Modern perspectives on statistics for spatio‐temporal data

Wikle, Christopher K.

doi:10.1002/wics.1341

Cited by 53 publications

(40 citation statements)

References 34 publications

(60 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…, Conn et al. , Wikle ) to better understand the spatiotemporal aggregation of juvenile delta smelt and will be best informed by additional experimental or observational data.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Prey availability, environmental constraints, and aggregation dictate population distribution of an imperiled fish

Simonis¹,

Merz

2019

Ecosphere

View full text Add to dashboard Cite

Accelerated global loss of fish species underscores an urgent need to elucidate constraints on spatial and temporal population patterns and to mitigate human‐mediated impacts accordingly. We analyzed 20 yr of trawl data using a spatiotemporally explicit, hierarchically Bayesian model to define density distributions of an imperiled anadromous fish—the delta smelt, Hypomesus transpacificus—in the heavily modified and highly used San Francisco Estuary. To reduce management reliance on an estuary‐wide annual population index that minimizes data resolution and ignores uncertainty, our model took a spatiotemporal approach and explicitly accounted for sampling and observation variation through both stochastic and deterministic elements. Our model demonstrated that juvenile smelt density has decreased over annual time and displays a distinct seasonal pattern (the combination of growth into the sampled size class, mortality, and movement into and out of the sampling frame) of increases in March–June followed by decreases in June–August. Smelt density was highly spatiotemporally autocorrelated and strongly tracked prey availability yet was also constrained by local hydrological factors (salinity, turbidity, velocity). Specifically, juvenile smelt preferred slightly saline, turbid, generally slow‐moving water with ample copepod prey. However, poor swimming capabilities reduced the capacity of juvenile smelt to mix throughout the estuary and find optimal habitat, emphasizing the importance of accounting for spatiotemporal autocorrelation in species distribution models. Further, whole‐estuary outflow appeared to influence the spatial distribution of covariates and juvenile smelt, such that smelt densities tended to peak closer to the marine zone at a lower maximal value when outflow was high. The predictions of our model accurately matched observed patterns of juvenile delta smelt catch while embracing uncertainty (although the predictions tended to exaggerate the total smelt catch variation via over‐predicted tail values). These results highlight that data‐driven improvements can be made to the analytical methods currently used to guide delta smelt management and lay the groundwork for mitigating the impacts of water use, flow regimes, habitat alteration, and climate change on an endangered species in a highly impacted and economically consequential habitat.

show abstract

“…, Conn et al. , Wikle ) to better understand the spatiotemporal aggregation of juvenile delta smelt and will be best informed by additional experimental or observational data.…”

Section: Discussionmentioning

confidence: 99%

“…We developed a hierarchical Bayesian implementation of a descriptive partially observed population model (Ver Hoef and Jansen , Cressie and Wikle , Conn et al. , Wikle ) based on the standard mixed‐effects model formulation of explicitly correlated data (Wikle , Sauer and Link , Ross et al. ).…”

Section: Statistical Population Modelmentioning

confidence: 99%

Prey availability, environmental constraints, and aggregation dictate population distribution of an imperiled fish

Simonis¹,

Merz

2019

Ecosphere

View full text Add to dashboard Cite

show abstract

“…A basic assumption in spatial analysis is that points located close to each other have more association than points located farther away (Logan 2012;Wickle 2015;Anselin et al 2006). Therefore, Moran's I spatial dependency test was performed to quantify the degree of spatial association among the neighboring observations and to determine whether spatial clustering of regression residuals is present (Tiefelsdorf 1998).…”

Section: Spatial Dependency Analysismentioning

confidence: 99%

Application of the double kernel density approach to the multivariate analysis of attributeless event point datasets

Zusman

Broitman

Portnov

2015

Lett Spat Resour Sci

View full text Add to dashboard Cite

Attributeless event point datasets (AEPDs) are datasets composed of discrete events or observations defined by their geographical location only and lacking any other additional attributes. Examples of such datasets include spotted criminal events, road accidents and residential locations of disease patients. A commonly used approach to the analysis of such datasets involves their aggregation into predefined areal units, such as neighborhoods or census tracts. However, this approach does not perform effectively when the events of interests are geographically localized and the number of areal units available for aggregation is small. An alternative approach to the analysis of AEPDs is based on double kernel density (DKD) smoothing, according to which events of interest are transformed into continuous density surfaces and then normalized by the density of the entire population from which the events of interest are drawn. In the present study, the applicability of the DKD approach to multivariate analysis is tested for estimation consistency, sensitivity to the number of input observations and potential bias attributed to the spatial dependency of neighboring observations. Our analysis indicates that the DKD approach provides reasonably stable and consistent estimates, if the following three preconditions are met: (a) the kernel estimation parameters are properly defined, (b) the number of reference points, used for transformation of continuous DKD surfaces into discrete observations, is sufficiently large, and (c) the spatial dependency of neighboring observations is taken into account using spatial analysis tools.

show abstract

“…These models, at least when implemented in a way that fully accounts for uncertainty in data, process, and parameters, can be quite computationally challenging, mainly due to the very large number of parameters that must be estimated. Solutions to this challenge require reducing the dimension of the state space, regularizing the parameter space, the incorporating of additional information (prior knowledge), and novel computational approaches (see the summary in Wikle, ). Parsimonious alternatives include analog methods (e.g., McDermott & Wikle, ; Zhao & Giannakis, ) and individual (agent)‐based models (e.g., Hooten & Wikle, ).…”

Section: Introductionmentioning

confidence: 99%

Deep echo state networks with uncertainty quantification for spatio‐temporal forecasting

McDermott

Wikle

2018

Environmetrics

Self Cite

View full text Add to dashboard Cite

Long‐lead forecasting for spatio‐temporal systems can entail complex nonlinear dynamics that are difficult to specify a priori. Current statistical methodologies for modeling these processes are often highly parameterized and, thus, challenging to implement from a computational perspective. One potential parsimonious solution to this problem is a method from the dynamical systems and engineering literature referred to as an echo state network (ESN). ESN models use reservoir computing to efficiently compute recurrent neural network forecasts. Moreover, multilevel (deep) hierarchical models have recently been shown to be successful at predicting high‐dimensional complex nonlinear processes, particularly those with multiple spatial and temporal scales of variability (such as those we often find in spatio‐temporal environmental data). Here, we introduce a deep ensemble ESN (D‐EESN) model. Despite the incorporation of a deep structure, the presented model is computationally efficient. We present two versions of this model for spatio‐temporal processes that produce forecasts and associated measures of uncertainty. The first approach utilizes a bootstrap ensemble framework, and the second is developed within a hierarchical Bayesian framework (BD‐EESN). This more general hierarchical Bayesian framework naturally accommodates non‐Gaussian data types and multiple levels of uncertainties. The methodology is first applied to a data set simulated from a novel non‐Gaussian multiscale Lorenz‐96 dynamical system simulation model and, then, to a long‐lead United States (U.S.) soil moisture forecasting application. Across both applications, the proposed methodology improves upon existing methods in terms of both forecast accuracy and quantifying uncertainty.

show abstract

Modern perspectives on statistics for spatio‐temporal data

Cited by 53 publications

References 34 publications

Prey availability, environmental constraints, and aggregation dictate population distribution of an imperiled fish

Prey availability, environmental constraints, and aggregation dictate population distribution of an imperiled fish

Application of the double kernel density approach to the multivariate analysis of attributeless event point datasets

Deep echo state networks with uncertainty quantification for spatio‐temporal forecasting

Contact Info

Product

Resources

About