given year. We also discuss assumptions and apply these methods to a breeding population of hawksbill sea turtles (Eretmochelys imbricata) from the Caribbean. We anticipate that this method could also be useful for in-water studies of well defined populations.

Book chapter↗
USGS1981

Improved population estimates through the use of auxiliary information

When estimating the size of a population of birds, the investigator may have, in addition to an estimator based on a statistical sample, information on one of several auxiliary variables, such as: (1) estimates of the population made on previous occasions, (2) measures of habitat variables associated with the size of the population, and (3) estimates of the population sizes of other species that correlate with the species of interest. Although many studies have described the relationships between each of these kinds of data and the population size to be estimated, very little work has been done to improve the estimator by incorporating such auxiliary information. A statistical methodology termed 'empirical Bayes' seems to be appropriate to these situations. The potential that empirical Bayes methodology has for improved estimation of the population size of the Mallard (Anas platyrhynchos) is explored. In the example considered, three empirical Bayes estimators were found to reduce the error by one-fourth to one-half of that of the usual estimator.

Book chapter↗
USGS1999

Suite versus composite statistics

Suite and composite methodologies, two statistically valid approaches for producing statistical descriptive measures, are investigated for sample groups representing a probability distribution where, in addition, each sample is probability distribution. Suite and composite means (first moment measures) are always equivalent. Composite standard deviations (second moment measures) are always larger than suite standard deviations. Suite and composite values for higher moment measures have more complex relationships. Very seldom, however, are they equivalent, and they normally yield statistically significant but different results. Multiple samples are preferable to single samples (including composites) because they permit the investigator to examine sample-to-sample variability. These and other relationships for suite and composite probability distribution analyses are investigated and reported using granulometric data.

Sedimentary Geology↗
USGS2003

A hydrologic network supporting spatially referenced regression modeling in the Chesapeake Bay watershed

The U.S. Geological Survey has developed a methodology for statistically relating nutrient sources and land-surface characteristics to nutrient loads of streams. The methodology is referred to as SPAtially Referenced Regressions On Watershed attributes (SPARROW), and relates measured stream nutrient loads to nutrient sources using nonlinear statistical regression models. A spatially detailed digital hydrologic network of stream reaches, stream-reach characteristics such as mean streamflow, water velocity, reach length, and travel time, and their associated watersheds supports the regression models. This network serves as the primary framework for spatially referencing potential nutrient source information such as atmospheric deposition, septic systems, point-sources, land use, land cover, and agricultural sources and land-surface characteristics such as land use, land cover, average-annual precipitation and temperature, slope, and soil permeability. In the Chesapeake Bay watershed that covers parts of Delaware, Maryland, Pennsylvania, New York, Virginia, West Virginia, and Washington D.C., SPARROW was used to generate models estimating loads of total nitrogen and total phosphorus representing 1987 and 1992 land-surface conditions. The 1987 models used a hydrologic network derived from an enhanced version of the U.S. Environmental Protection Agency's digital River Reach File, and course resolution Digital Elevation Models (DEMs). A new hydrologic network was created to support the 1992 models by generating stream reaches representing surface-water pathways defined by flow direction and flow accumulation algorithms from higher resolution DEMs. On a reach-by-reach basis, stream reach characteristics essential to the modeling were transferred to the newly generated pathways or reaches from the enhanced River Reach File used to support the 1987 models. To complete the new network, watersheds for each reach were generated using the direction of surface-water flow derived from the DEMs. This network improves upon existing digital stream data by increasing the level of spatial detail and providing consistency between the reach locations and topography. The hydrologic network also aids in illustrating the spatial patterns of predicted nutrient loads and sources contributed locally to each stream, and the percentages of nutrient load that reach Chesapeake Bay.

Environmental Monitoring and Assessment↗
USGS2008

Seabed mapping and characterization of sediment variability using the usSEABED data base

We present a methodology for statistical analysis of randomly located marine sediment point data, and apply it to the US continental shelf portions of usSEABED mean grain size records. The usSEABED database, like many modern, large environmental datasets, is heterogeneous and interdisciplinary. We statistically test the database as a source of mean grain size data, and from it provide a first examination of regional seafloor sediment variability across the entire US continental shelf. Data derived from laboratory analyses ("extracted") and from word-based descriptions ("parsed") are treated separately, and they are compared statistically and deterministically. Data records are selected for spatial analysis by their location within sample regions: polygonal areas defined in ArcGIS chosen by geography, water depth, and data sufficiency. We derive isotropic, binned semivariograms from the data, and invert these for estimates of noise variance, field variance, and decorrelation distance. The highly erratic nature of the semivariograms is a result both of the random locations of the data and of the high level of data uncertainty (noise). This decorrelates the data covariance matrix for the inversion, and largely prevents robust estimation of the fractal dimension. Our comparison of the extracted and parsed mean grain size data demonstrates important differences between the two. In particular, extracted measurements generally produce finer mean grain sizes, lower noise variance, and lower field variance than parsed values. Such relationships can be used to derive a regionally dependent conversion factor between the two. Our analysis of sample regions on the US continental shelf revealed considerable geographic variability in the estimated statistical parameters of field variance and decorrelation distance. Some regional relationships are evident, and overall there is a tendency for field variance to be higher where the average mean grain size is finer grained. Surprisingly, parsed and extracted noise magnitudes correlate with each other, which may indicate that some portion of the data variability that we identify as "noise" is caused by real grain size variability at very short scales. Our analyses demonstrate that by applying a bias-correction proxy, usSEABED data can be used to generate reliable interpolated maps of regional mean grain size and sediment character.

Continental Shelf Research↗
USGS2014

Fitting statistical distributions to sea duck count data: implications for survey design and abundance estimation

Determining appropriate statistical distributions for modeling animal count data is important for accurate estimation of abundance, distribution, and trends. In the case of sea ducks along the U.S. Atlantic coast, managers want to estimate local and regional abundance to detect and track population declines, to define areas of high and low use, and to predict the impact of future habitat change on populations. In this paper, we used a modified marked point process to model survey data that recorded flock sizes of Common eiders, Long-tailed ducks, and Black, Surf, and White-winged scoters. The data come from an experimental aerial survey, conducted by the United States Fish & Wildlife Service (USFWS) Division of Migratory Bird Management, during which east-west transects were flown along the Atlantic Coast from Maine to Florida during the winters of 2009–2011. To model the number of flocks per transect (the points), we compared the fit of four statistical distributions (zero-inflated Poisson, zero-inflated geometric, zero-inflated negative binomial and negative binomial) to data on the number of species-specific sea duck flocks that were recorded for each transect flown. To model the flock sizes (the marks), we compared the fit of flock size data for each species to seven statistical distributions: positive Poisson, positive negative binomial, positive geometric, logarithmic, discretized lognormal, zeta and Yule–Simon. Akaike’s Information Criterion and Vuong’s closeness tests indicated that the negative binomial and discretized lognormal were the best distributions for all species for the points and marks, respectively. These findings have important implications for estimating sea duck abundances as the discretized lognormal is a more skewed distribution than the Poisson and negative binomial, which are frequently used to model avian counts; the lognormal is also less heavy-tailed than the power law distributions (e.g., zeta and Yule–Simon), which are becoming increasingly popular for group size modeling. Choosing appropriate statistical distributions for modeling flock size data is fundamental to accurately estimating population summaries, determining required survey effort, and assessing and propagating uncertainty through decision-making processes.

Statistical Methodology↗
USGS2013

Temporal variation and scale in movement-based resource selection functions

A common population characteristic of interest in animal ecology studies pertains to the selection of resources. That is, given the resources available to animals, what do they ultimately choose to use? A variety of statistical approaches have been employed to examine this question and each has advantages and disadvantages with respect to the form of available data and the properties of estimators given model assumptions. A wealth of high resolution telemetry data are now being collected to study animal population movement and space use and these data present both challenges and opportunities for statistical inference. We summarize traditional methods for resource selection and then describe several extensions to deal with measurement uncertainty and an explicit movement process that exists in studies involving high-resolution telemetry data. Our approach uses a correlated random walk movement model to obtain temporally varying use and availability distributions that are employed in a weighted distribution context to estimate selection coefficients. The temporally varying coefficients are then weighted by their contribution to selection and combined to provide inference at the population level. The result is an intuitive and accessible statistical procedure that uses readily available software and is computationally feasible for large datasets. These methods are demonstrated using data collected as part of a large-scale mountain lion monitoring study in Colorado, USA.

Statistical Methodology↗
USGS2015

Deployment of paired pushnets from jet-propelled kayaks to sample ichthyoplankton

Accessing and effectively sampling the off-channel habitats that are considered crucial for early life stages of freshwater fishes constitute a difficult challenge when common ichthyoplankton survey methods, such as push nets, are used. We describe a new method of deploying push nets from jet-propelled kayaks to enable the sampling of previously inaccessible off-channel habitats. The described rig is also functional in more open and accessible habitats, such as the main channel of rivers or reservoirs. Although further evaluation is necessary to ensure that results are comparable across studies, the described push-net system offers a statistically rigorous methodology that generates replicate samples from a wide range of freshwater habitats that were previously inaccessible to this gear type.

North American Journal of Fisheries Management↗
USGS2021

Comment on “Which earthquake accounts matter” by Susan E. Hough and Stacey S. Martin

In their analysis of the U.S. Geological Survey’s (USGS) “Did You Feel It?” (DYFI) data Hough and Martin (2021) claim, among other assertions, that the following: Socioeconomic and geopolitical factors can introduce biases in the USGS’ characterization of earthquakes and their effects, especially if online data collection systems are not designed to be broadly accessible; These biases can, in turn, potentially cascade in myriad ways, potentially shaping our understanding of an earthquake’s impact and the characterization of seismic hazard; and Caution should be urged when relying on data from the DYFI system to characterize the distribution of shaking from large earthquakes in India and other parts of the world (outside of the United States). Claims of inequity in access, systematic data biases, or urging caution in the usage of data from critical governmental earthquake information systems should not be made, nor taken, lightly. Several assertions made by Hough and Martin (hereafter, H&M) about the nature of DYFI contributors—and the data they provide—leave a false narrative concerning DYFI system accessibility and quality that H&M have not adequately substantiated. I describe several shortcomings of H&M’s demographic statistics and methodology, focusing on four main concerns. First, DYFI has revolutionized and greatly facilitated access to reporting intensities, in contrast to H&M claims to the contrary. Second, because DYFI does not directly collect demographic data other than the observer’s location, any demographic analyses require extraordinary inferences, well outside the normal bounds of sociodemographic analyses. Third, independent of accessibility and the geographic distribution of contributions from the public, the macroseismic data collected are nonetheless representative of the shaking and impact at each location, of quality, rapid, and thus extremely useful. Lastly, H&M fail to cite critical and pertinent prior, highly relevant scholarly studies, and as such, they misrepresent the novelty of their own work as well as miss key practical matters detailed in those prior studies. Prior to rebutting what H&M claim DYFI does not do, I will remind the reader the ways in which DYFI excels.

Seismological Research Letters↗
USGS2005

Statistical approaches to the analysis of point count data: A little extra information can go a long way

Point counts are a standard sampling procedure for many bird species, but lingering concerns still exist about the quality of information produced from the method. It is well known that variation in observer ability and environmental conditions can influence the detection probability of birds in point counts, but many biologists have been reluctant to abandon point counts in favor of more intensive approaches to counting. However, over the past few years a variety of statistical and methodological developments have begun to provide practical ways of overcoming some of the problems with point counts. We describe some of these approaches, and show how they can be integrated into standard point count protocols to greatly enhance the quality of the information. Several tools now exist for estimation of detection probability of birds during counts, including distance sampling, double observer methods, time-depletion (removal) methods, and hybrid methods that combine these approaches. Many counts are conducted in habitats that make auditory detection of birds much more likely than visual detection. As a framework for understanding detection probability during such counts, we propose separating two components of the probability a bird is detected during a count into (1) the probability a bird vocalizes during the count and (2) the probability this vocalization is detected by an observer. In addition, we propose that some measure of the area sampled during a count is necessary for valid inferences about bird populations. This can be done by employing fixed-radius counts or more sophisticated distance-sampling models. We recommend any studies employing point counts be designed to estimate detection probability and to include a measure of the area sampled.

Book chapter↗
USGS2002

Long lead statistical forecasts of area burned in western U.S. wildfires by ecosystem province

A statistical forecast methodology exploits large-scale patterns in monthly U.S. Climatological Division Palmer Drought Severity Index (PDSI) values over a wide region and several seasons to predict area burned in western U.S. wildfires by ecosystem province a season in advance. The forecast model, which is based on canonical correlations, indicates that a few characteristic patterns determine predicted wildfire season area burned. Strong negative associations between anomalous soil moisture (inferred from PDSI) immediately prior to the fire season and area burned dominate in most higher elevation forested provinces, while strong positive associations between anomalous soil moisture a year prior to the fire season and area burned dominate in desert and shrub and grassland provinces. In much of the western U.S., above- and below-normal fire season forecasts were successful 57% of the time or better, as compared with a 33% skill for a random guess, and with a low probability of being surprised by a fire season at the opposite extreme of that forecast.

Conference Paper↗
USGS2021

Techniques to improve ecological interpretability of black box machine learning models

Statistical modeling of ecological data is often faced with a large number of variables as well as possible nonlinear relationships and higher-order interaction effects. Gradient boosted trees (GBT) have been successful in addressing these issues and have shown a good predictive performance in modeling nonlinear relationships, in particular in classification settings with a categorical response variable. They also tend to be robust against outliers. However, their black-box nature makes it difficult to interpret these models. We introduce several recently developed statistical tools to the environmental research community in order to advance interpretation of these black-box models. To analyze the properties of the tools, we applied gradient boosted trees to investigate biological health of streams within the contiguous USA, as measured by a benthic macroinvertebrate biotic index. Based on these data and a simulation study, we demonstrate the advantages and limitations of partial dependence plots (PDP), individual conditional expectation (ICE) curves and accumulated local effects (ALE) in their ability to identify covariate–response relationships. Additionally, interaction effects were quantified according to interaction strength (IAS) and Friedman’s H 2 "> H 2 statistic. Interpretable machine learning techniques are useful tools to open the black-box of gradient boosted trees in the environmental sciences. This finding is supported by our case study on the effect of impervious surface on the benthic condition, which agrees with previous results in the literature. Overall, the most important variables were ecoregion, bed stability, watershed area, riparian vegetation and catchment slope. These variables were also present in most identified interaction effects. In conclusion, graphical tools (PDP, ICE, ALE) enable visualization and easier interpretation of GBT but should be supported by analytical statistical measures. Future methodological research is needed to investigate the properties of interaction tests. Supplementary materials accompanying this paper appear on-line.

Journal of Agricultural, Biological, and Environme↗
USGS1989

Population trends of North American shorebirds based on the International Shorebird Survey

Shorebirds Charadrii are prime candidates for population decline because of their dependence on wetlands that are being lost at a rapid pace. Thirty-six of the 49 species of shorebirds that breed in North America spend most of the year in Latin America. Because populations of most species breed and winter at remote sites, it may be most feasible to monitor their numbers at migration stopovers. In this study, we used statistical trend analysis methods, developed for the North America Breeding Bird Survey, to analyze data on shorebird populations during southbound migration in the United States. Survey data were collected by volunteers in the International Shorebird Survey (ISS). The analyses indicate that whimbrels Numenius phaeopus , short-billed dowitchers Limnodromus griseus , and sanderlings Calidris alba have undergone statistically significant declines. Methodological concerns over both the ISS and the trend analysis procedures are discussed in detail and biological interpretations of the results are suggested.

Biological Conservation↗
USGS2022

Simulating burn severity maps at 30 meters in two forested regions in California

Climate change is altering wildfire behavior and vegetation regimes in California’s forested ecosystems. Present day fires are seeing an increase in high burn severity area and high severity patch size. The ability to predict future burn severity patterns would support better policy and land management decisions. Here we demonstrate a methodology to first, statistically estimate individual burn severity classes at 30 meters and second, cluster and smooth high severity patches onto a landscape. Our goal here was not to exactly replicate observed burn severity maps, but rather to utilize observed maps as one realization of a random process dependent on climate, topography, fire weather, and fuels, to inform creation of additional realizations through our simulation technique. We developed two sets of empirical models with two different vegetation datasets to test if coarse vegetation could accurately model for burn severity. While visual acuity can be used to assess the performance of our simulation process, we also employ the Ripley’s K function to compare spatial point processes at different scales to test if the simulation is capturing an appropriate amount of clustering. We utilize FRAGSTATS to obtain high severity patch metrics to test the contiguity of our high severity simulation. Ripley’s K function helped identify the number of clustering iterations and FRAGSTATS showed how different focal window sizes affected our ability to cluster high severity patches. High severity patch simulations was comparable between the coarse and fine resolution vegetation datasets. Improving our ability to simulate burn severity will help advance our understanding of the potential influence of land and fuels management on ecosystem-level response variables that are important for decision-makers. Simulated burn severity maps can support managing habitat and estimating risks of habitat loss, protecting infrastructure and homes, improving future wildfire emissions projections, and better mapping and planning for fuels treatment scenarios.

California↗
USGS2017

A multimodal wave spectrum-based approach for statistical downscaling of local wave climate

Characterization of wave climate by bulk wave parameters is insufficient for many coastal studies, including those focused on assessing coastal hazards and long-term wave climate influences on coastal evolution. This issue is particularly relevant for studies using statistical downscaling of atmospheric fields to local wave conditions, which are often multimodal in large ocean basins (e.g. the Pacific). Swell may be generated in vastly different wave generation regions, yielding complex wave spectra that are inadequately represented by a single set of bulk wave parameters. Furthermore, the relationship between atmospheric systems and local wave conditions is complicated by variations in arrival time of wave groups from different parts of the basin. Here, we address these two challenges by improving upon the spatiotemporal definition of the atmospheric predictor used in statistical downscaling of local wave climate. The improved methodology separates the local wave spectrum into “wave families,” defined by spectral peaks and discrete generation regions, and relates atmospheric conditions in distant regions of the ocean basin to local wave conditions by incorporating travel times computed from effective energy flux across the ocean basin. When applied to locations with multimodal wave spectra, including Southern California and Trujillo, Peru, the new methodology improves the ability of the statistical model to project significant wave height, peak period, and direction for each wave family, retaining more information from the full wave spectrum. This work is the base of statistical downscaling by weather types, which has recently been applied to coastal flooding and morphodynamic applications.

Journal of Physical Oceanography↗
USGS2004

Geospatial techniques for developing a sampling frame of watersheds across a region

Current land-management decisions that affect the persistence of native salmonids are often influenced by studies of individual sites that are selected based on judgment and convenience. Although this approach is useful for some purposes, extrapolating results to areas that were not sampled is statistically inappropriate because the sampling design is usually biased. Therefore, in recent investigations of coastal cutthroat trout (Oncorhynchus clarki clarki) located above natural barriers to anadromous salmonids, we used a methodology for extending the statistical scope of inference. The purpose of this paper is to apply geospatial tools to identify a population of watersheds and develop a probability-based sampling design for coastal cutthroat trout in western Oregon, USA. The population of mid-size watersheds (500-5800 ha) west of the Cascade Range divide was derived from watershed delineations based on digital elevation models. Because a database with locations of isolated populations of coastal cutthroat trout did not exist, a sampling frame of isolated watersheds containing cutthroat trout had to be developed. After the sampling frame of watersheds was established, isolated watersheds with coastal cutthroat trout were stratified by ecoregion and erosion potential based on dominant bedrock lithology (i.e., sedimentary and igneous). A stratified random sample of 60 watersheds was selected with proportional allocation in each stratum. By comparing watershed drainage areas of streams in the general population to those in the sampling frame and the resulting sample (n = 60), we were able to evaluate the how representative the subset of watersheds was in relation to the population of watersheds. Geospatial tools provided a relatively inexpensive means to generate the information necessary to develop a statistically robust, probability-based sampling design.

Conference Paper↗