stfl_trend_1stn_1stat: Estimate trend for 1 station and 1 statistic

In bcgov/StreamFlowTrend: Trend analysis on annual streamflow metrics from HYDAT database

Description

Estimate trend for 1 station for 1 statistics using least squares (LM), Mann-Kendall (MK), or robust least squares (LMrob) with automatic HAC corrections.

Usage

stfl_trend_1stn_1stat(
  data,
  log.Y = FALSE,
  Year.var = "Year",
  Value.var = "Value",
  Parameter.var = "Parameter",
  Statistic.var = "Statistic",
  STATION_NUMBER.var = "STATION_NUMBER",
  methods = c("LMrob", "LM", "MK"),
  offset = NULL,
  nsim.bbmks = 2000,
  trend.min.n = stfl_options()$trend.min.n
)

Arguments

data	Data frame with X and Y variables
log.Y	Should the analysis be done on the log(Y+offset) scale where log() referes to natural logarithms. If log(y) makes no sense (e.g., if statistic is DoY), the log() is ignored and set to FALSE automatically. Refer to the offset argument for more details. In most cases, the analysis on the log(Y) scale is preferred because of the simple interpretation of the trend. For example, an slope of .02 on the log(Y) corresponds to a 2% increase/year regardless of the underlying units. In some cases, a log(Y) transform makes no sense. For example, the Day_of_Year (DoY) when a minimum occurs is arbitrary and depends on when the start of the year begins (e.g. a water year starts on 1 October). The stfl_get_avail_stat() function has a list of available statistics and if a log(Y) transform is allowable.
Year.var	Name of the Year variable (usually the year the data was collected)
Value.var	Character string with the name of the variable containing the value to be analyzed.
Parameter.var	Character string with the name of the variable containing the name of the Parameter in the daily data and in the output results.
Statistic.var	Character string with the name of the variable containing the statistic that was computed
STATION_NUMBER.var	Character string with the name of the variable containing the station number in the daily data and in the output results..
methods	Which methods should be used to estimate trend lines. Possible values are one or more of LM: Ordinary regression using the lm() function MK: Mann-Kendall estimates using Sen estimator LMrob: Robust linear regression fit using the lmrob() function from the robustbase package
offset	What is the value of the offset used when taking log(Y+offset). If unspecified, then the offset is 1/2 of the smallest positive values. The offset is needed to avoid taking log(0).
nsim.bbmks	Number of bootstrapped simulations for Mann-Kendall block boot strap
trend.min.n	Minimum number of data points for trend computation

Value

List with the following elements

• data.augData frame augmented with the LM, MK, and LMrob predictions as specified in the methods argument

• estimatesData frame with estimates and standard errors from methods specified in methods argument and the following columns Parameter, Statistic, Year, Value, log.Y, estimate (intercept or Y variable), std error (ignoring autocorrelation), p.value ignoring autocorrelation, estimated autocorrelation, std.error adjusted for HAC, p.value adjusted or HAC

Examples

## Not run: 
# needs the HYDAT package installed
station.id <- "08NM116"
data <- stfl_get_annual_stat(station.id, Statistic=c("MEAN"))
temp <- stfl_trend_1stn_1stat(data, log.Y=TRUE)
temp

## End(Not run)

bcgov/StreamFlowTrend documentation built on Dec. 19, 2021, 7:38 a.m.