4.2. Python functions
Prev  4. The python module  Next

4.2. Python functions

dout, weights, masks, axes = spanlib.stackData(*data)

This fonction concatenates several dataset that have the same time axis. It is useful for analysing for example several variables at the same time. It takes into account weights, masks and axes.

  • *data [intent:input] :: One or more data objects to stack. They must all have the same time length.
  • dout [intent:output] :: Stacked data
  • weights [intent:output] :: Associated stacked weights
  • masks [intent:output] :: Associated stacked masks
  • axes [intent:output] :: Associated stacked axes 
dout = spanlib.unStackData(din, weights, masks, axes)

This function is the reverse operation of stakData. It splits stacked datasets into a list.

  • din [intent:input] :: Stacked data (see stackData function)
  • weights [intent:input] :: Associated stacked weights
  • masks [intent:input] :: Associated stacked masks
  • axes [intent:input] :: Associated stacked axes
  • dout [intent:output] :: List of unstacked data 
packed_data, packed_weights, mask = spanlib.pack(data, weights=None)

This function packs a dataset in 2D space by removing all masked points and returning a space-time array. It performs this operation also on the weights. It is used for removing unnecessary points and simplifying the input format for analysis functions.

  • data [intent:input] :: Flatten in space an [x,y,t] array by removing its masked point
  • weights [intent:input, default:None] :: Weights to be flatten also
  • packed_data [intent:output] :: Space-time packed array
  • packed_weights [intent:output] :: Packed weights that were guessed or used
  • mask [intent:output] :: Mask that were guessed or used 
phases = spanlib.computePhases(data, nphases=8, offset=.5, firstphase=0)

This computes temporal phase composites of a spatio-temporal dataset. The dataset is expected to be oscillatory in time. It corresponds to a reoganisation of the time axis to to represents the dataset over its cycle in a arbitrary number of phases. It is useful, for example, to have a synthetic view of an reconstructed MSSA oscillation.

  • data [intent:input] :: Space-time data oscillatory in time data.shape is rank 2 and dim 0 is space
  • nphases [intent:input, default:8] :: Number of phases (divisions of the cycle) offset :: Normalised offset to keep higher values only [default:
  • offset [intent:input, default:.5] ::
  • firstphase [intent:input, default:0] :: Position of the first phase in the 360 degree cycle
  • phases [intent:output] :: Space-phase array 
<SpAn_object> = spanlib.SpAn(data, weights=None, npca=None, window=None, nmssa=None)

This function creates an object for future analyses. It optionally initializes some parameters.

  • data [intent:input] :: Data on which to run the PC Analysis Last dimensions must represent the spatial dimensions. Analysis will be run on the first dimension.
  • weights [intent:input, default:1. everywhere] :: If you which to apply weights on some points, set weights to "0" where you wish to mask. The input data mask will be applied, using the union of all none spacial dimension mask. If the data are on a regular grid, area weights will be generated, if the cdutil (CDAT) module is available. npca :: Number of principal components to return [default: 10] nmssa :: Number of MSSA modes retained [default: 4] window :: MSSA window parameter [default: time_length/3.]
  • npca [intent:input, default:None] ::
  • window [intent:input, default:None] ::
  • nmssa [intent:input, default:None] ::
  • <SpAn_object> [intent:output] :: Object created for further analysis
eof, pc, ev = <SpAn_object>.pca(npca=None)
  • npca [intent:input, default:None] :: Number of principal components to return, default will be 10
  • eof [intent:output] :: EOF array
  • pc [intent:output] :: Principal Components array
  • ev [intent:output] :: Eigein Values array 
eof, pc, ev = <SpAn_object>.mssa(nmssa=None, pca=None, window=None)
  • nmssa [intent:input, default:None] :: Number of MSSA modes retained
  • pca [intent:input, default:None] :: If True, performs a preliminary PCA
  • window [intent:input, default:None] :: MSSA window parameter
  • eof [intent:output] :: EOF array
  • pc [intent:output] :: Principal Components array
  • ev [intent:output] :: Eigen Values array 
ffrec = <SpAn_object>.reconstruct(start=1, end=None, mssa=None, pca=None, phases=False, nphases=8, offset=.5, firstphase=0)

This function performs recontructions to retreive the the contribution of a selection of modes to the original field. By default, it recontructs from available PCA and MSSA results. Recontruction of MSSA modes also calls recontruction from of pre-PCA to get back to the original space. This function can optionally performs phase composites (useful for pairs of MSSA modes = oscillations) on MSSA recontructions.

  • start [intent:input, default:1] :: First mode
  • end [intent:input, default:None] :: Last mode
  • mssa [intent:input, default:None] :: Reconstruct MSSA if True
  • pca [intent:input, default:None] :: Reconstruct PCA if True
  • phases [intent:input, default:False] :: Operate phase reconstruction True/False (default is False)
  • nphases [intent:input, default:8] ::
  • offset [intent:input, default:.5] ::
  • firstphase [intent:input, default:0] ::
  • ffrec [intent:output] :: Reconstructed field
Prev  Up  Next
4. The python module  Home  4.3. Python examples
SourceForge.net Logo
This document was generated using xsltproc and perl.