importNGWMN_wml2.R 6.1 KB
Newer Older
David Watkins's avatar
David Watkins committed
1
#' Function to return data from the National Ground Water Monitoring Network waterML2 format
2
3
4
#'
#' This function accepts a url parameter for a WaterML2 getObservation. This function is still under development,
#' but the general functionality is correct.
David Watkins's avatar
David Watkins committed
5
#' 
6
7
8
#' @param input character or raw, containing the url for the retrieval or a path to the data file, or raw XML.
#' @param asDateTime logical, if \code{TRUE} returns date and time as POSIXct, if \code{FALSE}, character
#' @param tz character to set timezone attribute of datetime. Default is an empty quote, which converts the 
David Watkins's avatar
David Watkins committed
9
#' datetimes to UTC (properly accounting for daylight savings times based on the data's provided time zone offset).
10
11
#' Possible values to provide are "America/New_York","America/Chicago", "America/Denver","America/Los_Angeles",
#' "America/Anchorage","America/Honolulu","America/Jamaica","America/Managua","America/Phoenix", and "America/Metlakatla"
David Watkins's avatar
David Watkins committed
12
#' @return mergedDF a data frame source, time, value, uom, uomTitle, comment, gmlID
13
14
15
16
17
#' @export
#' @importFrom xml2 read_xml
#' @importFrom xml2 xml_find_all
#' @importFrom xml2 xml_text
#' @importFrom xml2 xml_attr
18
#' @importFrom xml2 xml_find_first
19
20
#' @importFrom lubridate parse_date_time
#' @examples
David Watkins's avatar
David Watkins committed
21
#' \dontrun{
22
#' url <- "http://cida.usgs.gov/ngwmn_cache/sos?request=GetObservation&service=SOS&version=2.0.0
23
24
#' &observedProperty=urn:ogc:def:property:OGC:GroundWaterLevel&responseFormat=text/xml&featureOf
#' Interest=VW_GWDP_GEOSERVER.USGS.403836085374401"
David Watkins's avatar
David Watkins committed
25
#' data <- importNGWMN_wml2(url)
David Watkins's avatar
David Watkins committed
26
#' 
27
#' url <- "http://cida.usgs.gov/ngwmn_cache/sos?request=GetObservation&service=SOS&version=2.0.0
28
29
#' &observedProperty=urn:ogc:def:property:OGC:GroundWaterLevel&responseFormat=text/xml&featureOf
#' Interest=VW_GWDP_GEOSERVER.USGS.474011117072901"
David Watkins's avatar
David Watkins committed
30
#' data <- importNGWMN_wml2(url)
David Watkins's avatar
David Watkins committed
31
#' }
David Watkins's avatar
David Watkins committed
32
#' 
33
34
#' 
#TODO: separate id and agency name, give also as separate dimensions
David Watkins's avatar
David Watkins committed
35
importNGWMN_wml2 <- function(input, asDateTime=FALSE, tz=""){
36
  if(tz != ""){
David Watkins's avatar
David Watkins committed
37
    tz <- match.arg(tz, OlsonNames())
38
39
40
41
42
43
44
45
46
47
48
49
  }else{tz = "UTC"}
  
  raw <- FALSE
  if(class(input) == "character" && file.exists(input)){
    returnedDoc <- read_xml(input)
  }else if(class(input) == 'raw'){
    returnedDoc <- read_xml(input)
    raw <- TRUE
  } else {
    returnedDoc <- xml_root(getWebServiceData(input, encoding='gzip'))
  }
  
David Watkins's avatar
David Watkins committed
50
51
52
53
54
55
56
57
58
59
60
  response <- xml_name(returnedDoc)
  if(response == "GetObservationResponse"){
    
    timeSeries <- xml_find_all(returnedDoc, "//wml2:MeasurementTimeseries") #each parameter/site combo
    
    if(0 == length(timeSeries)){
      df <- data.frame()
      if(!raw){
        attr(df, "url") <- input
      }
      return(df)
61
62
    }
    
David Watkins's avatar
David Watkins committed
63
    mergedDF <- NULL
64
    
David Watkins's avatar
David Watkins committed
65
66
67
    for(t in timeSeries){
      gmlID <- xml_attr(t,"id")
      TVP <- xml_find_all(t, ".//wml2:MeasurementTVP")#time-value pairs
68
69
70
71
72
73
74
75
76
77
78
      rawTime <- xml_text(xml_find_all(TVP,".//wml2:time"))
      
      valueNodes <- xml_find_all(TVP,".//wml2:value")
      values <- as.numeric(xml_text(valueNodes))
      nVals <- length(values)
      gmlID <- rep(gmlID, nVals)
      
      #df of date, time, dateTime
      oneCol <- rep(NA, nVals) 
      timeDF <- data.frame(date=oneCol, time=oneCol, dateTime=oneCol)
      splitTime <- data.frame(matrix(unlist(strsplit(rawTime, "T")), nrow=nVals, byrow = TRUE), stringsAsFactors=FALSE)
79
80
81
82
83
84
85
      if(ncol(splitTime) > 1){ #some sites only have a date
        names(splitTime) <- c("date", "time")
      }else{
        names(splitTime) <- "date"
        splitTime <- mutate(splitTime, time = NA)
      }
      
86
87
88
      timeDF <- mutate(splitTime, dateTime = NA)
      logicVec <- nchar(rawTime) > 19
      timeDF$dateTime[logicVec] <- rawTime[logicVec]
David Watkins's avatar
David Watkins committed
89
      if(asDateTime){
90
        timeDF$dateTime <- parse_date_time(timeDF$dateTime, c("%Y","%Y-%m-%d","%Y-%m-%dT%H:%M","%Y-%m-%dT%H:%M:%S",
David Watkins's avatar
David Watkins committed
91
92
93
94
                                        "%Y-%m-%dT%H:%M:%OS","%Y-%m-%dT%H:%M:%OS%z"), exact = TRUE)
        #^^setting tz in as.POSIXct just sets the attribute, does not convert the time!
        attr(time, 'tzone') <- tz 
      }
95
96
       
      
David Watkins's avatar
David Watkins committed
97
      
David Watkins's avatar
David Watkins committed
98
      uom <- xml_attr(valueNodes, "uom", default = NA)
David Watkins's avatar
David Watkins committed
99
      source <- xml_attr(xml_find_all(TVP, ".//wml2:source"), "title")
David Watkins's avatar
David Watkins committed
100
      comment <- xml_text(xml_find_all(TVP, ".//wml2:comment"))
David Watkins's avatar
David Watkins committed
101
      
102
      df <- cbind.data.frame(source, timeDF, value=values, uom, comment, gmlID,
David Watkins's avatar
David Watkins committed
103
104
105
106
107
108
109
                             stringsAsFactors=FALSE)
      if (is.null(mergedDF)){
        mergedDF <- df
      } else {
        similarNames <- intersect(colnames(mergedDF), colnames(df))
        mergedDF <- full_join(mergedDF, df, by=similarNames)
      }
110
    }
David Watkins's avatar
David Watkins committed
111
112
113
114
115
    
    if(!raw){
      url <- input
      attr(mergedDF, "url") <- url
    }
David Watkins's avatar
David Watkins committed
116
117
118
    if(asDateTime){
      mergedDF$date <- as.Date(mergedDF$date)
    }
119
120
121
    nonDateCols <- grep("date",names(mergedDF), value=TRUE, invert = TRUE)
    
    mergedDF[nonDateCols][mergedDF[nonDateCols] == "" | mergedDF[nonDateCols]== -999999.0] <- NA
David Watkins's avatar
David Watkins committed
122
123
    attr(mergedDF, "gml:identifier") <- xml_text(xml_find_all(returnedDoc, ".//gml:identifier")) 
    attr(mergedDF, "generationDate") <- xml_text(xml_find_all(returnedDoc, ".//wml2:generationDate")) 
124
125
126
127
    meta <- xml_find_all(returnedDoc, ".//gmd:contact")
    attr(mergedDF, "contact") <- xml_attr(meta, "href")
    attr(mergedDF, "responsibleParty") <- xml_text(xml_find_all(meta, ".//gco:CharacterString"))
    
David Watkins's avatar
David Watkins committed
128
129
    
  }else if(response == "GetFeatureOfInterestResponse"){
130
131
    featureMembers <- xml_find_all(returnedDoc, ".//sos:featureMember")
    site <- xml_text(xml_find_all(featureMembers,".//gml:identifier"))
David Watkins's avatar
David Watkins committed
132
133
    site <- substring(site, 8)
    
134
135
    #some sites don't have a description
    siteDesc <- xml_text(xml_find_first(featureMembers, ".//gml:description"))
David Watkins's avatar
David Watkins committed
136
    
137
138
139
140
    siteLocs <- strsplit(xml_text(xml_find_all(featureMembers, ".//gml:pos")), " ")
    siteLocs <- data.frame(matrix(unlist(siteLocs), nrow=length(siteLocs), byrow=TRUE), stringsAsFactors = FALSE)
    names(siteLocs) <- c("dec_lat_va", "dec_lon_va")
    siteLocs <- mutate(siteLocs, dec_lat_va=as.numeric(dec_lat_va), dec_lon_va=as.numeric(dec_lon_va))
David Watkins's avatar
David Watkins committed
141
142
    mergedDF <- cbind.data.frame(site, description = siteDesc, siteLocs, stringsAsFactors = FALSE) 
  }
David Watkins's avatar
David Watkins committed
143
144
  else{
    stop("Unrecognized response from the web service")
145
146
147
  }
  return(mergedDF)
}