|
| 1 | + |
| 2 | +join_qw_uv <- function(qw_data, # data from readWQP |
| 3 | + uv_flow_qw, # data from readNWISuv |
| 4 | + hour_threshold = 24, # hours threshold for joining |
| 5 | + join_by_qw = "ActivityStartDateTime", |
| 6 | + join_by_uv = "dateTime", |
| 7 | + qw_val = "ResultMeasureValue", |
| 8 | + qw_rmk = "ResultDetectionConditionText", |
| 9 | + qw_det_val = "DetectionQuantitationLimitMeasure.MeasureValue", |
| 10 | + qw_val_uv, # water quality value column in uv data |
| 11 | + qw_rmk_uv, # water quality remark column in uv data |
| 12 | + flow_val = "X_00060_00000", # uv flow parameter |
| 13 | + flow_rmk = "X_00060_00000_cd"){ # uv flow parameter cd |
| 14 | + |
| 15 | + library(data.table) |
| 16 | + req_cols <- c(join_by_qw, qw_val, qw_rmk, qw_det_val) |
| 17 | + if(!all(req_cols %in% names(qw_data))){ |
| 18 | + stop(paste('qw_data missing columns:', req_cols[!req_cols %in% names(qw_data)])) |
| 19 | + } |
| 20 | + |
| 21 | + req_cols_uv <- c(join_by_uv) |
| 22 | + if(!all(req_cols_uv %in% names(uv_flow_qw))){ |
| 23 | + stop(paste('uv_data missing columns:', req_cols_uv[!req_cols_uv %in% names(uv_flow_qw)])) |
| 24 | + } |
| 25 | + |
| 26 | + data.table::setDT(qw_data)[, eval(parse(text = paste("join_date :=", join_by_qw)))] |
| 27 | + |
| 28 | + data.table::setDT(uv_flow_qw)[, eval(parse(text = paste("join_date :=", join_by_uv)))] |
| 29 | + |
| 30 | + # rolling join |
| 31 | + x <- uv_flow_qw[qw_data, on = .(join_date), roll = "nearest"] |
| 32 | + |
| 33 | + setnames(x, c(qw_val, join_by_uv, join_by_qw, qw_rmk, qw_det_val), |
| 34 | + c("val_qw","uv_date", "qw_date", "qw_rmk", "qw_det_val")) |
| 35 | + |
| 36 | + x <- x[order(qw_date)] |
| 37 | + |
| 38 | + x_tib <- as_tibble(x) |
| 39 | + |
| 40 | + if(!is.na(flow_val) | flow_val != ""){ |
| 41 | + x_tib$flow_uv <- x_tib[[flow_val]] |
| 42 | + } |
| 43 | + if(!is.na(flow_rmk) | flow_rmk != ""){ |
| 44 | + x_tib$flow_rmk_uv <- x_tib[[flow_rmk]] |
| 45 | + } |
| 46 | + |
| 47 | + if(!is.na(qw_val_uv) | qw_val_uv != ""){ |
| 48 | + x_tib$qw_val_uv <- x_tib[[qw_val_uv]] |
| 49 | + } |
| 50 | + if(!is.na(qw_rmk_uv) | qw_rmk_uv != ""){ |
| 51 | + x_tib$qw_rmk_uv <- x_tib[[qw_rmk_uv]] |
| 52 | + } |
| 53 | + |
| 54 | + toMatch <- c("NON-DETECT", "NON DETECT", "NOT DETECTED", |
| 55 | + "DETECTED NOT QUANTIFIED", "BELOW QUANTIFICATION LIMIT") |
| 56 | + |
| 57 | + x_tib <- x_tib |> |
| 58 | + mutate(delta_time = difftime(qw_date, uv_date, units = "hours"), |
| 59 | + qw_val_uv = if_else(abs(as.numeric(delta_time)) >= hour_threshold, |
| 60 | + NA, qw_val_uv), |
| 61 | + qualifier = if_else(grepl(paste(toMatch,collapse="|"), |
| 62 | + toupper(qw_rmk)), |
| 63 | + "<", ""), |
| 64 | + value = if_else(qualifier == "<", qw_det_val, val_qw), |
| 65 | + date = as.Date(qw_date)) |> |
| 66 | + select(any_of(c("uv_date", "qw_date", "delta_time", "date", |
| 67 | + "qw_val_uv", "qw_rmk_uv", |
| 68 | + "value", "qualifier", |
| 69 | + "flow_uv", "flow_rmk_uv"))) |> |
| 70 | + rename(dateTime = qw_date) |
| 71 | + |
| 72 | + |
| 73 | + compressedData <- EGRET::compressData(x_tib[, c("date", |
| 74 | + "qualifier", |
| 75 | + "value")], |
| 76 | + verbose = FALSE) |
| 77 | + Sample <- EGRET::populateSampleColumns(compressedData) |
| 78 | + Sample <- Sample |> |
| 79 | + left_join(x_tib |> |
| 80 | + select(-qualifier) |> |
| 81 | + rename(qw_dateTime = dateTime, |
| 82 | + uv_dateTime = uv_date, |
| 83 | + Date = date, |
| 84 | + ConcHigh = value), |
| 85 | + by = c("Date", "ConcHigh")) |
| 86 | + |
| 87 | + return(Sample) |
| 88 | + |
| 89 | +} |
0 commit comments