Base

area_weighting(var)

Apply area weighting to the variable var using the cosine of latitude: \(\cos (\phi)\).

Parameters:

Name	Type	Description	Default
var	DataArray	Variable to weight e.g. ds.t_surf to weight the surface temperature, where ds is the dataset for the experiment which contains all variables.	required

Returns:

Type	Description
DataArrayWeighted	Area weighted version of var.

Source code in isca_tools/utils/base.py

def area_weighting(var: xr.DataArray) -> DataArrayWeighted:
    """
    Apply area weighting to the variable `var` using the `cosine` of latitude: $\cos (\phi)$.

    Args:
        var: Variable to weight e.g. `ds.t_surf` to weight the surface temperature, where
            `ds` is the dataset for the experiment which contains all variables.

    Returns:
        Area weighted version of `var`.
    """
    weights = np.cos(np.deg2rad(var.lat))
    weights.name = "weights"
    return var.weighted(weights)

dp_from_pressure(p, dim='lev')

Compute layer pressure thickness Δp, preserving extra dims and coord order.

Parameters:

Name	Type	Description	Default
p	DataArray	Pressure [Pa], with vertical dimension dim (n_lev). Can include other dims (e.g., time, lat, lon).	required
dim	str	Name of the vertical coordinate.	'lev'

Returns:

Type	Description
DataArray	xr.DataArray: Pressure thickness Δp [Pa], same shape and coords as p.

Source code in isca_tools/utils/base.py

def dp_from_pressure(p: xr.DataArray, dim: str = "lev") -> xr.DataArray:
    """Compute layer pressure thickness Δp, preserving extra dims and coord order.

    Args:
        p: Pressure [Pa], with vertical dimension `dim` (n_lev).
            Can include other dims (e.g., time, lat, lon).
        dim: Name of the vertical coordinate.

    Returns:
        xr.DataArray: Pressure thickness Δp [Pa], same shape and coords as `p`.
    """

    def _dp_1d(p_1d: np.ndarray) -> np.ndarray:
        # ensure increasing order (bottom→top) for calculation
        reversed_flag = p_1d[0] < p_1d[-1]
        if reversed_flag:
            p_1d = p_1d[::-1]

        # edges & dp calculation
        p_edge_mid = 0.5 * (p_1d[:-1] + p_1d[1:])
        p_edge_bot = p_1d[0] + 0.5 * (p_1d[0] - p_1d[1])
        p_edge_top = p_1d[-1] - 0.5 * (p_1d[-2] - p_1d[-1])
        p_edges = np.concatenate([[p_edge_bot], p_edge_mid, [p_edge_top]])
        dp = p_edges[:-1] - p_edges[1:]
        dp = np.abs(dp)  # ensure positive

        # if we reversed order, un-reverse result to match original orientation
        if reversed_flag:
            dp = dp[::-1]
        return dp

    dp = xr.apply_ufunc(
        _dp_1d,
        p,
        input_core_dims=[[dim]],
        output_core_dims=[[dim]],
        vectorize=True,
        dask="parallelized",
        output_dtypes=[float],
    )

    dp.name = "dp"
    dp.attrs.update({"long_name": "pressure thickness", "units": "Pa"})
    return dp

get_memory_usage()

Get current process’s memory in MB.

Returns:

Name	Type	Description
mem_mb	float	Memory usage in MB

Source code in isca_tools/utils/base.py

def get_memory_usage() -> float:
    """
    Get current process’s memory in MB.

    Returns:
        mem_mb: Memory usage in MB
    """
    process = psutil.Process(os.getpid())
    mem_mb = process.memory_info().rss / (1024 * 1024)
    return mem_mb

has_out_of_range(val, min_range, max_range)

Check if any number within val is outside the range between min_range and max_range.

Parameters:

Name	Type	Description	Default
val	Union[List, Tuple, ndarray, float]	Numbers to check	required
min_range	float	Minimum allowed value.	required
max_range	float	Maximum allowed value.	required

Returns:

Type	Description
bool	True if there is a value outside the range between min_range and max_range.

Source code in isca_tools/utils/base.py

def has_out_of_range(val: Union[List, Tuple, np.ndarray, float], min_range: float, max_range: float) -> bool:
    """
    Check if any number within `val` is outside the range between `min_range` and `max_range`.

    Args:
        val: Numbers to check
        min_range: Minimum allowed value.
        max_range: Maximum allowed value.

    Returns:
        True if there is a value outside the range between `min_range` and `max_range`.
    """
    # If it's a single number, make it a list
    vals = val if isinstance(val, (list, tuple, np.ndarray)) else [val]
    return any((x < min_range or x > max_range) for x in vals)

len_safe(x)

Return length of x which can have multiple values, or just be a number.

Parameters:

Name	Type	Description	Default
x		Variable to return length of.	required

Returns:

Type	Description
int	Number of elements in x.

Source code in isca_tools/utils/base.py

def len_safe(x) -> int:
    """
    Return length of `x` which can have multiple values, or just be a number.

    Args:
        x: Variable to return length of.

    Returns:
        Number of elements in `x`.
    """
    if isinstance(x, numbers.Number):
        return 1
    try:
        return len(x)
    except TypeError:
        raise TypeError(f"Unsupported type with no length: {type(x)}")

parse_int_list(value, format_func=lambda x: str(x), all_values=None)

Takes in a value or list of values e.g. [1, 2, 3] and converts it into a list of strings where each string has the format given by format_func e.g. ['1', '2', '3'] for the default case.

There are three string options for value: * value='x:y', will return all integers between x and y inclusive. * value='firstX' will return first X values of all_values. * value='firstY' will return first Y values of all_values.

Parameters:

Name	Type	Description	Default
value	Union[str, int, List]	Variable to convert into list of strings	required
format_func	Callable	How to format each integer within the string.	lambda x: str(x)
all_values	Optional[List]	List of all possible integers, must be provided if value='firstX' or value='firstY'.	None

Returns:

Type	Description
List	List, where each integer in value is converted using format_func.

Source code in isca_tools/utils/base.py

def parse_int_list(value: Union[str, int, List], format_func: Callable = lambda x: str(x),
                   all_values: Optional[List] = None) -> List:
    """
    Takes in a value or list of values e.g. `[1, 2, 3]` and converts it into a list of strings where
    each string has the format given by `format_func` e.g. `['1', '2', '3']` for the default case.

    There are three string options for `value`:
    * `value='x:y'`, will return all integers between `x` and `y` inclusive.
    * `value='firstX'` will return first X values of `all_values`.
    * `value='firstY'` will return first Y values of `all_values`.

    Args:
        value: Variable to convert into list of strings
        format_func: How to format each integer within the string.
        all_values: List of all possible integers, must be provided if `value='firstX'` or `value='firstY'`.

    Returns:
        List, where each integer in `value` is converted using `format_func`.
    """
    if isinstance(value, list):
        pass
    elif isinstance(value, int):
        value = [value]
    elif isinstance(value, str):
        value = value.strip()       # remove blank space
        # Can specify just first or last n years
        if re.search(r'^first(\d+)', value):
            if all_values is None:
                raise ValueError(f'With value={value}, must provide all_values')
            n_req = int(re.search(r'^first(\d+)', value).group(1))
            if n_req > len(all_values):
                warnings.warn(f"Requested {value} but there are only "
                              f"{len(all_values)} available:\n{all_values}")
            value = all_values[:n_req]
        elif re.search(r'^last(\d+)', value):
            if all_values is None:
                raise ValueError(f'With value={value}, must provide all_values')
            n_req = int(re.search(r'^last(\d+)', all_values).group(1))
            if n_req > len(all_values):
                warnings.warn(f"Requested {value} but there are only "
                              f"{len(all_values)} available:\n{all_values}")
            value = all_values[-n_req:]
        elif ':' in value:
            # If '1979:2023' returns all integers from 1979 to 2023
            start, end = map(int, value.split(':'))
            value = list(range(start, end + 1))
        else:
            value = [int(value)]
    else:
        raise ValueError(f"Unsupported format: {value}")
    return [format_func(i) for i in value]

print_log(text, logger=None)

Quick function to add to log if log exists, otherwise print it.

Parameters:

Name	Type	Description	Default
text	str	Text to be printed.	required
logger	Optional[Logger]		None

Returns:

Source code in isca_tools/utils/base.py

def print_log(text: str, logger:Optional[logging.Logger] = None) -> None:
    """
    Quick function to add to log if log exists, otherwise print it.

    Args:
        text: Text to be printed.
        logger:

    Returns:

    """
    logger.info(text) if logger else print(text)

round_any(x, base, round_type='round')

Rounds x to the nearest multiple of base with the rounding done according to round_type.

Parameters:

Name	Type	Description	Default
x	Union[float, ndarray]	Number or array to round.	required
base	float	Rounds x to nearest integer multiple of value of base.	required
round_type	str	One of the following, indicating how to round x - 'round' 'ceil' 'float'	'round'

Returns:

Type	Description
Union[float, ndarray]	Rounded version of x.

Example

round_any(3, 5) = 5
round_any(3, 5, 'floor') = 0

Source code in isca_tools/utils/base.py

def round_any(x: Union[float, np.ndarray], base: float, round_type: str = 'round') -> Union[float, np.ndarray]:
    """
    Rounds `x` to the nearest multiple of `base` with the rounding done according to `round_type`.

    Args:
        x: Number or array to round.
        base: Rounds `x` to nearest integer multiple of value of `base`.
        round_type: One of the following, indicating how to round `x` -

            - `'round'`
            - `'ceil'`
            - `'float'`

    Returns:
        Rounded version of `x`.

    Example:
        ```
        round_any(3, 5) = 5
        round_any(3, 5, 'floor') = 0
        ```
    """
    if round_type == 'round':
        return base * np.round(x / base)
    elif round_type == 'ceil':
        return base * np.ceil(x / base)
    elif round_type == 'floor':
        return base * np.floor(x / base)
    else:
        raise ValueError(f"round_type specified was {round_type} but it should be one of the following:\n"
                         f"round, ceil, floor")

run_func_loop(func, max_wait_time=300, wait_interval=20, func_check=None, logger=None)

Safe way to run a function, such that if hit error, will try again every wait_interval seconds up to a maximum of max_wait_time seconds.

If func_check is given and returns True at any point, it will exit the loop without executing func.

Most obvious usage is for creating a directory e.g. os.makedirs, especially to a server where connection cuts in and out.

Parameters:

Name	Type	Description	Default
func	Callable	Function to run. Must have no arguments.	required
max_wait_time	int	Maximum number of seconds to try and run func.	300
wait_interval	int	Interval in seconds to wait between running func.	20
func_check	Optional[Callable]	Function that returns a boolean. If it returns True at any point, the loop will exit the loop without executing func.	None
logger	Optional[Logger]	Logger to record information	None

Returns:

Type	Description
	Whatever func returns.

Source code in isca_tools/utils/base.py

def run_func_loop(func: Callable, max_wait_time: int = 300, wait_interval: int = 20,
                  func_check: Optional[Callable] = None, logger: Optional[logging.Logger] = None):
    """
    Safe way to run a function, such that if hit error, will try again every `wait_interval` seconds up to a
    maximum of `max_wait_time` seconds.

    If `func_check` is given and returns `True` at any point, it will exit the loop without executing `func`.

    Most obvious usage is for creating a directory e.g. `os.makedirs`, especially to a server where connection
    cuts in and out.

    Args:
        func: Function to run. Must have no arguments.
        max_wait_time: Maximum number of seconds to try and run `func`.
        wait_interval: Interval in seconds to wait between running `func`.
        func_check: Function that returns a boolean. If it returns `True` at any point, the loop
            will exit the loop without executing `func`.
        logger: Logger to record information

    Returns:
        Whatever `func` returns.
    """
    i = 0
    j = 0
    success = False
    start_time = time.time()
    output = None
    while not success and (time.time() - start_time) < max_wait_time:
        if func_check is not None:
            if func_check():
                print_log("func_check passed so did not exectute func", logger)
                success = True
                break
        try:
            output = func()
            success = True
        except PermissionError as e:
            i += 1
            if i == 1:
                # Only print on first instance of error
                print_log(f'Permission Error: {e}', logger)
            time.sleep(wait_interval)
        except Exception as e:
            j += 1
            if j == 1:
                # Only print on first instance of error
                print_log(f'Unexpected Error: {e}', logger)
            time.sleep(wait_interval)
    if not success:
        raise ValueError(f"Making output directory - Failed to run function after {max_wait_time} seconds.")
    return output

split_list_max_n(lst, n)

Split lst into balanced chunks with at most n elements each.

Parameters:

Name	Type	Description	Default
lst	Union[List, ndarray]	List to split.	required
n	int	Maximum number of elements in each chunk of lst.	required

Returns:

Type	Description
List	List of n chunks of lst

Source code in isca_tools/utils/base.py

def split_list_max_n(lst: Union[List, np.ndarray], n: int) -> List:
    """
    Split `lst` into balanced chunks with at most `n` elements each.

    Args:
        lst: List to split.
        n: Maximum number of elements in each chunk of `lst`.

    Returns:
        List of `n` chunks of `lst`
    """
    k = int(np.ceil(len(lst) / n))  # Number of chunks needed
    avg = int(np.ceil(len(lst) / k))
    return [lst[i * avg : (i + 1) * avg] for i in range(k)]

top_n_peaks_ind(var, n=1, min_ind_spacing=0)

Return the indices of the N largest values of var, such that the indices of these values are ≥min_ind_spacing apart.

Parameters:

Name	Type	Description	Default
var	ndarray	1D array containing variable values. Assumed in an order	required
n	int	Number of peaks to select.	1
min_ind_spacing	int	Minimum index spacing between selected peaks.	0

Returns:

Type	Description
ndarray	Indices of n peak values of var.

Source code in isca_tools/utils/base.py

def top_n_peaks_ind(
    var: np.ndarray,
    n: int = 1,
    min_ind_spacing: int = 0,
) -> np.ndarray:
    """Return the indices of the N largest values of `var`, such that the indices of these values
     are ≥`min_ind_spacing` apart.

    Args:
        var: 1D array containing variable values. Assumed in an order
        n: Number of peaks to select.
        min_ind_spacing: Minimum index spacing between selected peaks.

    Returns:
        Indices of `n` peak values of `var`.
    """
    # Sort indices by descending value of var
    order = np.argsort(var)[::-1]
    selected_ind = []

    for i in order:
        # Check spacing constraint
        if all(abs(i - s) >= min_ind_spacing for s in selected_ind):
            selected_ind.append(i)
            if len(selected_ind) == n:
                break

    return np.array(selected_ind, dtype=int)

weighted_RMS(var, weight=None, dim=None)

Compute (weighted) RMS of a DataArray or numpy array along specified dimension(s).

Parameters:

Name	Type	Description	Default
var	Union[DataArray, ndarray]	Variable to compute RMS for (shape [...]).	required
weight	Optional[Union[DataArray, ndarray]]	Weights (same shape as var along dim). If None, computes unweighted RMS.	None
dim	Optional[Union[str, int, List[Union[str, int]]]]	Dimension(s) to reduce over. - For xarray: names of dimensions. - For numpy: integer axis or list of axes.	None

Returns:

Name	Type	Description
rms	Union[DataArray, ndarray]	Same type as input, reduced along dim.

Source code in isca_tools/utils/base.py

def weighted_RMS(
    var: Union[xr.DataArray, np.ndarray],
    weight: Optional[Union[xr.DataArray, np.ndarray]] = None,
    dim: Optional[Union[str, int, List[Union[str, int]]]] = None
) -> Union[xr.DataArray, np.ndarray]:
    """
    Compute (weighted) RMS of a DataArray or numpy array along specified dimension(s).

    Args:
        var: Variable to compute RMS for (shape [...]).
        weight: Weights (same shape as var along `dim`).
            If None, computes unweighted RMS.
        dim: Dimension(s) to reduce over.
            - For xarray: names of dimensions.
            - For numpy: integer axis or list of axes.

    Returns:
        rms: Same type as input, reduced along `dim`.
    """

    # --- Handle dim input uniformly ---
    if isinstance(dim, (str, int)):
        dims = [dim]
    elif dim is None:
        # all dims
        if isinstance(var, xr.DataArray):
            dims = list(var.dims)
        else:
            dims = list(range(var.ndim))
    else:
        dims = dim

    # --- xarray branch ---
    if isinstance(var, xr.DataArray):
        if weight is None:
            rms_sq = (var ** 2).mean(dim=dims)
        else:
            rms_sq = ((var ** 2) * weight).sum(dim=dims) / weight.sum(dim=dims)
        return np.sqrt(rms_sq)

    # --- numpy branch ---
    else:
        if weight is None:
            rms_sq = np.nanmean(var ** 2, axis=tuple(dims))
        else:
            rms_sq = np.nansum((var ** 2) * weight, axis=tuple(dims)) / np.nansum(weight, axis=tuple(dims))
        return np.sqrt(rms_sq)