Input Data Format¶

WRTDS requires two DataFrames and an optional metadata dictionary.

Daily DataFrame¶

One row per day of discharge record.

Column	Type	Description
`Date`	datetime	Date of observation
`Q`	float	Daily mean discharge in m^3^/s (must be > 0)

daily = pd.DataFrame({
    'Date': pd.date_range('1980-01-01', '2020-12-31'),
    'Q': discharge_values,
})

During construction, populate_daily adds the following derived columns:

Column	Description
`LogQ`	Natural log of Q
`Julian`	Days since 1850-01-01
`DecYear`	Decimal year (e.g. 2010.5 = July 2010)
`Month`	Month (1-12)
`Day`	Day of month
`MonthSeq`	Sequential month index from start of record

Sample DataFrame¶

Water-quality observations. Two input formats are supported.

Format 1: ConcLow / ConcHigh (Preferred)¶

Column	Type	Description
`Date`	datetime	Date of observation
`ConcLow`	float	Lower concentration bound
`ConcHigh`	float	Upper concentration bound

For uncensored observations, set ConcLow == ConcHigh (the measured value).

For left-censored observations (below detection limit), set ConcLow = 0 (or NaN) and ConcHigh to the detection limit.

Format 2: Conc / Remark¶

Column	Type	Description
`Date`	datetime	Date of observation
`Conc`	float	Concentration value
`Remark`	str	`'<'` for left-censored, empty/NaN for uncensored

This format is automatically converted to ConcLow/ConcHigh internally using compress_data.

Derived Sample Columns¶

After preparation, the Sample DataFrame gains:

Column	Description
`Uncen`	1 if uncensored, 0 if censored
`ConcAve`	Average of ConcLow and ConcHigh
`Q`	Discharge (merged from Daily by nearest date)
`LogQ`	Natural log of Q
`Julian`	Days since 1850-01-01
`DecYear`	Decimal year
`Month`, `Day`	Calendar month and day
`MonthSeq`	Sequential month index
`SinDY`	sin(2 * pi * DecYear) — seasonal harmonic
`CosDY`	cos(2 * pi * DecYear) — seasonal harmonic

Info Dictionary¶

Optional metadata dictionary. Missing keys are filled from DEFAULT_INFO:

DEFAULT_INFO = {
    'station_name': '',
    'param_name': '',
    'drainage_area_km2': None,
    'pa_start': 10,       # Period-of-analysis start month (10 = October = water year)
    'pa_long': 12,        # Period-of-analysis length in months
}

The pa_start and pa_long parameters define the period of analysis. The default (pa_start=10, pa_long=12) uses the water year (October through September). To use calendar years, set pa_start=1, pa_long=12.

w = WRTDS(daily, sample, info={
    'station_name': 'Choptank River at Greensboro',
    'param_name': 'Nitrate-N',
    'drainage_area_km2': 292.0,
    'pa_start': 10,
    'pa_long': 12,
})