validation

measurement.validation

Classes

Measurement

Bases: MeasurementBase

Class to store the measurements and their validation status.

This class holds the value of a given variable and station at a specific time, as well as auxiliary information such as maximum and minimum values, depth and direction, for vector quantities. All of these have a raw version where a backup of the original data is kept, should this change at any point.

Flags to monitor its validation status, if the data is active (and therefore can be used for reporting) and if it has actually been used for that is also included.

Attributes:

Name	Type	Description
depth	int	Depth of the measurement.
direction	Decimal	Direction of the measurement, useful for vector quantities.
raw_value	Decimal	Original value of the measurement.
raw_maximum	Decimal	Original maximum value of the measurement.
raw_minimum	Decimal	Original minimum value of the measurement.
raw_direction	Decimal	Original direction of the measurement.
raw_depth	int	Original depth of the measurement.
is_validated	bool	Flag to indicate if the measurement has been validated.
is_active	bool	Flag to indicate if the measurement is active. An inactive measurement is not used for reporting

Attributes

overwritten: bool property

Indicates if any of the values associated to the entry have been overwritten.

Returns:

Name	Type	Description
bool	bool	True if any raw field is different to the corresponding standard field.

raws: tuple[str, ...] property

Return the raw fields of the measurement.

Returns:

Type	Description
tuple[str, ...]	tuple[str]: Tuple with the names of the raw fields of the measurement.

Functions

clean()

Check consistency of validation, reporting and backs-up values.

Source code in measurement/models.py

def clean(self) -> None:
    """Check consistency of validation, reporting and backs-up values."""
    # Check consistency of validation
    if not self.is_validated and not self.is_active:
        raise ValidationError("Only validated entries can be declared as inactive.")

    # Backup values to raws, if needed
    for r in self.raws:
        value = getattr(self, r.removeprefix("raw_"))
        if value and not getattr(self, r):
            setattr(self, r, value)

Station

Bases: PermissionsBase

Main representation of a station, including several metadata.

Attributes:

Name	Type	Description
visibility	str	Visibility level of the object, including an "internal" option.
station_id	int	Primary key.
station_code	str	Unique code for the station.
station_name	str	Brief description of the station.
station_type	StationType	Type of the station.
country	Country	Country where the station is located.
region	Region	Region within the Country where the station is located.
ecosystem	Ecosystem	Ecosystem associated with the station.
institution	Institution	Institutional partner responsible for the station.
place_basin	PlaceBasin	Place-Basin association.
station_state	bool	Is the station operational?
timezone	str	Timezone of the station.
delta_t	DeltaT	Interval of data adquisition (in minutes).
station_latitude	Decimal	Latitude of the station, in degrees [-90 to 90].
station_longitude	Decimal	Longitude of the station, in degrees [-180 to 180].
station_altitude	int	Altitude of the station.
influence_km	Decimal	Area of influence in km2.
station_file	ImageField	Photography of the station.
station_external	bool	Is the station external?

Functions

__str__()

Return the station code.

Source code in station/models.py

def __str__(self) -> str:
    """Return the station code."""
    return str(self.station_code)

clean()

Set the default delta_t value if not provided.

Source code in station/models.py

def clean(self) -> None:
    """Set the default delta_t value if not provided."""
    super().clean()
    if not self.delta_t:
        self.delta_t = DeltaT.get_default()

get_absolute_url()

Return the absolute url of the station.

Source code in station/models.py

def get_absolute_url(self) -> str:
    """Return the absolute url of the station."""
    return reverse("station:station_detail", kwargs={"pk": self.pk})

set_object_permissions()

Set object-level permissions.

This method is called by the save method of the model to set the object-level permissions based on the visibility level of the object. In addition to the standard permissions for the station, the view_measurements permission is set which controls who can view the measurements associated to the station.

Source code in station/models.py

def set_object_permissions(self) -> None:
    """Set object-level permissions.

    This method is called by the save method of the model to set the object-level
    permissions based on the visibility level of the object. In addition to the
    standard permissions for the station, the view_measurements permission is set
    which controls who can view the measurements associated to the station.
    """
    super().set_object_permissions()

    standard_group = Group.objects.get(name="Standard")
    anonymous_user = get_anonymous_user()

    # Assign view_measurements permission based on permissions level
    if self.visibility == "public":
        assign_perm("view_measurements", standard_group, self)
        assign_perm("view_measurements", anonymous_user, self)
        if self.owner:
            remove_perm("view_measurements", self.owner, self)
    elif self.visibility == "internal":
        assign_perm("view_measurements", standard_group, self)
        remove_perm("view_measurements", anonymous_user, self)
        if self.owner:
            remove_perm("view_measurements", self.owner, self)
    elif self.visibility == "private":
        remove_perm("view_measurements", standard_group, self)
        remove_perm("view_measurements", anonymous_user, self)
        if self.owner:
            assign_perm("view_measurements", self.owner, self)

Variable

Bases: PermissionsBase

A variable with a physical meaning.

Such as precipitation, wind speed, wind direction, soil moisture, including the associated unit. It also includes metadata to help identify what is a reasonable value for the data, to flag outliers and to help with the validation process.

The nature of the variable can be one of the following:

• sum: Cumulative value over a period of time.
• average: Average value over a period of time.
• value: One-off value.

Attributes:

Name	Type	Description
variable_id	AutoField	Primary key.
variable_code	CharField	Code of the variable, eg. airtemperature.
name	CharField	Human-readable name of the variable, eg. Air temperature.
unit	ForeignKey	Unit of the variable.
maximum	DecimalField	Maximum value allowed for the variable.
minimum	DecimalField	Minimum value allowed for the variable.
diff_error	DecimalField	If two sequential values in the time-series data of this variable differ by more than this value, the validation process can mark this with an error flag.
outlier_limit	DecimalField	The statistical deviation for defining outliers, in times the standard deviation (sigma).
null_limit	DecimalField	The max % of null values (missing, caused by e.g. equipment malfunction) allowed for hourly, daily, monthly data. Cumulative values are not deemed trustworthy if the number of missing values in a given period is greater than the null_limit.
nature	CharField	Nature of the variable, eg. if it represents a one-off value, the average over a period of time or the cumulative value over a period

Attributes

is_cumulative: bool property

Return True if the nature of the variable is sum.

Functions

__str__()

Return the string representation of the object.

Source code in variable/models.py

def __str__(self) -> str:
    """Return the string representation of the object."""
    return str(self.name)

clean()

Validate the model fields.

Source code in variable/models.py

def clean(self) -> None:
    """Validate the model fields."""
    if self.maximum < self.minimum:
        raise ValidationError(
            {
                "maximum": "The maximum value must be greater than the minimum "
                "value."
            }
        )
    if not self.variable_code.isidentifier():
        raise ValidationError(
            {
                "variable_code": "The variable code must be a valid Python "
                "identifier. Only letters, numbers and underscores are allowed, and"
                " it cannot start with a number."
            }
        )
    return super().clean()

get_absolute_url()

Get the absolute URL of the object.

Source code in variable/models.py

def get_absolute_url(self) -> str:
    """Get the absolute URL of the object."""
    return reverse("variable:variable_detail", kwargs={"pk": self.pk})

Functions

flag_suspicious_daily_count(data, period, null_limit)

Finds suspicious records count for daily data.

Parameters:

Name	Type	Description	Default
data	Series	The count of records per day.	required
period	Decimal	The expected period for the measurements, in minutes.	required
null_limit	Decimal	The percentage of null data allowed.	required

Returns:

Type	Description
DataFrame	A dataframe with the suspicious data.

Source code in measurement/validation.py

def flag_suspicious_daily_count(
    data: pd.Series, period: Decimal, null_limit: Decimal
) -> pd.DataFrame:
    """Finds suspicious records count for daily data.

    Args:
        data: The count of records per day.
        period: The expected period for the measurements, in minutes.
        null_limit: The percentage of null data allowed.

    Returns:
        A dataframe with the suspicious data.
    """
    expected_data_count = 24 * 60 / float(period)
    suspicious = pd.DataFrame(index=data.index)
    suspicious["daily_count_fraction"] = (data / expected_data_count).round(2)
    suspicious["suspicious_daily_count"] = ~suspicious["daily_count_fraction"].between(
        1 - float(null_limit) / 100, 1
    ) | (suspicious["daily_count_fraction"] > 1)

    return suspicious

flag_suspicious_data(data, maximum, minimum, period, allowed_difference)

Finds suspicious data in the database.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with the data to be evaluated.	required
maximum	Decimal	The maximum allowed value.	required
minimum	Decimal	The minimum allowed value.	required
period	Decimal	The expected period for the measurements, in minutes.	required
allowed_difference	Decimal	The allowed difference between the measurements.	required

Returns:

Type	Description
DataFrame	A dataframe with the suspicious data.

Source code in measurement/validation.py

def flag_suspicious_data(
    data: pd.DataFrame,
    maximum: Decimal,
    minimum: Decimal,
    period: Decimal,
    allowed_difference: Decimal,
) -> pd.DataFrame:
    """Finds suspicious data in the database.

    Args:
        data: The dataframe with the data to be evaluated.
        maximum: The maximum allowed value.
        minimum: The minimum allowed value.
        period: The expected period for the measurements, in minutes.
        allowed_difference: The allowed difference between the measurements.

    Returns:
        A dataframe with the suspicious data.
    """
    time_lapse = flag_time_lapse_status(data, period)
    value_difference = flag_value_difference(data, allowed_difference)
    value_limits = flag_value_limits(data, maximum, minimum)
    return pd.concat([time_lapse, value_difference, value_limits], axis=1)

flag_time_lapse_status(data, period)

Flags if period of the time entries is correct.

It is assume that the first entry is correct. A tolerance of 2% of the period is used when deciding on the suspicious status.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with allowed_difference = Variable. the data.	required
period	Decimal	The expected period for the measurements, in minutes.	required

Returns:

Type	Description
Series	A series with the status of the time lapse.

Source code in measurement/validation.py

def flag_time_lapse_status(data: pd.DataFrame, period: Decimal) -> pd.Series:
    """Flags if period of the time entries is correct.

    It is assume that the first entry is correct. A tolerance of 2% of the period
    is used when deciding on the suspicious status.

    Args:
        data: The dataframe with allowed_difference = Variable. the data.
        period: The expected period for the measurements, in minutes.

    Returns:
        A series with the status of the time lapse.
    """
    flags = pd.DataFrame(index=data.index, columns=["suspicious_time_lapse"])
    low = pd.Timedelta(f"{period}min") * (1 - 0.02)
    high = pd.Timedelta(f"{period}min") * (1 + 0.02)
    flags["suspicious_time_lapse"] = ~data.time.diff().between(
        low, high, inclusive="both"
    )
    flags["suspicious_time_lapse"].iloc[0] = False
    return flags

flag_value_difference(data, allowed_difference)

Flags if the differences in value of the measurements is correct.

It is assume that the first entry is correct.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with allowed_difference = Variable. the data.	required
allowed_difference	Decimal	The allowed difference between the measurements.	required

Returns:

Type	Description
Series	A series with the status of the value.

Source code in measurement/validation.py

def flag_value_difference(data: pd.DataFrame, allowed_difference: Decimal) -> pd.Series:
    """Flags if the differences in value of the measurements is correct.

    It is assume that the first entry is correct.

    Args:
        data: The dataframe with allowed_difference = Variable. the data.
        allowed_difference: The allowed difference between the measurements.

    Returns:
        A series with the status of the value.
    """
    flags = pd.DataFrame(index=data.index, columns=["suspicious_value_difference"])
    flags["suspicious_value_difference"] = data["value"].diff().abs() > float(
        allowed_difference
    )
    flags["suspicious_value_difference"].iloc[0] = False
    return flags

flag_value_limits(data, maximum, minimum)

Flags if the values and limits of the measurements are within limits.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with allowed_difference = Variable. the data.	required
maximum	Decimal	The maximum allowed value.	required
minimum	Decimal	The minimum allowed value.	required

Returns:

Type	Description
DataFrame	A dataframe with suspicious columns indicating a problem.

Source code in measurement/validation.py

def flag_value_limits(
    data: pd.DataFrame, maximum: Decimal, minimum: Decimal
) -> pd.DataFrame:
    """Flags if the values and limits of the measurements are within limits.

    Args:
        data: The dataframe with allowed_difference = Variable. the data.
        maximum: The maximum allowed value.
        minimum: The minimum allowed value.

    Returns:
        A dataframe with suspicious columns indicating a problem.
    """
    flags = pd.DataFrame(index=data.index)
    flags["suspicious_value_limits"] = (data["value"] < minimum) | (
        data["value"] > maximum
    )
    if "maximum" in data.columns:
        flags["suspicious_maximum_limits"] = (data["maximum"] < minimum) | (
            data["maximum"] > maximum
        )
    if "minimum" in data.columns:
        flags["suspicious_minimum_limits"] = (data["minimum"] < minimum) | (
            data["minimum"] > maximum
        )
    return flags

generate_daily_summary(data, suspicious, period, null_limit, is_cumulative)

Generates a daily report of the data.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with the data to be evaluated.	required
suspicious	DataFrame	The dataframe with the suspicious data.	required
period	Decimal	The expected period for the measurements, in minutes.	required
null_limit	Decimal	The percentage of null data allowed.	required
is_cumulative	bool	If the data is cumulative and should be aggregated by sum.	required

Returns:

Type	Description
DataFrame	A dataframe with the daily report.

Source code in measurement/validation.py

def generate_daily_summary(
    data: pd.DataFrame,
    suspicious: pd.DataFrame,
    period: Decimal,
    null_limit: Decimal,
    is_cumulative: bool,
) -> pd.DataFrame:
    """Generates a daily report of the data.

    Args:
        data: The dataframe with the data to be evaluated.
        suspicious: The dataframe with the suspicious data.
        period: The expected period for the measurements, in minutes.
        null_limit: The percentage of null data allowed.
        is_cumulative: If the data is cumulative and should be aggregated by sum.

    Returns:
        A dataframe with the daily report.
    """
    report = pd.DataFrame(index=data.time.dt.date.unique())

    # Group the data by day and calculate the mean or sum
    datagroup = data.groupby(data.time.dt.date)
    report["value"] = (
        datagroup["value"].sum() if is_cumulative else datagroup["value"].mean()
    )

    if "maximum" in data.columns:
        report["maximum"] = datagroup["maximum"].max()
    if "minimum" in data.columns:
        report["minimum"] = datagroup["minimum"].min()

    # Count the number of entries per day and flag the suspicious ones
    count_report = flag_suspicious_daily_count(
        datagroup["value"].count(), period, null_limit
    )

    # Group the suspicious data by day and calculate the sum
    suspiciousgroup = suspicious.groupby(data.time.dt.date)
    suspicious_report = suspiciousgroup.sum().astype(int)
    suspicious_report["total_suspicious_entries"] = suspicious_report.sum(axis=1)

    # Put together the final report
    report = pd.concat([report, suspicious_report, count_report], axis=1)
    report = report.sort_index().reset_index().rename(columns={"index": "date"})
    report.date = pd.to_datetime(report.date)
    return report

generate_validation_report(station, variable, start_time, end_time, maximum, minimum, is_validated=False)

Generates a report of the data.

Parameters:

Name	Type	Description	Default
station	str	Station of interest.	required
variable	str	Variable of interest.	required
start_time	str	Start time.	required
end_time	str	End time.	required
maximum	Decimal	The maximum allowed value.	required
minimum	Decimal	The minimum allowed value.	required
is_validated	bool	Whether to retrieve validated or non-validated data.	False

Returns:

Type	Description
tuple[DataFrame, DataFrame]	A tuple with the summary report and the granular report.

Source code in measurement/validation.py

def generate_validation_report(
    station: str,
    variable: str,
    start_time: str,
    end_time: str,
    maximum: Decimal,
    minimum: Decimal,
    is_validated: bool = False,
) -> tuple[pd.DataFrame, pd.DataFrame]:
    """Generates a report of the data.

    Args:
        station: Station of interest.
        variable: Variable of interest.
        start_time: Start time.
        end_time: End time.
        maximum: The maximum allowed value.
        minimum: The minimum allowed value.
        is_validated: Whether to retrieve validated or non-validated data.

    Returns:
        A tuple with the summary report and the granular report.
    """
    period = Station.objects.get(station_code=station).delta_t.delta_t
    var = Variable.objects.get(variable_code=variable)

    data = get_data_to_validate(station, variable, start_time, end_time, is_validated)
    if data.empty:
        return pd.DataFrame(), pd.DataFrame()

    suspicious = flag_suspicious_data(data, maximum, minimum, period, var.diff_error)
    summary = generate_daily_summary(
        data, suspicious, period, var.null_limit, var.is_cumulative
    )
    granular = pd.concat([data, suspicious], axis=1)
    return summary, granular

get_data_to_validate(station, variable, start_time, end_time, is_validated=False)

Retrieves data to be validated.

Parameters:

Name	Type	Description	Default
station	str	Station of interest.	required
variable	str	Variable of interest.	required
start_time	str	Start time.	required
end_time	str	End time.	required
is_validated	bool	Whether to retrieve validated or non-validated data.	False

Returns:

Type	Description
DataFrame	A dictionary with the report for the chosen days.

Source code in measurement/validation.py

def get_data_to_validate(
    station: str,
    variable: str,
    start_time: str,
    end_time: str,
    is_validated: bool = False,
) -> pd.DataFrame:
    """Retrieves data to be validated.

    Args:
        station: Station of interest.
        variable: Variable of interest.
        start_time: Start time.
        end_time: End time.
        is_validated: Whether to retrieve validated or non-validated data.

    Returns:
        A dictionary with the report for the chosen days.
    """
    tz = zoneinfo.ZoneInfo(Station.objects.get(station_code=station).timezone)
    start_time_ = datetime.strptime(start_time, "%Y-%m-%d").replace(tzinfo=tz)
    end_time_ = datetime.strptime(end_time, "%Y-%m-%d").replace(tzinfo=tz)

    df = pd.DataFrame.from_records(
        Measurement.objects.filter(
            station__station_code=station,
            variable__variable_code=variable,
            time__gte=start_time_,
            time__lte=end_time_,
            is_validated=is_validated,
        ).values()
    )

    if df.empty:
        return df

    return df.sort_values("time")

reset_validated_days(station, variable, start_date, end_date)

Resets validation and active status for the selected data.

It also deletes the associated report data.

TODO: should this also reset any modified value, minimum or maximum entries?

Parameters:

Name	Type	Description	Default
station	str	Station code	required
variable	str	Variable code	required
start_date	str	Start date	required
end_date	str	End date	required

Source code in measurement/validation.py

def reset_validated_days(
    station: str, variable: str, start_date: str, end_date: str
) -> None:
    """Resets validation and active status for the selected data.

    It also deletes the associated report data.

    TODO: should this also reset any modified value, minimum or maximum entries?

    Args:
        station (str): Station code
        variable (str): Variable code
        start_date (str): Start date
        end_date (str): End date
    """
    tz = zoneinfo.ZoneInfo(Station.objects.get(station_code=station).timezone)

    # To update we use the exact date range.
    start_date_ = datetime.strptime(start_date, "%Y-%m-%d").replace(tzinfo=tz)
    end_date_ = datetime.strptime(end_date, "%Y-%m-%d").replace(tzinfo=tz)
    Measurement.objects.filter(
        station__station_code=station,
        variable__variable_code=variable,
        time__date__range=(start_date_.date(), end_date_.date()),
    ).update(is_validated=False, is_active=True)

    # To remove reports we use an extendaed date range to include the whole month.
    start_date_, end_date_ = reporting.reformat_dates(station, start_date, end_date)
    reporting.remove_report_data_in_range(station, variable, start_date_, end_date_)

reset_validated_entries(ids)

Resets validation and activation status for the selected data.

TODO: should this also reset any modified value, minimum or maximum entries?

Parameters:

Name	Type	Description	Default
ids	list	List of measurement ids to reset.	required

Source code in measurement/validation.py

def reset_validated_entries(ids: list) -> None:
    """Resets validation and activation status for the selected data.

    TODO: should this also reset any modified value, minimum or maximum entries?

    Args:
        ids (list): List of measurement ids to reset.
    """
    times = []
    for _id in ids:
        current = Measurement.objects.get(id=_id)
        current.is_validated = False
        current.is_active = True
        current.save()
        times.append(current.time)

    station = current.station.station_code
    variable = current.variable.variable_code
    start_time, end_time = reporting.reformat_dates(
        station, min(times).strftime("%Y-%m-%d"), max(times).strftime("%Y-%m-%d")
    )

    reporting.remove_report_data_in_range(station, variable, start_time, end_time)

save_validated_days(data)

Saves the validated days to the database and launches the report calculation.

Only the data that is flagged as "validate?" will be saved. The only updated field is is_active. To update the value, maximum or minimum, use save_validated_entries.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with the validated data.	required

Source code in measurement/validation.py

def save_validated_days(data: pd.DataFrame) -> None:
    """Saves the validated days to the database and launches the report calculation.

    Only the data that is flagged as "validate?" will be saved. The only updated field
    is is_active. To update the value, maximum or minimum, use save_validated_entries.

    Args:
        data: The dataframe with the validated data.
    """
    validate = data[data["validate?"]]
    for _, row in validate.iterrows():
        Measurement.objects.filter(
            station__station_code=row["station"],
            variable__variable_code=row["variable"],
            time__date=row["date"],
        ).update(is_validated=True, is_active=not row["deactivate?"])

    station = validate["station"].iloc[0]
    variable = validate["variable"].iloc[0]
    start_time = validate["date"].min()
    end_time = validate["date"].max()

    reporting.launch_reports_calculation(station, variable, start_time, end_time)

save_validated_entries(data)

Saves the validated data to the database.

Only the data that is flagged as "validate?" will be saved. Possible updated fields are: value, maximum, minimum and is_active.

Parameters:

Name	Type	Description	Default
data	DataFrame	The dataframe with the validated data.	required

Source code in measurement/validation.py

def save_validated_entries(data: pd.DataFrame) -> None:
    """Saves the validated data to the database.

    Only the data that is flagged as "validate?" will be saved. Possible updated fields
    are: value, maximum, minimum and is_active.

    Args:
        data: The dataframe with the validated data.
    """
    times: list[datetime] = []
    for _, row in data[data["validate?"]].iterrows():
        current = Measurement.objects.get(id=row["id"])
        times.append(current.time)

        update = {"is_validated": True, "is_active": not row["deactivate?"]}
        if current.value != row["value"]:
            update["value"] = row["value"]
        if "maximum" in row and current.maximum != row["maximum"]:
            update["maximum"] = row["maximum"]
        if "minimum" in row and current.minimum != row["minimum"]:
            update["minimum"] = row["minimum"]

        Measurement.objects.filter(id=row["id"]).update(**update)

    tz = zoneinfo.ZoneInfo(current.station.timezone)
    station = current.station.station_code
    variable = current.variable.variable_code
    start_time = min(times).astimezone(tz).strftime("%Y-%m-%d")
    end_time = max(times).astimezone(tz).strftime("%Y-%m-%d")

    reporting.launch_reports_calculation(station, variable, start_time, end_time)