Unmonitored Sites Research Articles

Deriving global parameter estimates for the Noah land surface model using FLUXNET and machine learning

AbstractWith their origins in numerical weather prediction and climate modeling, land surface models aim to accurately partition the surface energy balance. An overlooked challenge in these schemes is the role of model parameter uncertainty, particularly at unmonitored sites. This study provides global parameter estimates for the Noah land surface model using 85 eddy covariance sites in the global FLUXNET network. The at‐site parameters are first calibrated using a Latin Hypercube‐based ensemble of the most sensitive parameters, determined by the Sobol method, to be the minimum stomatal resistance (rs,min), the Zilitinkevich empirical constant (Czil), and the bare soil evaporation exponent (fxexp). Calibration leads to an increase in the mean Kling‐Gupta Efficiency performance metric from 0.54 to 0.71. These calibrated parameter sets are then related to local environmental characteristics using the Extra‐Trees machine learning algorithm. The fitted Extra‐Trees model is used to map the optimal parameter sets over the globe at a 5 km spatial resolution. The leave‐one‐out cross validation of the mapped parameters using the Noah land surface model suggests that there is the potential to skillfully relate calibrated model parameter sets to local environmental characteristics. The results demonstrate the potential to use FLUXNET to tune the parameterizations of surface fluxes in land surface models and to provide improved parameter estimates over the globe.

Predicting sulphur and nitrogen deposition using a simple statistical method

Data from 32 long-term (1994–2012) monitoring sites were used to assess temporal development and spatial variability of sulphur (S) and inorganic nitrogen (N) concentrations in bulk precipitation, and S in throughfall, for the Czech Republic. Despite large variance in absolute S and N concentration/deposition among sites, temporal coherence using standardised data (Z score) was demonstrated. Overall significant declines of SO4 concentration in bulk and throughfall precipitation, as well as NO3 and NH4 concentration in bulk precipitation, were observed. Median Z score values of bulk SO4, NO3 and NH4 and throughfall SO4 derived from observations and the respective emission rates of SO2, NOx and NH3 in the Czech Republic and Slovakia showed highly significant (p < 0.001) relationships. Using linear regression models, Z score values were calculated for the whole period 1900–2012 and then back-transformed to give estimates of concentration for the individual sites. Uncertainty associated with the concentration calculations was estimated as 20% for SO4 bulk precipitation, 22% for throughfall SO4, 18% for bulk NO3 and 28% for bulk NH4. The application of the method suggested that it is effective in the long-term reconstruction and prediction of S and N deposition at a variety of sites. Multiple regression modelling was used to extrapolate site characteristics (mean precipitation chemistry and its standard deviation) from monitored to unmonitored sites. Spatially distributed temporal development of S and N depositions were calculated since 1900. The method allows spatio-temporal estimation of the acid deposition in regions with extensive monitoring of precipitation chemistry.

Operational Surface Water Detection and Monitoring Using Radarsat 2

Traditional on-site methods for mapping and monitoring surface water extent are prohibitively expensive at a national scale within Canada. Despite successful cost-sharing programs between the provinces and the federal government, an extensive number of water features within the country remain unmonitored. Particularly difficult to monitor are the potholes in the Canadian Prairie region, most of which are ephemeral in nature and represent a discontinuous flow that influences water pathways, runoff response, flooding and local weather. Radarsat-2 and the Radarsat Constellation Mission (RCM) offer unique capabilities to map the extent of water bodies at a national scale, including unmonitored sites, and leverage the current infrastructure of the Meteorological Service of Canada to monitor water information in remote regions. An analysis of the technical requirements of the Radarsat-2 beam mode, polarization and resolution is presented. A threshold-based procedure to map locations of non-vegetated water bodies after the ice break-up is used and complemented with a texture-based indicator to capture the most homogeneous water areas and automatically delineate their extents. Some strategies to cope with the radiometric artifacts of noise inherent to Synthetic Aperture Radar (SAR) images are also discussed. Our results show that Radarsat-2 Fine mode can capture 88% of the total water area in a fully automated way. This will greatly improve current operational procedures for surface water monitoring information and impact a number of applications including weather forecasting, hydrological modeling, and drought/flood predictions.

Spatial–temporal interpolation of non methane hydrocarbon levels in Kuwait

ABSTRACTThis article handles the prediction of hourly concentrations ofnon methane hydrocarbon (NMHC) pollutants at 15 unmonitored sites in Kuwait using the data recorded from 6 monitored stations at successive time points. The trend model depends on hourly meteorological variables and seasonal effects. The stochasticcomponent of the trend model which has spatiotemporal features is modeled as autoregressive temporal process. A spatial predictive distribution for residuals of the AR model is developed for the unmonitored sites. By transforming the predicted residuals back to the original data scales, we impute Kuwait’s hourly NMHC field.

Airborne Thermal Data Identifies Groundwater Discharge at the North-Western Coast of the Dead Sea

A qualitative and quantitative monitoring of groundwater discharge was conducted based on an airborne thermal campaign undertaken along the north-western coast of the Dead Sea in January 2011 to contribute to the relatively scarce information on groundwater discharge to date in the region. The application of airborne thermal data exploits thermal contrasts that exist between discharging groundwater and background sea surface temperatures of the Dead Sea. Using these contrasts, 72 discharge sites were identified from which only 42 were known from previous in situ measurements undertaken at terrestrial springs by the Israel Hydrological Service. Six of these sites represent submarine springs and at a further 24 locations groundwater appears to seep through the sediment. Although the abundance of groundwater seepage sites suggests a significant, but so far unknown groundwater source, the main contribution appears to originate from terrestrial springs. In an attempt to provide a quantitative approach for terrestrial springs, a linear bootstrap regression model between in situ spring discharge and respective thermal discharge plumes (r2 = 0.87 p &lt; 0.001) is developed and presented here. While the results appear promising and could potentially be applied to derive discharge values at unmonitored sites, several influence factors need to be clarified before a robust and reliable model to efficiently derive a complete quantitative picture of groundwater discharge can be proposed.

Modelling the Spatial Distribution of Industrial and Traffic Pollution Mixtures across Canada

Background: Air pollution has been associated with a variety of adverse health outcomes. Air pollution health effects may vary by pollutant mixture, with source acting as an important indicator of mixture composition. However, pollutant mixtures are often complex and difficult to assess. This study used source apportionment to identify pollutant sources and quantify their contributions to pollutant mixtures at regulatory monitoring sites. Land-use regression (LUR) modelling was then used to predict the spatial distribution of sector-based air pollution mixtures. Methods: Source apportionment was conducted using positive matrix factorization (PMF) on volatile organic compounds (VOCs) measured from 2006 to 2009 at 47 regulatory monitoring sites across Canada. Factor masses generated for each industrial or trafic source at each site were modelled as the dependant variable in LUR analyses. Model predictors included sector-specific facilities and land-use zoning within buffer areas ranging from 500 meters (m) to 20 kilometers (km) generated using ArcMap-10.0. For traffic models, road density and distance to major roads were also included. Models were selected based on adjusted r-squared, akaike’s information criterion, and variance inflation factor using SAS-9.2. Results: PMF models identified source impacts from five major industrial sectors as well as traffic. LUR models explained approximately 70% of the variability in sector-specific pollution for four of the industrial sectors. Major predictors included sector-specific facilities within 500m to 10km. Traffic models are still under development. An improvement is expected when wind speed and direction are incorporated into the models. Conclusions: Our results demonstrate that land-use information can be combined with source attribution obtained from PMF receptor modelling to predict pollutant concentrations associated with specific industrial sectors at un-monitored sites. These models will be applied in future health studies to examine whether air pollution health effects vary by industrial sector.

Multivariate Receptor Models for Spatially Correlated Multipollutant Data

The goal of multivariate receptor modeling is to estimate the profiles of major pollution sources and quantify their impacts based on ambient measurements of pollutants. Traditionally, multivariate receptor modeling has been applied to multiple air pollutant data measured at a single monitoring site or measurements of a single pollutant collected at multiple monitoring sites. Despite the growing availability of multipollutant data collected from multiple monitoring sites, there has not yet been any attempt to incorporate spatial dependence that may exist in such data into multivariate receptor modeling. We propose a spatial statistics extension of multivariate receptor models that enables us to incorporate spatial dependence into estimation of source composition profiles and contributions given the prespecified number of sources and the model identification conditions. The proposed method yields more precise estimates of source profiles by accounting for spatial dependence in the estimation. More importantly, it enables predictions of source contributions at unmonitored sites as well as when there are missing values at monitoring sites. The method is illustrated with simulated data and real multipollutant data collected from eight monitoring sites in Harris County, Texas. Supplementary materials for this article, including data and R code for implementing the methods, are available online on the journal web site.

Geometric calculation of view factors for stream surface radiation modelling in the presence of riparian forest

Many efforts to model stream temperature by using an energy budget approach have not accounted for view factors in modelling stream surface radiative exchanges, used informal approaches for computing them, or relied on calibration, which is not applicable for prediction at unmonitored sites or for predicting the effects of changes in riparian vegetation. In this paper, equations are derived for calculating view factors on the basis of geometric considerations for streams with and without riparian forest. The solutions can accommodate vegetation overhanging the stream surface. Example calculations illustrate the substantial variability of view factors across the stream width, which has implications for the estimation of view factors from point-scale radiation measurements over the stream surface, and the important influence of overhanging vegetation on view factors for narrow streams. View factors computed from the geometric model agreed well with view factors computed from hemispherical photography for streams ranging from 1 to almost 50 m wide, indicating that the model appears to be reasonably robust to deviations from the simplified geometry assumed by the model. In addition to their use in modelling stream surface energy exchanges, the solutions could also be adapted for application to energy balance and microclimate modelling in linear forest openings, such as seismic lines used in oil and gas exploration. Copyright © 2013 John Wiley & Sons, Ltd.

Proglacial river stage, discharge, and temperature datasets from the Akuliarusiarsuup Kuua River northern tributary, Southwest Greenland, 2008–2011

Abstract. Pressing scientific questions concerning the Greenland ice sheet's climatic sensitivity, hydrology, and contributions to current and future sea level rise require hydrological datasets to resolve. While direct observations of ice sheet meltwater losses can be obtained in terrestrial rivers draining the ice sheet and from lake levels, few such datasets exist. We present a new hydrologic dataset from previously unmonitored sites in the vicinity of Kangerlussuaq, Southwest Greenland. This dataset contains measurements of river stage and discharge for three sites along the Akuliarusiarsuup Kuua (Watson) River's northern tributary, with 30 min temporal resolution between June 2008 and July 2011. Additional data of water temperature, air pressure, and lake stage are also provided. Flow velocity and depth measurements were collected at sites with incised bedrock or structurally reinforced channels to maximize data quality. However, like most proglacial rivers, high turbulence and bedload transport introduce considerable uncertainty to the derived discharge estimates. Eleven propagating error sources were quantified, and reveal that largest uncertainties are associated with flow depth observations. Mean discharge uncertainties (approximately the 68% confidence interval) are two to four times larger (&amp;amp;pm;19% to &amp;amp;pm;43%) than previously published estimates for Greenland rivers. Despite these uncertainties, this dataset offers a rare collection of direct measurements of ice sheet runoff to the global ocean and is freely available for scientific use at http://dx.doi.org/10.1594/PANGAEA.762818.

Prediction of non-methane hydrocarbons in Kuwait using regression and Bayesian kriged Kalman model

This article describes the hierarchical Bayesian approach for predicting average hourly concentrations of ambient non-methane hydrocarbons (NMHC) in Kuwait where records of six monitor stations located in different sites are observed at successive time points. Our objective is to predict the concentration level of NMHC in unmonitored areas. Here an attempt is made for the prediction of unmeasured concentration of NMHC at two additional locations in Kuwait. We will implement a kriged Kalman filter (KKF) hierarchical Bayesian approach assuming a Gaussian random field, a technique that allows the pooling of data from different sites in order to predict the exposure of the NMHC in different regions of Kuwait. In order to increase the accuracy of the KKF we will use other statistical models such as imputation, regression, principal components, and time series analysis in our approach. We considered four different types of imputation techniques to address the missing data. At the primary level, the logarithmic field is modeled as a trend plus Gaussian stochastic residual model. The trend model depends on hourly meteorological predictors which are common to all sites. The residuals are then modeled using KKF, and the prediction equation is derived conditioned on adjoining hours. On this basis we developed a spatial predictive distribution for these residuals at unmonitored sites. By transforming the predicted residuals back to the original data scales, we can impute Kuwait’s hourly non-methane hydrocarbons field.

Comparing spatio‐temporal models for particulate matter in Piemonte

In the last two decades, increasing attention has been given to air pollution around the world, mainly because of its impact on human health and on the environment. In the Po valley (northern Italy), one of the most troublesome pollutant is PM10 (particulate matter with an aerodynamic diameter of less than 10 μm). In order to assess PM10 concentration over an entire region, environmental agencies need models to predict PM10 at unmonitored sites. To choose among possible predictive models and then meet the agencies' request, we focus on the class of Bayesian hierarchical models as they provide a flexible framework for incorporating relevant covariates as well as spatio‐temporal interactions. We consider six alternative models for PM10 concentration in Piemonte region (north‐western Po Valley), during the winter season October 2005–March 2006. Our aim is to choose a model that is satisfactory in terms of goodness of fit, interpretability, parsimony, prediction capability and computational costs. In order to support this choice, we propose a comparison approach based on a set of criteria summarized in a table that can be easily communicated to non‐statisticians. The comparison findings allow to provide Piemonte environmental agencies with an effective statistical model for building reliable PM10 concentration maps, equipped with the corresponding uncertainty measure. Copyright © 2011 John Wiley &amp; Sons, Ltd.

Near‐surface temperatures and heat balance of bare outcrops exposed to solar radiation

ABSTRACTSubsurface temperatures in rocks naturally fluctuate under the influence of local meteorological conditions. These fluctuations play a role in mechanical weathering, thus creating the environmental conditions conducive to natural hazards such as rockfalls and providing important sediment source terms for landscape evolution. However, the physics that control heat penetration into rocks are not fully understood, which makes the underground thermal state difficult to interpret when temperature measurements are available and even more difficult to estimate for unmonitored sites. This is an important lacuna given possible impacts of future climate change on mechanical weathering processes.The natural daily variations of subsurface temperatures were investigated on a bare gneiss outcrop exposed to solar radiation, where temperatures at various depths (up to 50 cm), as well as the solar radiation reaching a pyranometer, were monitored hourly for several months. This detailed times series of thermal data was used to gain insight into the heat balance at the inclined free surface of the rock mass.Attention was focused on two major contributors to the heat balance; the heat flux entering the rock mass through conduction and the incoming shortwave (solar) radiation. A Fourier decomposition of the temperature measurements provided an estimate of the in situ thermal conductivity of the rock and was used to calculate the conductive term. The shortwave radiation term was determined on the basis of the pyranometer measurements adjusted to account for the angle of incidence of the sun.It is shown that, throughout clear‐sky periods, heat exchanges at the surface are mainly controlled by direct solar radiation during the day, and by a roughly constant outgoing heat flux during the night. Subsurface temperatures can be reliably estimated with a semi‐infinite medium model whose boundary condition is derived from an analytical insolation model that takes atmospheric attenuation into account. Copyright © 2011 John Wiley &amp; Sons, Ltd.

Temporal and spatial trends in precipitation chemistry in the Georgia Basin, British Columbia

During the 1960s and 1970s, scientific evidence on the impacts of atmospheric acid deposition led to international negotiations to control emissions of compounds (primarily sulphur (S) and nitrogen (N) oxides) that undergo long-range transport. More recently, there has been growing interest in acid rain research in western Canada, e.g., the Georgia Basin, British Columbia, where S and N emissions are expected to increase during the next two decades. In the current study, long-term trends (1990–2007) in precipitation chemistry were evaluated at four stations (Saturna, British Columbia; Egbert, Ontario; Chapais Quebec; and Kejimkujik, Nova Scotia) using the Mann-Kendall statistical test. Precipitation chemistry at Saturna suggests that the Georgia Basin is less influenced by anthropogenic (S and N) emissions compared with monitoring stations in eastern Canada; nonetheless, long-term (1990–2007) trends in precipitation chemistry at Saturna [pH (significant increase) sulphate (significant decrease) and nitrate (decrease)] mirrored those at sites in eastern Canada suggesting all stations (across Canada) have responded similarly to large-scale emissions reductions under the Canada-United States Air Quality Agreement. The chemistry of precipitation and snow in the current study was well characterised by location and elevation, providing a way to estimate long-term mean annual base cation and chloride precipitation chemistry (ions showing no trend) at unmonitored sites for contiguous regions in the Georgia Basin.

Valuing coastal habitats: predicting high-tide roosts of non-breeding migratory shorebirds from landscape composition

Conservation of critical habitats for shorebirds (Charadriiformes: Charadrii), usually requires spatially explicit identification of each habitat. Most non-breeding shorebirds depend on two critical habitats: foraging (intertidal flats) and roosting (safe and open supratidal sites). We used published information on use of high-tide roosts by shorebirds to develop models of roost selection (probability of occurrence) and usage (mean abundance) for 12 species of shorebirds spending the non-breeding season in Moreton Bay, in subtropical eastern Australia. Selection of roosts was most strongly affected by the proximity to a large foraging area, field of view and to a lesser extent by composition of the surrounding landscape. Abundance of the most common species, such as Bar-tailed Godwit (Limosa lapponica), Eastern Curlew (Numenius madagascariensis) and Whimbrel (N. phaeopus), depended on distance to the nearest large foraging area or the size of the roost per se. Our results suggest that the shorebird species occurring in a subtropical Australian estuary use two generalised types of roosts: exposed ocean-front roosts typified by supratidal spits and sandbars and patches of claypan—saltmarsh in the upper reaches of the tidal range surrounded by mangroves. The amount of these habitats available in Moreton Bay was examined by visiting unmonitored sites with a predicted high probability of occupancy by two key species, Eastern Curlew and Bar-tailed Godwit. Few of the unmonitored sites currently supported many shorebirds. These data suggest that despite the high demand on coastal land for urban development in Moreton Bay, suitable roosting habitat is still available near their foraging areas.

Valley floor climate observations from the McMurdo dry valleys, Antarctica, 1986–2000

Climate observations from the McMurdo dry valleys, East Antarctica are presented from a network of seven valley floor automatic meteorological stations during the period 1986 to 2000. Mean annual temperatures ranged from −14.8°C to −30.0°C, depending on the site and period of measurement. Mean annual relative humidity is generally highest near the coast. Mean annual wind speed increases with proximity to the polar plateau. Site‐to‐site variation in mean annual solar flux and PAR is due to exposure of each station and changes over time are likely related to changes in cloudiness. During the nonsummer months, strong katabatic winds are frequent at some sites and infrequent at others, creating large variation in mean annual temperature owing to the warming effect of the winds. Katabatic wind exposure appears to be controlled to a large degree by the presence of colder air in the region that collects at low points and keeps the warm less dense katabatic flow from the ground. The strong influence of katabatic winds makes prediction of relative mean annual temperature based on geographical position (elevation and distance from the coast) alone, not possible. During the summer months, onshore winds dominate and warm as they progress through the valleys creating a strong linear relationship (r2 = 0.992) of increasing potential temperature with distance from the coast (0.09°C km−1). In contrast to mean annual temperature, summer temperature lends itself quite well to model predictions, and is used to construct a statistical model for predicting summer dry valley temperatures at unmonitored sites.

Contending with space–time interaction in the spatial prediction of pollution: Vancouver's hourly ambient PM10 field

AbstractIn this article we describe an approach for predicting average hourly concentrations of ambient PM10 in Vancouver. We know our solution also applies to hourly ozone fields and believe it may be quite generally applicable. We use a hierarchical Bayesian approach. At the primary level we model the logarithmic field as a trend model plus Gaussian stochastic residual. That trend model depends on hourly meteorological predictors and is common to all sites. The stochastic component consists of a 24‐hour vector response that we model as a multivariate AR(3) temporal process with common spatial parameters. Removing the trend and AR structure leaves ‘whitened’ time series of vector series. With this approach (as opposed to using 24 separate univariate time series models), there is little loss of spatial correlation in these residuals compared with that in just the detrended residuals (prior to removing the AR component). Moreover our multivariate approach enables predictions for any given hour to ‘borrow strength’ through its correlation with adjoining hours. On this basis we develop a spatial predictive distribution for these residuals at unmonitored sites. By transforming the predicted residuals back to the original data scales we can impute Vancouver's hourly PM10 field. Copyright © 2002 John Wiley &amp; Sons, Ltd.

Designing and Integrating Composite Networks for Monitoring Multivariate Gaussian Pollution Fields

SUMMARY Networks of ambient monitoring stations are used to monitor environmental pollution fields such as those for acid rain and air pollution. Such stations provide regular measurements of pollutant concentrations. The networks are established for a variety of purposes at various times so often several stations measuring different subsets of pollutant concentrations can be found in compact geographical regions. The problem of statistically combining these disparate information sources into a single ‘network’ then arises. Capitalizing on the efficiencies so achieved can then lead to the secondary problem of extending this network. The subject of this paper is a set of 31 air pollution monitoring stations in southern Ontario. Each of these regularly measures a particular subset of ionic sulphate, sulphite, nitrite and ozone. However, this subset varies from station to station. For example only two stations measure all four. Some measure just one. We describe a Bayesian framework for integrating the measurements of these stations to yield a spatial predictive distribution for unmonitored sites and unmeasured concentrations at existing stations. Furthermore we show how this network can be extended by using an entropy maximization criterion. The methods assume that the multivariate response field being measured has a joint Gaussian distribution conditional on its mean and covariance function. A conjugate prior is used for these parameters, some of its hyperparameters being fitted empirically.

Interpolating Vancouver's daily ambient PM10 field

In this article we develop a spatial predictive distribution for the ambient space – time response field of daily ambient PM10 in Vancouver, Canada. Observed responses have a consistent temporal pattern from one monitoring site to the next. We exploit this feature of the field by adopting a response model with two components, a common deterministic trend across all sites plus a stochastic residual. We are thereby able to whiten the temporal residuals without losing much of the spatial correlation in the original log-transformed series. This in turn enables us to develop an effective spatial predictive distribution for these residuals at unmonitored sites. By transforming the predicted residuals back to the original data scales, we can impute Vancouver's daily PM10 field for purposes such as human exposure and health impacts analysis. Copyright © 2000 John Wiley & Sons, Ltd.

Will continued monitoring of beaver-damaged resources minimize future damage?

Author(s): Wilson, Ben S.; McEwen, Gary M. | Abstract: The purpose of this study was to determine if continued monitoring and removal of beavers (Castor canadensis) from previously controlled beaver damage sites resulted in less additional damage than not monitoring such sites. Beavers were removed from 34 sites in nine southeast Texas counties from August 1996 through March 1997. Sixteen sites subsequently were monitored monthly and, if beavers had reinvaded, they were removed and the additional damage value was recorded. The remaining 18 sites were not monitored monthly, but they were visited for a final survey at the end of the study. The value of additional damage was recorded at that time. Damage following reinvasion occurred more often when sites were not monitored (5 of 7 sites, compared to only 2 of 7 reinvaded, monitored sites). In addition, when damage occurred at reinvaded sites, monetary value appeared to be greater without monitoring (average $940, n=5) than with monitoring (average $125, n=2). The larger average damage values for reinvaded unmonitored sites compared to reinvaded monitored sites would be important to landowners when deciding if property should be monitored. Factors that made some sites susceptible to reinvasion were also evaluated. Significantly more beavers were taken initially, per site, in the reinvaded sites compared to all other sites. This implies that better habitat and higher beaver density were the most important factors in determining a site’s susceptibility to reinvasion.

IMPROVING REGIONAL‐MODEL ESTIMATES OF URBAN‐RUNOFF QUALITY USING LOCAL DATA1

ABSTRACT: Urban water‐quality managers need load estimates of storm‐runoff pollutants to design effective remedial programs. Estimates are commonly made using published models calibrated to large regions of the country. This paper presents statistical methods, termed model‐adjustment procedures (MAPs), which use a combination of local data and published regional models to improve estimates of urban‐runoff quality. Each MAP is a form of regression analysis that uses a local data base as a calibration data set to adjust the regional model, in effect increasing the size of the local data base without additional, expensive data collection. The adjusted regional model can then be used to estimate storm‐runoff quality at unmonitored sites and storms in the locality. The four MAPs presented in this study are (1) single‐factor regression against the regional model prediction, Pu; (2) least‐squares regression against Pu; (3) least‐squares regression against Pu and additional local variables; and (4) weighted combination of Pu and a local‐regression prediction. Identification of the statistically most valid method among these four depends upon characteristics of the local data base. A MAP‐selection scheme based on statistical analysis of the calibration data set is presented and tested.