Multisite Time Series Research Articles

Ground- and ship-based microwave radiometer measurements during EUREC4A

Abstract. During the EUREC4A field study, microwave radiometric measurements were performed at Barbados Cloud Observatory (BCO) and aboard RV Meteor and RV Maria S Merian in the downstream winter trades of the North Atlantic. We present retrieved integrated water vapor (IWV), liquid water path (LWP), and temperature and humidity profiles as a unified, quality-controlled, multi-site data set on a 3 s temporal resolution for a core period between 19 January and 14 February 2020 in which all instruments were operational. Multi-channel radiometric measurements were performed at BCO and aboard RV Meteor between 22 and 31 GHz (K-band) and from 51 to 58 GHz (V-band). Combined radar–radiometer measurements of a W-band Doppler radar with a single-channel radiometer instrument were conducted at 89 GHz aboard RV Meteor and RV Maria S Merian. We present a novel retrieval method to retrieve LWP from single-channel 89 GHz measurements, evaluate retrieved quantities with independent measurements, and analyze retrieval uncertainties by site and instrument intercomparison. Mean IWV conditions of 31.8 kg m−2 match independent radiosoundings at BCO with a root-mean-square difference of 1.1 kg m−2. Mean LWP conditions in confidently liquid cloudy, non-precipitating conditions ranged between 63.1 g m−2 at BCO and 46.8 g m−2 aboard RV Maria S Merian. Aboard the ships, 90 % of LWP was below 120 g m−2 with a 30 % uncertainty for LWP of 50 g m−2. Up to 20 % of confidently liquid cloudy profiles ranged below the LWP detection limit due to optically thin clouds. The data set comprises of processed raw data (Level 1), full quality-controlled post-processed instrument data (Level 2), a unified temporal resolution (Level 3), and a ready-to-use multi-site time series of IWV and LWP (Level 4), available to the public via AERIS (https://doi.org/10.25326/454##v2.0; Schnitt et al., 2023a). The data set complements the airborne LWP measurements conducted during EUREC4A and provides a unique benchmark tool for satellite evaluation and model–observation studies.

Conditional Poisson Regression with Random Effects for the Analysis of Multi-site Time Series Studies.

The analysis of time series studies linking daily counts of a health indicator with environmental variables (e.g., mortality or hospital admissions with air pollution concentrations or temperature; or motor vehicle crashes with temperature) is usually conducted with Poisson regression models controlling for long-term and seasonal trends using temporal strata. When the study includes multiple zones, analysts usually apply a two-stage approach: first, each zone is analyzed separately, and the resulting zone-specific estimates are then combined using meta-analysis. This approach allows zone-specific control for trends. A one-stage approach uses spatio-temporal strata and could be seen as a particular case of the case-time series framework recently proposed. However, the number of strata can escalate very rapidly in a long time series with many zones. A computationally efficient alternative is to fit a conditional Poisson regression model, avoiding the estimation of the nuisance strata. To allow for zone-specific effects, we propose a conditional Poisson regression model with a random slope, although available frequentist software does not implement this model. Here, we implement our approach in the Bayesian paradigm, which also facilitates the inclusion of spatial patterns in the effect of interest. We also provide a possible extension to deal with overdispersed data. We first introduce the equations of the framework and then illustrate their application to data from a previously published study on the effects of temperature on the risk of motor vehicle crashes. We provide R code and a semi-synthetic dataset to reproduce all analyses presented.

Short-Term Effects of Climate Variability on Childhood Diarrhoea in Bangladesh: Multi-Site Time-Series Regression Analysis.

The aim of this study was to estimate the effects of climate on childhood diarrhoea hospitalisations across six administrative divisions in Bangladesh and to provide scientific evidence for local health authorities for disease control and prevention. Fortnightly hospital admissions (August/2013-June/2017) for diarrhoea in children under five years of age, and fortnightly average maximum temperature, relative humidity and rainfall recordings for six administrative divisions were modelled using negative binomial regression with distributed lag linear terms. Flexible spline functions were used to adjust models for seasonality and long-term trends. During the study period, 25,385 diarrhoea cases were hospitalised. Overall, each 1 °C rise in maximum temperature increased diarrhoea hospitalisations by 4.6% (IRR = 1.046; 95% CI, 1.007-1.088) after adjusting for seasonality and long-term trends in the unlagged model. Using lagged effects of maximum temperature, and adjusting for relative humidity and rainfall for each of the six administrative divisions, the relationship between maximum temperature and diarrhoea hospitalisations varied between divisions, with positive and negative effect estimates. The temperature-diarrhoea association may be confounded by seasonality and long-term trends. Our findings are a reminder that the effects of climate change may be heterogeneous across regions, and that tailored diarrhoea prevention strategies need to consider region-specific recommendations rather than relying on generic guidelines.

Assessing the causal effects of a stochastic intervention in time series data: are heat alerts effective in preventing deaths and hospitalizations?

The methodological development of this article is motivated by the need to address the following scientific question: does the issuance of heat alerts prevent adverse health effects? Our goal is to address this question within a causal inference framework in the context of time series data. A key challenge is that causal inference methods require the overlap assumption to hold: each unit (i.e., a day) must have a positive probability of receiving the treatment (i.e., issuing a heat alert on that day). In our motivating example, the overlap assumption is often violated: the probability of issuing a heat alert on a cooler day is near zero. To overcome this challenge, we propose a stochastic intervention for time series data which is implemented via an incremental time-varying propensity score (ItvPS). The ItvPS intervention is executed by multiplying the probability of issuing a heat alert on day $t$-conditional on past information up to day $t$-by an odds ratio $\delta_t$. First, we introduce a new class of causal estimands, which relies on the ItvPS intervention. We provide theoretical results to show that these causal estimands can be identified and estimated under a weaker version of the overlap assumption. Second, we propose nonparametric estimators based on the ItvPS and derive an upper bound for the variances of these estimators. Third, we extend this framework to multisite time series using a spatial meta-analysis approach. Fourth, we show that the proposed estimators perform well in terms of bias and root mean squared error via simulations. Finally, we apply our proposed approach to estimate the causal effects of increasing the probability of issuing heat alerts on each warm-season day in reducing deaths and hospitalizations among Medicare enrollees in 2837 US counties.

Temperatures and health costs of emergency department visits: A multisite time series study in China

BackgroundEvidence is limited regarding the association between temperatures and health costs. ObjectivesWe tried to investigate the association between temperatures and emergency department visits (EDVs) costs in China. MethodsDaily data on EDVs costs, weather, air pollution were collected from 17 sites in China during 2014–2018. A quasi–Poisson generalized additive regression with distributed lag nonlinear model was applied to assess the temperature–EDVs cost association. Random-effect meta-analysis was used to pool the estimates from each site. Attributable fractions and national attributable EDVs costs due to heat and cold were calculated. ResultsRelative risk (RR) due to extreme heat over 0–7 lag days was 1.14 [95% confidence intervals (CI): 1.08–1.19] and 1.11 (95% CI: 1.07–1.16) for EDVs examination (including treatment) and medicine cost, respectively. People aged 18–44 and those with genitourinary diseases were at higher risk from heat. 0.72% of examination cost and 0.57% of medicine cost were attributed to extreme heat, costing 274 million Chinese Yuan annually. Moderate heat had lower RR but higher attributable fraction of EDVs costs. Exposure to extreme cold over 0–21 lag days increased the risk of medicine cost for people aged 18–44 [RR: 1.30 (95% CI: 1.10–1.55)] and those with respiratory diseases [RR: 1.56 (95% CI: 1.14–2.14)], but had non-statistically significant attributable fraction of the total EDVs cost. ConclusionsExposure to heat and cold resulted in remarkable health costs. More resources and preparedness are needed to tackle such a challenge as our climate is rapidly changing.

Multivariate Streamflow Simulation Using Hybrid Deep Learning Models.

Reliable and accurate streamflow simulation has a vital role in water resource development, mainly in agriculture, environment, domestic water supply, hydropower generation, flood control, and early warning systems. In this context, these days, deep learning algorithms have got enormous attention due to their high-performance simulation capacity. In this study, we compared multilayer perceptron (MLP), long short-term memory (LSTM), and gated recurrent unit (GRU) with the proposed new hybrid models, including CNN-LSTM and CNN-GRU. Hence, we can simulate one-step daily streamflow in different agroclimatic conditions, rolling time windows, and a range of variable input combinations. The analysis used daily multivariate and multisite time series data collected from Awash River Basin (Borkena watershed: Ethiopia) and Tiber River Basin (Upper Tiber River Basin: Italy) stations. The datasets were subjected to rigorous quality control processes. Consequently, it rolled to a different time lag to remove noise in the time series and further split into training and testing datasets using a ratio of 80 : 20, respectively. Finally, the results showed that integrating the GRU layer with the convolutional layer and using monthly rolled average daily input time series could substantially improve the simulation of streamflow time series.

Speed and directedness predict colonization sequence post-disturbance.

Foundational ecological models characterize dispersal with two behavioral traits, speed and directional bias. We hypothesized that these two traits can predict the order of colonization by fishes in a heterogenous landscape. Colonization patterns following hydrological disturbance were documented from a 20-year multi-site time series of marsh fish, and we evaluated the ability of a two-parameter model to predict these patterns. The maximum aerobic swimming speed (UCRIT) for six coexisting fish species were estimated using endurance tests; field estimates of directedness and swimming speed were previously documented using encounter samplers. We incorporated interspecific variation in speed, direction, and density into several Agent Based Modelsto simulate dispersal following disturbance. Six virtual "species" with varying levels of directedness, "swam" in an artificial environment to reach a refuge habitat. The time of first arrival for each species was saved at the end of each run and used to calculate the probability of arrival order. Our simulated results generated predictions on order of arrival consistent with observed colonization patterns in our long-term dataset. Swim tunnel results revealed that fast (high UCRIT) estimates were characteristic of early colonizing species; whereas, slow (low UCRIT) estimates were characteristic of late colonizing species. Directional bias better predicted order of arrival than speed and was robust to inter-specific variation in density. This study demonstrated that two parameters were adequate to predict the order of species colonization in a complex landscape. These results support the use of relatively simple trait-based models to generate realistic community assembly dynamics.

Arthropod decline in grasslands and forests is associated with landscape-level drivers.

Recent reports of local extinctions of arthropod species1, and of massive declines in arthropod biomass2, point to land-use intensification as a major driver of decreasing biodiversity. However, to our knowledge, there are no multisite time series of arthropod occurrences across gradients of land-use intensity with which to confirm causal relationships. Moreover, it remains unclear which land-use types and arthropod groups are affected, and whether the observed declines in biomass and diversity are linked to one another. Here we analyse data from more than 1million individual arthropods (about 2,700species), from standardized inventories taken between 2008 and 2017 at 150grassland and 140forest sites in 3regions of Germany. Overall gamma diversity in grasslands and forests decreased over time, indicating loss of species across sites and regions. In annually sampled grasslands, biomass, abundance and number of species declined by 67%, 78% and 34%, respectively. The decline was consistent across trophic levels and mainly affected rare species; its magnitude was independent of local land-use intensity. However, sites embedded in landscapes with a higher cover of agricultural land showed a stronger temporal decline. In 30forest sites with annual inventories, biomass and species number-but not abundance-decreased by 41% and 36%, respectively. This was supported by analyses of all forest sites sampled in three-year intervals. The decline affected rare and abundant species, and trends differed across trophic levels. Our results show that there are widespread declines in arthropod biomass, abundance and the number of species across trophic levels. Arthropod declines in forests demonstrate that loss is not restricted to open habitats. Our results suggest that major drivers of arthropod decline act at larger spatial scales, and are (at least for grasslands) associated with agriculture at the landscape level. This implies that policies need to address the landscape scale to mitigate the negative effects of land-use practices.

Scaling-up implementation in community hospitals: a multisite interrupted time series design of the Mobilization of Vulnerable Elders (MOVE) program in Alberta

BackgroundAs the population ages, older hospitalized patients are at increased risk for hospital-acquired morbidity. The Mobilization of Vulnerable Elders (MOVE) program is an evidence-informed early mobilization intervention that was previously evaluated in Ontario, Canada. The program was effective at improving mobilization rates and decreasing length of stay in academic hospitals. The aim of this study was to scale-up the program and conduct a replication study evaluating the impact of the evidence-informed mobilization intervention on various units in community hospitals within a different Canadian province.MethodsThe MOVE program was tailored to the local context at four community hospitals in Alberta, Canada. The study population was patients aged 65 years and older who were admitted to medicine, surgery, rehabilitation and intensive care units between July 2015 and July 2016. The primary outcome was patient mobilization measured by conducting visual audits twice a week, three times a day. The secondary outcomes included hospital length of stay obtained from hospital administrative data, and perceptions of the intervention assessed through a qualitative assessment. Using an interrupted time series design, the intervention was evaluated over three time periods (pre-intervention, during, and post-intervention).ResultsA total of 3601 patients [mean age 80.1 years (SD = 8.4 years)] were included in the overall analysis. There was a significant increase in mobilization at the end of the intervention period compared to pre-intervention, with 6% more patients out of bed (95% confidence interval (CI) 1, 11; p-value = 0.0173). A decreasing trend in median length of stay was observed, where patients on average stayed an estimated 3.59 fewer days (95%CI -15.06, 7.88) during the intervention compared to pre-intervention period.ConclusionsMOVE is a low-cost, effective and adaptable intervention that improves mobilization in older hospitalized patients. This intervention has been replicated and scaled up across various units and hospital settings.

Climate Change Drives Widespread and Rapid Thermokarst Development in Very Cold Permafrost in the Canadian High Arctic

AbstractClimate warming in regions of ice‐rich permafrost can result in widespread thermokarst development, which reconfigures the landscape and damages infrastructure. We present multisite time series observations which couple ground temperature measurements with thermokarst development in a region of very cold permafrost. In the Canadian High Arctic between 2003 and 2016, a series of anomalously warm summers caused mean thawing indices to be 150–240% above the 1979–2000 normal resulting in up to 90 cm of subsidence over the 12‐year observation period. Our data illustrate that despite low mean annual ground temperatures, very cold permafrost (<−10 °C) with massive ground ice close to the surface is highly vulnerable to rapid permafrost degradation and thermokarst development. We suggest that this is due to little thermal buffering from soil organic layers and near‐surface vegetation, and the presence of near‐surface ground ice. Observed maximum thaw depths at our sites are already exceeding those projected to occur by 2090 under representative concentration pathway version 4.5.

Simulation of the minimum annual river flows based on the RCP climatic scenario, time horizon up to 2060-2080 and the Kaczawa River

In this paper a new simulations of minimum daily flow for Kaczawa River a left side tributary of the Odra River in south-west Poland are presented. Generated data were made based on very long series of 35 years of observed data and 24 sites of meteorological stations for south-west Poland gathered from the the Institute of Meteorology and Water Management National Research Institute (IMGW). For the data generation the spatial weather generator SWGEN producing the multisite daily time series was applied. Data were generated for the present (the year 2000 are used as a background) as well for future climate condition for 2060 and 2080 according Representative Concentration Pathways (RCPs) scenarios. The flow simulation in the river catchment is made using MIKE SHE hydrological model. Simulations are done for 2060 and 2080. The large number of new simulated series determined by the lead time, two climate change scenarios (RCP4.5 and RCP6.0), and number of generated years (1000 for each case) is equal to 5000 for a single station. Finally, Lognormal Pdf function for the minimum flow is presented as well probability of exceedance of minimum values.

Evaluation of multisite synthetic data generated by spatial weather generator and long climate data series

In this paper a new validation test for the spatial weather generator SWGEN producing the multisite daily time series of solar radiation, temperature and precipitation is presented. The method was tested by comparing statistics of 1000 years of generated data with extra long series of 35 years of observed weather parameters and 24 sites of meteorological stations for south-west Poland. The method evaluation showed that the means (sums) and variances of generated data were comparable with observed climatic data aggregated for months, seasons and years.

Using BiSON to detect solar internal g-modes

The unambiguous detection of individual solar internal g modes continues to elude us. With the aid of new additions to calibration procedures, as well as updated methods to combine multi-site time series more effectively, the noise and signal detection threshold levels in the low-frequency domain (where the g modes are expected to be found) have been greatly improved. In the BiSON 23-year dataset these levels now rival those of GOLF, and with much greater frequency resolution available, due to the long time series, there is an opportunity to place more constraints on the upper limits of individual g mode amplitudes. Here we detail recent work dedicated to the challenges of observing low-frequency oscillations using a ground-based network, including the role of the window function as well as the effect of calibration on the low frequency domain.

Multi-site time series analysis of acute effects of multiple air pollutants on respiratory mortality: A population-based study in Beijing, China

In large cities in China, the traffic-related air pollution has become the focus of attention, and its adverse effects on health have raised public concerns. We conducted a study to quantify the association between exposure to three major traffic-related pollutants — particulate matter <10μm in aerodynamic diameter (PM10), carbon monoxide (CO) and nitrogen dioxide (NO2) and the risk of respiratory mortality in Beijing, China at a daily spatiotemporal resolution. We used the generalized additive models (GAM) with natural splines and principal component regression method to associate air pollutants with daily respiratory mortality, covariates and confounders. The GAM analysis adjusting for the collinearity among pollutants indicated that PM10, CO and NO2 had significant effects on daily respiratory mortality in Beijing. An interquartile range increase in 2-day moving averages concentrations of day 0 and day 1 of PM10, CO and NO2 corresponded to 0.99 [95% confidence interval (CI): 0.30, 1.67], 0.89 (95% CI: 0.27, 1.51) and 0.95 (95% CI: 0.29, 1.61) percent increase in daily respiratory mortality, respectively. The effects were varied across the districts. The strongest effects were found in two rural districts and one suburban district but significant in only one district. In conclusion, high level of several traffic-related air pollutants is associated with an increased risk of respiratory mortality in Beijing over a short-time period. The high risk found in rural areas suggests a potential susceptible sub-population with undiagnosed respiratory diseases in these areas. Although the rural areas have relatively lower air pollution levels, they deserve more attention to respiratory disease prevention and air pollution reduction.

Spatial variable selection methods for investigating acute health effects of fine particulate matter components.

Multi-site time series studies have reported evidence of an association between short term exposure to particulate matter (PM) and adverse health effects, but the effect size varies across the United States. Variability in the effect may partially be due to differing community level exposure and health characteristics, but also due to the chemical composition of PM which is known to vary greatly by location and time. The objective of this article is to identify particularly harmful components of this chemical mixture. Because of the large number of highly-correlated components, we must incorporate some regularization into a statistical model. We assume that, at each spatial location, the regression coefficients come from a mixture model with the flavor of stochastic search variable selection, but utilize a copula to share information about variable inclusion and effect magnitude across locations. The model differs from current spatial variable selection techniques by accommodating both local and global variable selection. The model is used to study the association between fine PM (PM <2.5μm) components, measured at 115 counties nationally over the period 2000-2008, and cardiovascular emergency room admissions among Medicare patients.

An effective three-step algorithm for multi-site generation of stochastic weekly hydrological time series

A new method is presented to generate stationary multi-site hydrological time series. The proposed method can handle flexible time-step length, and it can be applied to both continuous and intermittent input series. The algorithm is a departure from standard decomposition models and the Box-Jenkins approach. It relies instead on the recent advances in statistical science that deal with generation of correlated random variables with arbitrary statistical distribution functions. The proposed method has been tested on 11 historic weekly input series, of which the first seven contain flow data and the last four have precipitation data. The article contains an extensive review of the results.Editor D. KoutsoyiannisCitation Ilich, N., 2014. An effective three-step algorithm for multi-site generation of stochastic weekly hydrological time series. Hydrological Sciences Journal, 59 (1), 85–98.

Reduced Hierarchical Models with Application to Estimating Health Effects of Simultaneous Exposure to Multiple Pollutants

Hierarchical models (HM) have been used extensively in multisite time series studies of air pollution and health to estimate health effects of a single pollutant adjusted for other pollutants and other time-varying factors. Recently, Environmental Protection Agency (EPA) has called for research quantifying health effects of simultaneous exposure to many air pollutants. However, straightforward application of HM in this context is challenged by the need to specify a random-effect distribution on a high-dimensional vector of nuisance parameters. Here we introduce reduced HM as a general statistical approach for analyzing correlated data with many nuisance parameters. For reduced HM we first calculate the integrated likelihood of the parameter of interest (e.g. excess number of deaths attributed to simultaneous exposure to high levels of many pollutants), and we then specify a flexible random-effect distribution directly on this parameter. Simulation studies show that the reduced HM performs comparably to the full HM in many scenarios, and even performs better in some cases, particularly when the multivariate random-effect distribution of the full HM is misspecified. Methods are applied to estimate relative risks of cardiovascular hospital admissions associated with simultaneous exposure to elevated levels of particulate matter and ozone in 51 US counties during 1999-2005.

A Meta-Analysis and Multisite Time-Series Analysis of the Differential Toxicity of Major Fine Particulate Matter Constituents

Health risk assessments of particulate matter less than 2.5 μm in diameter (PM(2.5)) often assume that all constituents of PM(2.5) are equally toxic. While investigators in previous epidemiologic studies have evaluated health risks from various PM(2.5) constituents, few have conducted the analyses needed to directly inform risk assessments. In this study, the authors performed a literature review and conducted a multisite time-series analysis of hospital admissions and exposure to PM(2.5) constituents (elemental carbon, organic carbon matter, sulfate, and nitrate) in a population of 12 million US Medicare enrollees for the period 2000-2008. The literature review illustrated a general lack of multiconstituent models or insight about probabilities of differential impacts per unit of concentration change. Consistent with previous results, the multisite time-series analysis found statistically significant associations between short-term changes in elemental carbon and cardiovascular hospital admissions. Posterior probabilities from multiconstituent models provided evidence that some individual constituents were more toxic than others, and posterior parameter estimates coupled with correlations among these estimates provided necessary information for risk assessment. Ratios of constituent toxicities, commonly used in risk assessment to describe differential toxicity, were extremely uncertain for all comparisons. These analyses emphasize the subtlety of the statistical techniques and epidemiologic studies necessary to inform risk assessments of particle constituents.

Estimating the acute health effects of coarse particulate matter accounting for exposure measurement error

In air pollution epidemiology, there is a growing interest in estimating the health effects of coarse particulate matter (PM) with aerodynamic diameter between 2.5 and 10 μm. Coarse PM concentrations can exhibit considerable spatial heterogeneity because the particles travel shorter distances and do not remain suspended in the atmosphere for an extended period of time. In this paper, we develop a modeling approach for estimating the short-term effects of air pollution in time series analysis when the ambient concentrations vary spatially within the study region. Specifically, our approach quantifies the error in the exposure variable by characterizing, on any given day, the disagreement in ambient concentrations measured across monitoring stations. This is accomplished by viewing monitor-level measurements as error-prone repeated measurements of the unobserved population average exposure. Inference is carried out in a Bayesian framework to fully account for uncertainty in the estimation of model parameters. Finally, by using different exposure indicators, we investigate the sensitivity of the association between coarse PM and daily hospital admissions based on a recent national multisite time series analysis. Among Medicare enrollees from 59 US counties between the period 1999 and 2005, we find a consistent positive association between coarse PM and same-day admission for cardiovascular diseases.

Impact of climate change on ambient ozone level and mortality in southeastern United States.

There is a growing interest in quantifying the health impacts of climate change. This paper examines the risks of future ozone levels on non-accidental mortality across 19 urban communities in Southeastern United States. We present a modeling framework that integrates data from climate model outputs, historical meteorology and ozone observations, and a health surveillance database. We first modeled present-day relationships between observed maximum daily 8-hour average ozone concentrations and meteorology measured during the year 2000. Future ozone concentrations for the period 2041 to 2050 were then projected using calibrated climate model output data from the North American Regional Climate Change Assessment Program. Daily community-level mortality counts for the period 1987 to 2000 were obtained from the National Mortality, Morbidity and Air Pollution Study. Controlling for temperature, dew-point temperature, and seasonality, relative risks associated with short-term exposure to ambient ozone during the summer months were estimated using a multi-site time series design. We estimated an increase of 0.43 ppb (95% PI: 0.14–0.75) in average ozone concentration during the 2040’s compared to 2000 due to climate change alone. This corresponds to a 0.01% increase in mortality rate and 45.2 (95% PI: 3.26–87.1) premature deaths in the study communities attributable to the increase in future ozone level.