A review of spatial approaches in road safety

Accident Analysis & Prevention, Journal Year: 2019, Volume and Issue: 135, P. 105323 - 105323

Published: Oct. 22, 2019

Language: Английский

---

sdmTMB: An R Package for Fast, Flexible, and User-Friendly Generalized Linear Mixed Effects Models with Spatial and Spatiotemporal Random Fields

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: March 27, 2022

Abstract Geostatistical spatial or spatiotemporal data are common across scientific fields. However, appropriate models to analyse these data, such as generalised linear mixed effects (GLMMs) with Gaussian Markov random fields (GMRFs), computationally intensive and challenging for many users implement. Here, we introduce the R package sdmTMB , which extends flexible interface familiar of lme4, glmmTMB mgcv include latent GMRFs using an SPDE-(stochastic partial differential equation) based approach. SPDE matrices constructed fmesher estimation is conducted via maximum marginal likelihood TMB Bayesian inference tmbstan rstan . We describe model explore case studies that illustrate ’s flexibility in implementing penalised smoothers, non-stationary processes (time-varying spatially varying coefficients), hurdle models, cross-validation anisotropy (directionally dependent correlation). Finally, compare functionality, speed, interfaces related software, demonstrating can be order magnitude faster than R- INLA hope will help open this useful class a wider field geostatistical analysts.

Language: Английский

---

spOccupancy: An R package for single‐species, multi‐species, and integrated spatial occupancy models

Methods in Ecology and Evolution, Journal Year: 2022, Volume and Issue: 13(8), P. 1670 - 1678

Published: May 16, 2022

Abstract Occupancy modelling is a common approach to assess species distribution patterns, while explicitly accounting for false absences in detection–nondetection data. Numerous extensions of the basic single‐species occupancy model exist multiple species, spatial autocorrelation and integrate data types. However, development specialized computationally efficient software incorporate such extensions, especially large datasets, scarce or absent. We introduce spOccupancy R package designed fit multi‐species spatially explicit models. all models within Bayesian framework using Pólya‐Gamma augmentation, which results fast inference. provides functionality integration datasets via joint likelihood framework. The leverages Nearest Neighbour Gaussian Processes account autocorrelation, enables potentially massive (e.g. 1,000s–100,000s sites). user‐friendly functions simulation, fitting, validation (by posterior predictive checks), comparison (using information criteria k‐fold cross‐validation) out‐of‐sample prediction. illustrate package's vignette, simulated analysis two bird case studies. platform variety single models, making it straightforward address detection biases even datasets.

Language: Английский

---

Optimizing the choice of a spatial weighting matrix in eigenvector‐based methods

Ecology, Journal Year: 2018, Volume and Issue: 99(10), P. 2159 - 2166

Published: July 24, 2018

Eigenvector-mapping methods such as Moran's eigenvector maps (MEM) are derived from a spatial weighting matrix (SWM) that describes the relations among set of sampled sites. The specification SWM is crucial step, but generally chosen arbitrarily, regardless sampling design characteristics. Here, we compare statistical performances different types SWMs (distance-based or graph-based) in contrasted realistic simulation scenarios. Then, present an optimization method and evaluate its compared to arbitrary choice most-widely used distance-based SWM. Results showed had lower power accuracy than other specifications, strongly underestimated signals. method, using correction procedure for multiple tests, correct type I error rate, higher Nevertheless, decreased when too many were compared, resulting trade-off between gain loss power. We advocate future studies should optimize small appropriate candidates. R functions implement available adespatial package detailed tutorial.

Language: Английский

---

Global Warming of Salmon and Trout Rivers in the Northwestern U.S.: Road to Ruin or Path Through Purgatory?

Transactions of the American Fisheries Society, Journal Year: 2018, Volume and Issue: 147(3), P. 566 - 587

Published: March 27, 2018

Abstract Large rivers constitute small portions of drainage networks but provide important migratory habitats and fisheries for salmon trout when where temperatures are sufficiently cold. Management conservation coldwater fishes in the current era rapid climate change require knowing how riverine thermal environments evolving potential detrimental biological impacts. Robust estimates warming rates, however, lacking due to limited long‐term temperature monitoring, so we compiled best available multidecadal records estimated trends at 391 sites 56,500‐km river network northwestern USA . Warming were prevalent during summer early fall months recent 20‐ 40‐year periods (0.18–0.35°C per decade 1996–2015 0.14–0.27°C 1976–2015), paralleled air trends, mediated by discharge regional local levels. To illustrate consequences later this century, trend used inform selection scenarios assess changes exposure adult Sockeye Salmon Oncorhynchus nerka migrating four population areas as well habitat shifts resident Brown Trout Salmo trutta Rainbow O. mykiss populations throughout region. Future 1–3°C would increase 5–16% (3–143 degree‐days) reduce thermally suitable 8–31% while causing their upstream shift. Effects those on persistence likely be context dependent, strategic restoration or adaptation strategies could ameliorate some impairments, effectiveness will tempered size rivers, high costs, pervasiveness effects. Most continue foreseeable future, it also appears inevitable that reaches gradually become too warm traditional habitats.

Language: Английский

---

Modeling urban growth using spatially heterogeneous cellular automata models: Comparison of spatial lag, spatial error and GWR

Computers Environment and Urban Systems, Journal Year: 2020, Volume and Issue: 81, P. 101459 - 101459

Published: Jan. 22, 2020

Language: Английский

---

Fuel treatment effectiveness in the context of landform, vegetation, and large, wind‐driven wildfires

Ecological Applications, Journal Year: 2020, Volume and Issue: 30(5)

Published: Feb. 22, 2020

Large wildfires (>50,000 ha) are becoming increasingly common in semiarid landscapes of the western United States. Although fuel reduction treatments used to mitigate potential wildfire effects, they can be overwhelmed wind-driven events with extreme fire behavior. We evaluated drivers severity and treatment effectiveness 2014 Carlton Complex, a record-setting complex north-central Washington State. Across varied topography, vegetation, distinct progressions, we combination simultaneous autoregression (SAR) random forest (RF) approaches model how mitigated severity. Predictor variables included type, time since treatment, topographic indices, vegetation fuels, weather summarized by progression interval. found that two spatial regression methods generally complementary instructive as combined approach for landscape analyses Simultaneous improves upon traditional linear models incorporating information about neighboring pixel burn severity, which avoids type I errors coefficient estimates incorrect inferences. Random modeling provides flexible environment capable capturing interactions nonlinearities while still accounting autocorrelation through use spatially explicit predictor variables. All areas burned higher proportions moderate high-severity during early but thin underburn, underburn only, past were more effective than thin-only pile treatments. Treatment units had much greater percentages unburned low area later progressions under milder conditions, differences between less pronounced. Our results provide evidence strategic placement fuels effectively reduce localized spread even severe weather. During located on leeward slopes tended have lower windward slopes. As managers evaluate options increasing resilience future climate change wildfires, may guided retrospective studies large events.

Language: Английский

---

Plant Endemism Centres and Biodiversity Hotspots in Greece

Biology, Journal Year: 2021, Volume and Issue: 10(2), P. 72 - 72

Published: Jan. 20, 2021

Biodiversity hotspots (BH) cover a small fraction of the Earth's surface, yet host numerous endemics. Human-induced biodiversity loss has been increasing worldwide, despite attempts to halt extinction crisis. There is thus an urgent need efficiently allocate available conservation funds in optimised prioritization scheme. Identifying BH and endemism centres (EC) therefore valuable tool planning. Even though Greece one most plant species-rich European countries, few studies have dealt with identification or EC none ever incorporated phylogenetic information extended national scale. Consequently, we are unaware extent that Special Areas Conservation (SAC) Natura 2000 network protect Greek diversity. Here, located for first time at scale framework, areas serving as EC, assessed effectiveness SAC safeguarding them. mainly near mountainous areas, supposedly floristically impoverished, such central Aegean islands. A critical re-assessment might be needed minimize risk endemics, by focusing efforts also on fall outside established SAC.

Language: Английский

---

Overview and introduction to development of non-ergodic earthquake ground-motion models

Bulletin of Earthquake Engineering, Journal Year: 2022, Volume and Issue: 21(11), P. 5121 - 5150

Published: Aug. 17, 2022

Abstract This paper provides an overview and introduction to the development of non-ergodic ground-motion models, GMMs. It is intended for a reader who familiar with standard approach developing ergodic starts brief summary GMMs then describes different methods that are used in emphasis on Gaussian process (GP) regression, as currently method preferred by most researchers contributing this special issue. Non-ergodic modeling requires definition locations source site characterizing systematic effects; domain divided into cells describing path effects. Modeling cell-specific anelastic attenuation GP, considerations constraints extrapolation also discussed. An updated unifying notation presented, which has been adopted authors

Language: Английский

---

Modeling the rarest of the rare: a comparison between multi‐species distribution models, ensembles of small models, and single‐species models at extremely low sample sizes

Ecography, Journal Year: 2023, Volume and Issue: 2023(6)

Published: April 10, 2023

Species distribution models are useful for estimating the and environmental preferences of rare species, but these same species challenging to model on account sparse data. We contrast a traditional single‐species approach (generalized linear models, GLMs) with two promising frameworks modeling species: ensembles small (ESMs), which average across simple models; multi‐species (MSDMs), allow rarer benefit from statistical ‘borrowing strength' more common species. Using virtual within community real we evaluated how accuracy was influenced by number occurrences (N = 2–64), niche breadth, similarity numerous species' niches. For discriminating between presence absence, ESMs just terms (ESM‐L) performed best N ≤ 4, whereas GLMs polynomial (ESM‐P) were ≥ 8. calibrating response influential variables, MSDM hierarchical communities (HMSC) ESM‐P niches similar those other dissimilar niches, did 8, no well calibrated smaller sample sizes. identifying uninfluential ESM‐L archetype (SAMs), type MSDM, Models narrow others had highest discrimination capacity compared generalist and/or ‘Borrowing in MSDMs can assist some inference tasks, does not necessarily improve predictions species; simpler, may be better at given task. The algorithm depends goal (discrimination versus calibration), size, breadth similarity. Keywords: borrowing strength, calibration, data‐deficient discrimination, presence–absence,

Language: Английский

---