Skip to contents

Individual quality-control tests for streamflow

streamflowQC_tests.Rd

Individual tests called by streamflowQC_daily and streamflowQC_subdaily. They are exported so each source of evidence can be run, inspected, deactivated, and tuned independently.

Usage

streamflowQC_range(x, lower=0, upper=Inf)

streamflowQC_duplicate(
  x, min.month.values=20L, min.year.values=300L, min.nonzero=1L
)

streamflowQC_gap(x, interval=NULL)

streamflowQC_climatology(
  x, group=c("dayofyear", "month", "month_slot"), window=5L,
  z=6, min.samples=30L, offset=0.01
)

streamflowQC_flatline(
  x, run=11L, high.run=Inf, high.prob=0.99,
  tolerance=0, flag.zero=FALSE
)

streamflowQC_spike(x, factor=5, prob=0.99)

streamflowQC_rate(
  x, group=c("month", "month_slot"), z=6, min.samples=30L,
  max.rise=Inf, max.fall=Inf, drop.factor=5, offset=0.01
)

streamflowQC_fluctuation(
  x, window=16L, minimum=8L, low.diff.prob=0.05
)

streamflowQC_highflow(
  x, log.z=6, amax.ratio=2, return.period=1000,
  min.years=10L, offset=0.01
)

streamflowQC_spatial(
  x, metadata=NULL, station.id="station", coords=c("lon", "lat"),
  area=NULL, elevation=NULL, elevation.scale=500,
  n.neighbours=5L, max.distance=400, min.neighbours=2L,
  min.support=1L, min.overlap=90L, min.correlation=0.5,
  window=3L, high.prob=0.99, support.prob=0.95, f=3
)

streamflowQC_breakpoint(
  x, min.years=10L, alpha=0.01, min.relative.change=0.5,
  min.completeness=0.8,
  indicators=c("mean", "sd", "minimum", "maximum")
)

Arguments

x

A numeric zoo object with one station per column.

lower, upper

Physical limits, scalar or station-specific.

min.month.values, min.year.values

Minimum paired calendar positions for exact copied-month and copied-year comparisons.

min.nonzero

Minimum positive observations in a candidate copied block.

interval

Expected interval: days for a Date index or seconds for a POSIXt index. NULL uses the modal difference.

group

Seasonal grouping. "month_slot" preserves the calendar month and hour:minute slot.

window

For climatology, circular calendar-day window width; for fluctuation, number of observations; for spatial QC, observations before and after the target event.

z

Standardized climatology or first-difference threshold.

min.samples

Minimum climatology or first-difference reference sample.

offset

Positive constant added before taking logarithms.

run, high.run

Minimum ordinary and high-flow flat-line lengths in observations.

high.prob

Target high-flow quantile.

tolerance

Maximum consecutive difference inside a flat line.

flag.zero

Whether repeated zero-flow runs may be flagged.

factor

Multiplicative isolated peak/dip threshold.

prob

Quantile used as the absolute spike threshold.

max.rise, max.fall

Optional absolute consecutive-change limits, scalar or station-specific.

drop.factor

Multiplicative abrupt-fall threshold.

minimum

Number of sign reversals that a fluctuation window must exceed.

low.diff.prob

Quantile of absolute changes treated as noise.

log.z

Global log-flow standard-deviation threshold.

amax.ratio

Largest/second-largest annual-maximum ratio threshold.

return.period

GEV return-period threshold in years.

min.years

Minimum annual values required for GEV or breakpoint analysis.

metadata

NULL, or station metadata containing all stations.

station.id

Name of the unique station identifier field.

coords

Names of longitude and latitude fields, in that order.

area

NULL, or the name of a positive numeric catchment-area field.

elevation

NULL, or the name of an optional numeric elevation field.

elevation.scale

Positive elevation-decay scale used in neighbour ranking.

n.neighbours

Maximum selected neighbours per station.

max.distance

Maximum geographic distance in km.

min.neighbours

Minimum simultaneous spatial predictions.

min.support

Minimum neighbours with a supporting high-flow event.

min.overlap

Minimum paired observations for selection and regression.

min.correlation

Minimum preferred Spearman log-flow correlation.

support.prob

Neighbour high-flow quantile.

f

Minimum standardized spatial residual.

alpha

Holm-adjusted Pettitt-test significance level.

min.relative.change

Minimum relative pre/post median change.

min.completeness

Minimum daily completeness of an annual indicator.

indicators

Unique subset of "mean", "sd", "minimum", and "maximum".

Details

streamflowQC_range treats negative discharge and optional station-specific upper violations as physical errors. Missing values are handled by the workflow completeness calculation. Non-finite non-missing values fail the range check.

streamflowQC_duplicate finds exactly copied calendar months or years. It is intended for daily data; requiring a positive value prevents ordinary all-zero intermittent-flow blocks from being flagged by default.

streamflowQC_gap marks the first finite observation after a missing value or missing interval. It complements, rather than replaces, station completeness. A numeric zoo cannot retain transmission syntax, so the QARTOD syntax test is enforced structurally by the workflow validators.

streamflowQC_climatology applies the GSIM transformation log(Q + 0.01) and mean plus/minus six standard deviations. Daily references use a circular calendar-day window; sub-daily references may use month or month-slot groups.

streamflowQC_flatline flags every member of an extended near-constant run. Daily defaults implement more than 10 equal positive values from GSIM. The sub-daily workflow converts the UK-Flow15-QC seven-day and one-day high-flow rules from hours to input observations.

streamflowQC_spike requires an isolated centre value to change and reverse by a factor of five or by more than the station 0.99 quantile. streamflowQC_rate combines robust monthly log-difference screening, optional absolute limits, and the five-fold rapid-drop rule. streamflowQC_fluctuation flags windows with recurrent first-derivative sign reversals after suppressing the smallest changes.

streamflowQC_highflow returns one combined flag plus separate six-standard-deviation, AMAX-ratio, and GEV components. The GEV is fitted to annual maxima by base-R maximum likelihood. The combined flag deliberately counts as one evidence family because all three statistics describe the same extreme event.

streamflowQC_spatial selects neighbours using correlation and optional geographic/elevation proximity, fits target-on-neighbour regressions in log-specific-discharge space when catchment area is available, and combines predictions using inverse-error weighted medians. A rare target event is flagged only when its residual score exceeds f and fewer than min.support neighbours exceed their support quantile in the search window.

streamflowQC_breakpoint aggregates sub-daily input to daily means, calculates complete annual indicators, applies Pettitt rank tests, and Holm-adjusts p-values. Its station flag is separate from point QC because an inhomogeneity may represent real regulation or catchment change.

Value

Range, duplicate, gap, climatology, flatline, spike, rate, and fluctuation tests return a logical zoo matching x.

streamflowQC_highflow returns a list with combined flags, component flag objects, return.period, and station fit diagnostics.

streamflowQC_spatial returns a list with flags, standardized scores, flow estimate, neighbour-event support.count, and selected neighbours.

streamflowQC_breakpoint returns one data.frame row per station.

References

Fileni, F. et al. (2026). UK-Flow15-QC: A quality control framework for better river flow data in hydrological research. EGUsphere preprint. doi:10.5194/egusphere-2026-277.

Gudmundsson, L. et al. (2018). The Global Streamflow Indices and Metadata Archive (GSIM) - Part 2: quality control, time-series indices and homogeneity assessment. Earth System Science Data, 10, 787–804. doi:10.5194/essd-10-787-2018.

U.S. Integrated Ocean Observing System (2018). Manual for Real-Time Quality Control of Stream Flow Data Version 1.0. doi:10.25923/gszc-ha43.

Zhao, Q. et al. (2018). Research on the Data-Driven Quality Control Method of Hydrological Time Series Data. Water, 10, 1712. doi:10.3390/w10121712.

See also

streamflowQC_daily, streamflowQC_subdaily

Examples

dates <- as.Date("2020-01-01") + 0:19
x <- zoo::zoo(cbind(A=c(rep(10, 11), 11:19),
                    B=seq(5, 24)), dates)
streamflowQC_flatline(x)
#>                A     B
#> 2020-01-01  TRUE FALSE
#> 2020-01-02  TRUE FALSE
#> 2020-01-03  TRUE FALSE
#> 2020-01-04  TRUE FALSE
#> 2020-01-05  TRUE FALSE
#> 2020-01-06  TRUE FALSE
#> 2020-01-07  TRUE FALSE
#> 2020-01-08  TRUE FALSE
#> 2020-01-09  TRUE FALSE
#> 2020-01-10  TRUE FALSE
#> 2020-01-11  TRUE FALSE
#> 2020-01-12 FALSE FALSE
#> 2020-01-13 FALSE FALSE
#> 2020-01-14 FALSE FALSE
#> 2020-01-15 FALSE FALSE
#> 2020-01-16 FALSE FALSE
#> 2020-01-17 FALSE FALSE
#> 2020-01-18 FALSE FALSE
#> 2020-01-19 FALSE FALSE
#> 2020-01-20 FALSE FALSE
streamflowQC_range(x)
#>                A     B
#> 2020-01-01 FALSE FALSE
#> 2020-01-02 FALSE FALSE
#> 2020-01-03 FALSE FALSE
#> 2020-01-04 FALSE FALSE
#> 2020-01-05 FALSE FALSE
#> 2020-01-06 FALSE FALSE
#> 2020-01-07 FALSE FALSE
#> 2020-01-08 FALSE FALSE
#> 2020-01-09 FALSE FALSE
#> 2020-01-10 FALSE FALSE
#> 2020-01-11 FALSE FALSE
#> 2020-01-12 FALSE FALSE
#> 2020-01-13 FALSE FALSE
#> 2020-01-14 FALSE FALSE
#> 2020-01-15 FALSE FALSE
#> 2020-01-16 FALSE FALSE
#> 2020-01-17 FALSE FALSE
#> 2020-01-18 FALSE FALSE
#> 2020-01-19 FALSE FALSE
#> 2020-01-20 FALSE FALSE