Conceived and designed the experiments: SMF NC. Performed the experiments: SMF. Analyzed the data: SMF. Contributed reagents/materials/analysis tools: SMF BJM. Wrote the paper: SMF NC MPW JAT BJM NKD.
The authors have declared that no competing interests exist.
The influences of relative humidity and ambient temperature on the transmission of influenza A viruses have recently been established under controlled laboratory conditions. The interplay of meteorological factors during an actual influenza epidemic is less clear, and research into the contribution of wind to epidemic spread is scarce. By applying geostatistics and survival analysis to data from a large outbreak of equine influenza (A/H3N8), we quantified the association between hazard of infection and air temperature, relative humidity, rainfall, and wind velocity, whilst controlling for premiseslevel covariates. The pattern of disease spread in space and time was described using extraction mapping and instantaneous hazard curves. Meteorological conditions at each premises location were estimated by kriging daily meteorological data and analysed as timelagged timevarying predictors using generalised Cox regression. Meteorological covariates timelagged by three days were strongly associated with hazard of influenza infection, corresponding closely with the incubation period of equine influenza. Hazard of equine influenza infection was higher when relative humidity was <60% and lowest on days when daily maximum air temperature was 20–25°C. Wind speeds >30 km hour^{−1} from the direction of nearby infected premises were associated with increased hazard of infection. Through combining detailed influenza outbreak and meteorological data, we provide empirical evidence for the underlying environmental mechanisms that influenced the local spread of an outbreak of influenza A. Our analysis supports, and extends, the findings of studies into influenza A transmission conducted under laboratory conditions. The relationships described are of direct importance for managing disease risk during influenza outbreaks in horses, and more generally, advance our understanding of the transmission of influenza A viruses under field conditions.
Influenza A viruses are enveloped RNA viruses of the family
In their animal model of human influenza A transmission, Lowen et.al. have shown that dry cool conditions (low relative humidity and cold ambient temperatures) increase the spread of influenza
Equine influenza virus (A/H3N8) is a highly contagious cause of low mortality, high morbidity respiratory disease capable of infecting all members of the horse family (
Contacttracing early in the 2007 outbreak revealed that the disease initially spread through a network of equestrian events, linked by the movement of infected horses prior to detection of the outbreak, producing clusters of infected premises in widespread locations
In this paper we present a comprehensive analysis of the influence of meteorological variables on time to infection based on an influenza A virus outbreak dataset. This spatiotemporal analysis aims to identify and quantify the association between four meteorological variables (air temperature, relative humidity, rainfall, wind velocity) and time to infection in the largest cluster of the 2007 equine influenza (A/H3N8) outbreak in Australia. We are unaware of any previously published analysis that combines such a large and spatiotemporally detailed influenza outbreak dataset with concurrent daily meteorological data, to allow meaningful estimation of the contribution of such factors in the spread of an influenza A outbreak.
The state government of New South Wales provided contacttracing and laboratory testing data on all horses investigated during the 2007 outbreak. This dataset was collected at the level of individual horses and aggregated to the premises level for analysis. Study designs that use groups as the unit of interest (such as herds or flocks) rather than individuals, are common in veterinary epidemiological research
There was a single ‘index’ for the 2007 outbreak of equine influenza in Australia: an equestrian event located 160 km north of Sydney, at which transmission was known to have occurred. This analysis focused on local spread within the single largest cluster of the outbreak, centred 60 km northwest of Sydney's city centre (
(a) From August–December 2007, around 70,000 horses were infected on over 9000 horse premises in two Australian States. (b) This study focused on the largest cluster (n = 3624 horse premises), northwest of Sydney, as defined by a 15 km buffer around the nine earliest infected premises (depicted in yellow) that were contacttraced to events where disease transmission was known to have occurred in the first week of the outbreak. Clinical signs were first observed on 17 August 2007 in a horse in quarantine at Eastern Creek Quarantine Station (red closed circle). The cluster is surrounded by national parks and Sydney urban areas. (For interpretation of the references to colour in this text, the reader is referred to the web version of the article.).
The dataset was imported into the R statistical package version 2.13.0
We applied semiparametric Cox regression modelling to estimate the association between potential risk factors and the times to infection of individual premises. A geodatabase was compiled in Microsoft Access 2007 (Microsoft Corporation, Redmond, WA, USA) to maintain all premises and meteorological data, with spatial covariates added using ArcMap 9.3 (ESRI, Redlands, CA, USA). The dataset was structured into a daily ‘counting process’ formulation to enable investigation of the effects of timevarying predictors
In the counting process generalisation of the Cox proportional hazards model, the hazard function depends on time in ways other than only through the baseline hazard function
Network and spatial spread in the early outbreak period (the first 14 days of this outbreak) is described in detail elsewhere
Explanatory covariates tested for associations with the time to infection of premises in the Northwest Sydney cluster are listed in
Variable group  Variable name  Variables (Units) 

RAIN 
Rainfall (mm day^{−1}) 

RH_9AM 
Relative humidity (%) measured daily at 9am 

RH_3PM 
Relative humidity (%) measured daily at 3pm 
TEMP_MAX 
Maximum daily air temperature (°C) 

TEMP_MIN 
Minimum daily air temperature (°C) 

WIND_SPD_{undir} 
Maximum daily wind speed – undirected (km hour^{−1}) 

WIND_SPD_{dir(k)} 
Maximum daily wind speed – directed (km hour^{−1}) 


AREA  Area (acres) 

HORSE_DENSITY  Horse density (horses acre^{−1}) 
HORSES_NUMBER  Number of horses  
SHARED_FENCE  Length of shared fence with other horse premises (m)  
VACC  Vaccination status (1 = Yes, 0 = No) 

VACC_DAYS  Days since vaccination 


ELEV  Elevation (m) 

HUMAN_DENS  Human population density within approximately 1 km of the premises (people km^{−2}) 
ROAD_DIST  Distance to nearest main road (km) 
Timechanging covariate.
Maximum daily wind speed was either based on wind from all directions (‘undirected’) or wind only from within 45° arcs centred on the direction of the
Main roads include freeways, highways, primary and arterial roads (Classes 1–3).
Hourly wind velocity data (wind direction and speed) and daily data for five other meteorological variables (rainfall, minimum and maximum daily air temperature, and relative humidity measured at 9 am and 3 pm) were obtained from 132 weather stations. All of these weather stations were operated by the Australian Bureau of Meteorology during the study period, and were located either within the cluster or within 20 km of the cluster boundary. Most stations reported only daily rainfall measurements. Ordinary kriging
Kriging is a geostatistical smoothing technique that involves modelling the underlying spatial dependency (autocorrelation) in spatially continuous data based on a covariance function (
(a) Exponential covariance function (with practical range = 0.25) and its related semivariance function. (b) Hourly wind velocity data from sixteen automated weather stations (open circles) within a 20 km buffer of the cluster's boundary were converted into their EasttoWest (‘u’) and NorthtoSouth (‘v’) components, and smoothed using kriging to predict hourly wind speed and direction at each premises (small grey dots). (c) For each premises on each day prior to infection or censoring, the (‘directed’) maximum wind speed originating from within 45( arcs centred on the direction of the nearest 1–3 infected premises was estimated for time lags of 1–5 days.
Hourly wind velocity data were available from sixteen of the weather stations, automatically measured on masts at 10(metres above the earth's surface. These wind data were supplied in a polar coordinate structure, comprising the average direction of origin of the wind (in degrees from true north) and the maximum wind speed (in kilometres hour(1), measured over the 10(minutes leading up to the observation time. To avoid the issue of northerly bearings being split at true north (i.e. true bearings of 1( and 359( seeming distant when they are only 2( apart), prior to variography and kriging, the wind velocity data was converted into a Cartesian coordinate system—defined by two components (
Kriging was then conducted on the two wind velocity vector components
Two approaches were taken to aggregate the hourly wind velocity vectors for each premises into daily maximum wind speed covariates. First, to test the hypothesis that increased wind speed from any direction was associated with increased hazard of infection we generated ‘undirected’ maximum daily wind speed covariates (‘WIND_SPD_{undir}’) without making any directional assumptions, taking the maximum of all hourly wind speed estimates for each premises on each day.
Next, to explore the directionality of wind exposure risk we generated ‘directed’ maximum daily wind speed covariates (‘WIND_SPD_{dir}’) based only on wind coming from within the direction of the nearest
Finally, these timevarying predictors were lagged by 1–5 days to serve as proxies for wind within the range of incubation periods that have been observed for equine influenza, producing 20 timelagged explanatory covariates: ‘WIND_SPD_{undir}
Instantaneous hazard curves were constructed for each timeinvariant covariate with the ‘epiR’ library in R
Univariable Cox models were then constructed and the statistical strength of the association between each variable (categorical or continuous) and the outcome assessed using likelihood ratio tests
All remaining variables (unconditionally statistically associated with the log hazard of infection at
Goodness of fit of the final model was assessed using ‘Martingale’ residuals. The influence of every individual observation was tested by omitting it and observing for change in the regression coefficients
The Northwest Sydney cluster of the 2007 equine influenza outbreak in Australia contained 3624 horse premises, of which 1922 were reported to be infected during the 131 day outbreak (cumulative incidence = 53.0%, 95% CI: 51.4, 54.7%).
Surfaces of spatial relative risk by four week period are included as
Surfaces of log relative risk were estimated in 4week intervals using adaptive kernel estimation, with upper 95% tolerance contours (solid white lines). With this method the amount of smoothing (bandwidth) is inversely proportional to the density of the population at risk.
The mean centre of the outbreak did not move predominantly in any single direction over the study period, moving Northwest at 3.0 km week^{−1} in the first 4 weeks, then Southwest at 3.9 km week^{−1} for 4 weeks, before moving back to the East at 4.1 km week^{−1} whilst the epidemic faded out.
The complete survival dataset included 3153 premises containing 1727 events (infections) during the study period. Data on 57 infected horse premises were excluded because their onset dates occurred in the first 14 days of the outbreak (a period when they could possibly have been infected by the movement of infected horses rather than by local spatial spread). Sixtyseven infected premises were missing a date of onset, and 347 premises (71 infected and 276 uninfected premises) were missing data on their number of horses. Once data on these premises (which were evenly distributed across the study extent) had been excluded, data on all variables were complete. The median survival time, the point at which half of the premises in this cluster were infected, was day 55 of this outbreak (95% CI : 52, 61). The instantaneous hazard, the proportion of infections per day in the population surviving uninfected until that day, peaked on day 28 (
Horse movement standstills were implemented from day 10, and vaccination commenced in this cluster on day 49. Dashed lines represent 95% confidence intervals, and dotted vertical reference lines denote the survival analysis study period (between days 14 and 131 of the outbreak).
Most horse premises were relatively close to a weather station, with the mean distance to the nearest weather station reporting wind data being 11.7 km (SD = 5.4 km, maximum = 27.4 km). For all meteorological data, there was a paucity of weather stations in the Northwest corner of the study extent (because this region is bordered by a national park).
Daily rainfall data were available from 127 weather stations in the study extent.
Over the study period, the median estimated daily rainfall per premises was 0.1 mm day^{−1} (IQR: 0 to 2.8 mm day^{−1}, maximum = 106.5 mm day^{−1}). No statistically significant associations were detected between timelagged rainfall covariates and hazard of infection (
Meteorological Factor  Timelag 

SE( 
LRT 


Rainfall (mm day^{−1}) 

0.006  0.033  0.0  1  0.852 

−0.005  0.028  0.0  1  0.870  

−0.045  0.037  1.5  1  0.215  

−0.024  0.027  0.9  1  0.342  

−0.028  0.031  0.9  1  0.344  
Relative humidity (%), 


29.9  4  <0.001  
measured daily at 9 am 


29.4  4  <0.001  


14.3  4  0.006  


47.1  4  <0.001  


41.9  4  <0.001  
Relative humidity (%), 


47.7  4  <0.001  
measured daily at 3 pm 


39.0  4  <0.001  


35.1  4  <0.001  


71.6  4  <0.001  


81.4  4  <0.001  
Maximum daily air 


58.4  4  <0.001  
temperature (°C) 


44.2  4  <0.001  


64.1  4  <0.001  


49.5  4  <0.001  


47.2  4  <0.001  
Minimum daily air 

−0.040  0.031  1.6  1  0.204 
temperature (°C) 

−0.033  0.032  1.1  1  0.289 

−0.068  0.032  4.6  1  0.031  

−0.052  0.033  2.5  1  0.111  

−0.027  0.032  0.7  1  0.395  
Maximum daily wind 


10.2  4  0.038  
speed (km hour^{−1}) 


19.6  4  <0.001  


52.0  4  <0.001  



35.5  4  <0.001  


14.9  4  0.005 
Maximum daily wind speed based on wind from all directions (‘undirected’), making no assumption concerning nearest infected premises assumption.
Relative humidity data measured twice daily (at 9 am and 3 pm) were available from eighteen weather stations (
Daily meteorological data provided by Australian Bureau of Meteorology weather stations (white closed circles) were smoothed using kriging, and timelagged by 1–5 days. (a) Smoothed estimate of relative humidity measured at 3 pm on Day 20 of the outbreak. Small grey dots denote the horse premises. (b) Restricted cubic splines of the crude relationship between hazard of infection and relative humidity (3 pm measurement) at timelags of 1–5 days over the entire study period. (c) Smoothed daily maximum air temperature on Day 20 and (d) the relationship between daily maximum air temperature and hazard of infection, by time lag.
A negative cubic relationship was observed between relative humidity and hazard of infection (
Daily surface air temperature data were available from 21 weather stations (
A highly nonlinear relationship was observed between infection and maximum daily air temperature (
Hourly wind velocity data were available from sixteen weather stations (
The univariate relationship between hazard of infection and wind speed, making no directional assumptions (’undirected’), is presented in
Estimates are based on hourly wind data from all directions and timelagged by 1–4 days. Dashed lines represent 95% confidence intervals.
Estimates are based only on hourly wind data from within 45° arcs centred on the direction of the
Meteorological Factor  Timelag  Term  LRT 


Maximum daily wind 


3.8  4  0.430 
speed (km hour^{−1}) 


9.1  4  0.058 


16.5  4  0.002  


6.6  4  0.159  


3.4  4  0.499  
Maximum daily wind 


14.0  4  0.007 
speed (km hour^{−1}) 


25.3  4  <0.001 


34.5  4  <0.001  


8.2  4  0.083  


24.6  4  <0.001  
Maximum daily wind 


41.2  4  <0.001 
speed (km hour^{−1}) 


49.5  4  <0.001 


75.6  4  <0.001  


38.0  4  <0.001  


52.3  4  <0.001 
Maximum daily wind speed (‘directed’) based on wind only from within 45° arcs centred on the direction of the
The following five candidate meteorological variables were consequently selected for multivariable analysis: linear terms for rainfall and minimum daily air temperature, both timelagged by 3 days, a restricted cubic spline for relative humidity measured at 3 pm timelagged by 5 days, and splines of maximum daily air temperature and maximum daily wind speed from the direction of the three nearest infected premises, both timelagged by 3 days.
Horse premises in the Northwest Sydney cluster were highly skewed in terms of their land area and the number of horses they held at the time of the outbreak. The median premises held 2 horses (IQR: 1, 5 horses; maximum: 139 horses) on 5.1 acres (IQR: 4.8, 15.2 acres; maximum: 2 225 acres). These variables were log transformed for all further analyses, with results backtransformed for presentation. Highly nonlinear crude relationships were observed between hazard of infection and premises area and horse density (
(a) The relationship between hazard of infection and premises area, and (b) the relationship between hazard of infection and local human population density (people residing within approximately 1 km of the horse premises). Dashed lines represent 95% confidence intervals.
Factor  Category  No.  Hazard ratio  (95% CI)  


Area (acres)  >15.2  788  0.99  (0.85, 1.15)  <0.001 
5.1–15.2  788  1.94  (1.69, 2.23)  
4.8–5.1  789  2.09  (1.83, 2.40)  
<4.8  788  1.00  
Horse density  >1.00  776  1.50  (1.29, 1.74)  <0.001 
(horses acre^{−1})  0.40–1.00  799  2.51  (2.18, 2.89)  
0.20–0.40  787  1.85  (1.63, 2.17)  
<0.20  791  1.00  
Number of horses  >5  662  3.28  (2.82, 3.82)  <0.001 
3–5  902  2.48  (2.14, 2.88)  
2  787  2.08  (1.79, 2.43)  
1  802  1.00  
Length of shared fence  >300  742  1.45  (1.29, 1.63)  <0.001 
with other horse  1–300  725  1.64  (1.47, 1.84)  
premises (m)  0  1686  1.00  
Vaccination status 
Yes  490  0.28  (0.04, 2.13)  0.137 
No  2663  1.00  


Elevation (m)  >115  785  0.72  (0.63, 0.82)  <0.001 
45–115  777  0.66  (0.58, 0.76)  
25–45  786  1.02  (0.90, 1.15)  
<25  805  1.00  
Human population  >500  1059  1.05  (0.94, 1.18)  <0.001 
density (people km^{−2})  1–500  954  1.29  (1.48, 1.44)  
0  1140  1.00  
Distance to nearest  >2.2  787  1.23  (1.08, 1.41)  0.021 
main road (km)  1.1–2.2  789  1.14  (1.00, 1.31)  
0.4–1.0  788  1.11  (0.97, 1.27)  
<0.4  786  1.00 
Timechanging covariate.
A trend existed across the study area in terms of premises elevation and surrounding human population density. Hazard of infection was higher on horses premises located at lower elevations (<45 m) and >2.2 km from main roads (
Premises area and premises horse density were the only highly correlated pairing (
The final model is presented in
Factor  Category  Hazard ratio  (95% CI)  


Rainfall (mm day^{−1}), 

0.91  (0.82, 1.00)  0.055 
Relative humidity (%), 

—  —  <0.001 
measured daily at 3pm, 

Maximum daily air 

—  —  <0.001 
temperature (°C), 

Maximum daily wind speed, 

—  —  <0.001 
(km hour^{−1}), 



Area (acres) 

—  —  <0.001 
Number of horses  >5  3.16  (2.70, 3.69)  <0.001 
3–5  2.19  (1.89, 2.55)  
2  1.93  (1.66, 2.26)  
1  1.00  
Length of shared fence  >300  1.30  (1.15, 1.48)  <0.001 
with other horse premises (m)  1–300  1.27  (1.13, 1.43)  
0  1.00  
Vaccination status 
Yes  0.28  (0.04, 2.09)  0.134 
No  1.00  


log_{10}(Elevation (m)) 

0.58  (0.51, 0.67)  <0.001 
Human population density 

—  —  <0.001 
(people km^{−2}) 
Timechanging covariate, timelagged 3 or 5 days as noted.
Maximum daily wind speed (‘directed’) based on wind only from within 45° arcs centred on the direction of the three nearest infected premises assuming that premises were infectious for 14 days and one of the three nearest infective premises was the source of infection.
The shape of the restricted cubic splines representing the nonlinear relationships between hazard of infection and relative humidity, maximum daily air temperature, maximum daily wind speed (from the direction of the nearest three infected premises), premises area and human population density, were all largely unchanged from their crude forms (as presented in
The final model accounted for a quarter of the variability in the data (Schemper and Stare pseudoR^{2} = 25.8%). No issues were identified based on inspection of martingale and deviance residuals, both overall, and when plotted against each variable included in the final model. Residual spatial structure was not evident in the empirical semivariogram of the deviance residuals, suggesting that spatial correlation was not unduly influencing our effect estimates (or their associated standard errors). Influence statistics identified only one important outlying premises, infected 36 days after the vaccination of the 2 horses on the property. These horses did not receive a second vaccination, whilst up to three doses may be required to attain protective immunity.
To our knowledge, this empirical analysis provides the first estimates of the contribution of humidity, air temperature and wind to the spread of an actual outbreak of influenza (‘in the field’). We have demonstrated that it is possible to detect an association between wind velocity and disease spread, and directly estimate the strength of such an association. This advances our understanding of the windborne spread of influenza from purely circumstantial association to a hypothesis statisticallytested with empirical data.
Our analysis shows that influenza spread in this cluster was highly dependent on relative humidity. Recent reviews
Recent research has suggested that in certain situations absolute humidity may better represent the relationship between air humidity and influenza A virus survival
The shape of the highly nonlinear relationship that we observed between hazard of equine influenza infection and maximum daily air temperature suggests two mechanisms of influenza transmission. Hazard was lowest on days when the maximum air temperature was between 20–25°C, and greatly increased on days with lower and higher maximum temperatures. Aerosol transmission of influenza A viruses has been shown to be enhanced in cooler conditions
There is some consensus in the literature that airborne transmission of influenza is at least possible; however, there is strong disagreement about its importance
In developing proxy covariates for the directed formulation of the daily wind speed covariates (‘WIND_SPD_{dir(k)}’) certain assumptions were required. Wind data was only included if it was within a 45° arc of the nearest
The associations that we have detected between increasing wind speed and hazard of infection need to be interpreted in the context of our study design. There is potential for ecological fallacy in aggregated data analyses such as this, in which the unit of interest is not an individual animal but a group. Furthermore, it is not possible in such observational epidemiological analyses to definitively identify windborne spread from any other transmission route (direct contact, cough droplet and spread on fomites). Nonetheless, the detected association, presumably representing windborne spread of equine influenza, is biologically plausible, and its increasing strength with increasing wind speed from the direction of nearby infected premises is difficult to explain by spread through other means alone.
At wind speeds of >30 km hour^{−1} an aerosol of influenza droplet nuclei would only need to be stable for minutes to be able to infect horses on nearby premises. Equine influenza viruses have been shown to survive for periods of hours to days in soil and water, even in direct sunlight
A recent timeseries analysis investigated correlation between the frequency of paediatric influenza A hospital admissions and several meteorological variables including wind velocity
When interpolating meteorological covariates and estimating nearest neighbour distances we used centroids to reduce the complexity of the analytical methods. For >99% of the premises in our dataset we estimate that the maximum distance between the centroid and premises boundary was <500 m. When the largest 1% of premises (in area) were excluded from the final model, the only regression coefficients to change by >20% were the two highest order spline components for relative humidity and maximum daily temperature, and these changes were not discernible in postadjustment plots. We therefore consider our findings to be insensitive to measurement bias introduced by representing premises by their centroids.
Environmental variables capable of influencing airborne disease spread (such as local horse density, tree density or terrain undulation) vary considerably in the different regions and clusters of premises infected during the 2007 equine influenza outbreak in Australia. A potential limitation of this analysis was that we focussed on only one cluster (the largest and most dense cluster in terms of population at risk) from a very large outbreak. There were two considered reasons for our detailed focus: counting process survival analysis involves analysing a very large dataset (204,909 observations on 3153 premises); and owing to a wide variance in local environmental characteristics and potential for differences in disease transmission dynamics, mixing clusters in the same analysis might dilute any meaningful results. Before generalising our findings to the whole outbreak, or indeed other outbreaks, followup research to assess the importance of the risk factors investigated in broadly dissimilar environments, is therefore required.
A classical geostatistical approach
In the cluster of infection investigated, disease did not appear to spread predominantly in any single direction. We purposefully focussed on this cluster rather than other large clusters in which a single global direction of spread has been noted
By restricting this analysis to a study period after the horse movement ban was put in place, we focussed this study on factors influencing the local spread of equine influenza. We also adjusted for a number of relevant confounders of the meteorological associations we aimed to estimate: vaccination status of horses on the premises, premises size (in terms of area and number of horses), whether premises were adjacent to another premises holding horses, and local human population density. A small misclassification bias is known to be present in the equine influenza dataset, due to underreporting of infected premises by owners either attempting to avoid movement restrictions or who failed to detect infection
In conclusion, by combining influenza outbreak and concurrent meteorological data, we have shown how relative humidity, air temperature and wind velocity combined to influence the spread of an actual influenza outbreak. Hazard of equine influenza infection was higher when relative humidity was <60% and lowest on days when daily maximum air temperature was 20–25°C. Wind speeds >30 km hour^{−1} from the direction of nearby infected premises were associated with increased hazard of infection. Our analysis supports, and extends, the findings of studies into influenza A transmission conducted under controlled conditions. The relationships described are of direct importance for managing disease risk during influenza outbreaks in horses, and more generally, advance our understanding of the transmission of influenza A viruses under natural conditions.
Survival analysis dataset formulation examples and correlations between explanatory variables in Cox regression modelling of factors associated with time to infection in the largest cluster of the 2007 outbreak of equine influenza in Australia.
(DOCX)
The authors gratefully acknowledge NSW DPI for making their equine influenza dataset available, Brendan Cowled (DAFF) for contributing to data compilation, Mark Stevenson (The EpiCentre, Massey University) for advice and initial R code for survival and spatial analyses, Graeme Garner (DAFF) and Evan Sergeant (AusVet) for advice on the manuscript. We also thank the Australian Bureau of Meteorology's National Climate Data Service and the Land and Property Information Division of the NSW Department Finance and Services for data provision.