Class DatumShiftTransform

All Implemented Interfaces:
Serializable, Parameterized, LenientComparable, org.opengis.referencing.operation.MathTransform
Direct Known Subclasses:
InterpolatedGeocentricTransform, InterpolatedTransform, MolodenskyFormula

public abstract class DatumShiftTransform extends AbstractMathTransform implements Serializable
Transforms between two CRS (usually geographic) based on different datum. A datum shift may be needed when two CRS use different ellipsoids as approximation of the shape of the Earth. Sometimes two CRS use the same ellipsoid but with different anchor point (i.e. their coordinate systems have their origin in different locations).

There is many different datum shift methods, ranging from transformations as simple as adding a constant offset to geographic coordinates, to more complex transformations involving conversions to geocentric coordinates and/or interpolations in a datum shift grid. The simple cases like adding a constant offset are handled by other MathTransform implementations like LinearTransform. More complex methods are subclasses of this DatumShiftTransform base class, but users should not assume that this is the case of every transforms performing a datum shift.

Datum shift methods overview

The two CRS's ellipsoids have slightly different scale and rotation in space, and their center are located in a slightly different position. Consequently, geodetic datum shifts are often approximated by a constant scale, rotation and translation applied on geocentric coordinates. Those approximations are handled in SIS by concatenations of EllipsoidToCentricTransform with LinearTransform instead of a specific DatumShiftTransform subclass.

If the geodetic datum shifts is approximated only by a geocentric translation without any scale or rotation, and if an error of a few centimetres it acceptable, then the MolodenskyTransform subclass can be used as an approximation of the above method. The Molodensky method requires less floating point operations since it applies directly on geographic coordinates, without conversions to geocentric coordinates.

Some countries go one step further and allow the above geocentric translations to be non-constant. Instead, a different geocentric translation is interpolated for each geographic input coordinates. This case is handled by the InterpolatedGeocentricTransform subclass, or its InterpolatedMolodenskyTransform variant if a few centimetres accuracy lost can be afforded.

A simpler alternative to the above is to interpolate translations to apply directly on geographic coordinates. This is the approach taken by NADCON and NTv2 grids. SIS handles those datum shifts with the InterpolatedTransform subclass.

Since:
0.7
Version:
1.0
See Also:
  • Field Details

    • serialVersionUID

      private static final long serialVersionUID
      Serial number for inter-operability with different versions.
      See Also:
    • context

      final ContextualParameters context
      The parameters used for creating this datum shift. They are used for formatting Well Known Text (WKT) and error messages. Subclasses shall not use the values defined in this object for computation purpose, except at construction time.
      See Also:
    • grid

      final DatumShiftGrid<?,?> grid
      The grid of datum shifts from source datum to target datum, or null if none.
      See Also:
    • scaleX

      private transient double scaleX
      Conversion from (λ,φ) coordinates in radians to grid indices (x,y).
      • x = (λ - λ₀) ⋅ scaleX = λ ⋅ scaleX + x₀
      • y = (φ - φ₀) ⋅ scaleY = φ ⋅ scaleY + y₀
      Those factors are extracted from the DatumShiftGrid.getCoordinateToGrid() transform for performance reasons.
    • scaleY

      private transient double scaleY
      Conversion from (λ,φ) coordinates in radians to grid indices (x,y).
      • x = (λ - λ₀) ⋅ scaleX = λ ⋅ scaleX + x₀
      • y = (φ - φ₀) ⋅ scaleY = φ ⋅ scaleY + y₀
      Those factors are extracted from the DatumShiftGrid.getCoordinateToGrid() transform for performance reasons.
    • x0

      private transient double x0
      Conversion from (λ,φ) coordinates in radians to grid indices (x,y).
      • x = (λ - λ₀) ⋅ scaleX = λ ⋅ scaleX + x₀
      • y = (φ - φ₀) ⋅ scaleY = φ ⋅ scaleY + y₀
      Those factors are extracted from the DatumShiftGrid.getCoordinateToGrid() transform for performance reasons.
    • y0

      private transient double y0
      Conversion from (λ,φ) coordinates in radians to grid indices (x,y).
      • x = (λ - λ₀) ⋅ scaleX = λ ⋅ scaleX + x₀
      • y = (φ - φ₀) ⋅ scaleY = φ ⋅ scaleY + y₀
      Those factors are extracted from the DatumShiftGrid.getCoordinateToGrid() transform for performance reasons.
  • Constructor Details

    • DatumShiftTransform

      DatumShiftTransform(org.opengis.parameter.ParameterDescriptorGroup descriptor, DatumShiftGrid<?,?> grid)
      Creates a datum shift transform for direct interpolations in a grid. It is caller responsibility to initialize the context parameters.
      Parameters:
      descriptor - the contextual parameter descriptor.
      grid - interpolation grid.
    • DatumShiftTransform

      DatumShiftTransform(org.opengis.parameter.ParameterDescriptorGroup descriptor, boolean isSource3D, boolean isTarget3D, DatumShiftGrid<?,?> grid)
      Creates a datum shift transform for interpolations in geocentric domain. It is caller responsibility to initialize the context parameters.
      Parameters:
      descriptor - the contextual parameter descriptor.
      isSource3D - true if the source coordinates have a height.
      isTarget3D - true if the target coordinates have a height.
      grid - interpolation grid in geocentric coordinates, or null if none.
  • Method Details

    • readObject

      private void readObject(ObjectInputStream in) throws IOException, ClassNotFoundException
      Invoked after deserialization. This method computes the transient fields.
      Parameters:
      in - the input stream from which to deserialize the datum shift grid.
      Throws:
      IOException - if an I/O error occurred while reading or if the stream contains invalid data.
      ClassNotFoundException - if the class serialized on the stream is not on the classpath.
    • computeConversionFactors

      private void computeConversionFactors()
      Computes the conversion factors needed for calls to DatumShiftGrid.interpolateInCell(double, double, double[]). This method takes only the 2<C extends javax.measure.Quantity<C>,T extends javax.measure.Quantity<T>> first dimensions. If a conversion factor can not be computed, then it is set to NaN.
    • diagonal

      private static double diagonal(org.opengis.referencing.operation.Matrix m, int j, int n)
      Returns the value on the diagonal of the given matrix, provided that all other non-translation terms are 0.
      Parameters:
      m - the matrix from which to get the scale factor on a row.
      j - the row for which to get the scale factor.
      n - index of the last column.
      Returns:
      the scale factor on the diagonal, or NaN.
    • ensureGeocentricTranslation

      static void ensureGeocentricTranslation(DatumShiftGrid<?,?> grid, javax.measure.Unit<javax.measure.quantity.Length> unit) throws IllegalArgumentException
      Ensures that the grid performs geocentric translations in the given units. This method is invoked by constructor for validation of given arguments.

      This method is defined here in order to ensure a consistent behavior of InterpolatedGeocentricTransform with InterpolatedMolodenskyTransform.

      Parameters:
      grid - the grid to validate.
      unit - the unit of semi-axis length of the source ellipsoid.
      Throws:
      IllegalArgumentException - if the given grid is not valid.
    • setContextParameters

      final void setContextParameters(double semiMajor, double semiMinor, javax.measure.Unit<javax.measure.quantity.Length> unit, org.opengis.referencing.datum.Ellipsoid target)
      Sets the semi-axis length in the context parameters. This is a helper method for constructors in some (not all) subclasses.
      Parameters:
      semiMajor - the semi-major axis length of the source ellipsoid.
      semiMinor - the semi-minor axis length of the source ellipsoid.
      unit - the unit of measurement of source ellipsoid axes.
      target - the target ellipsoid.
    • getParameterValues

      @Debug public org.opengis.parameter.ParameterValueGroup getParameterValues()
      Returns the internal parameter values of this DatumShiftTransform instance (ignoring context). The parameters returned by this method do not necessarily describe the whole datum shift process, because DatumShiftTransform instances are often preceeded and followed by linear conversions. It may be conversions between degrees and radians units, or conversions from geodetic coordinates to grid indices.
      Example: The chain of transforms of an InterpolatedGeocentricTransform is:
      1. Geographic to geocentric conversion
      2. Geocentric interpolation
      3. Geocentric to geographic conversion
      This method returns the parameters for the part in the middle of above example. The content of this part is highly implementation-dependent and used mostly for debugging purposes. The parameters that describe the process as a whole are rather given by getContextualParameters().
      Specified by:
      getParameterValues in interface Parameterized
      Overrides:
      getParameterValues in class AbstractMathTransform
      Returns:
      the internal parameter values for this transform.
      See Also:
    • getContextualParameters

      protected ContextualParameters getContextualParameters()
      Returns the parameters used for creating the complete transformation. Those parameters describe a sequence of normalizethisdenormalize transforms, not including axis swapping. Those parameters are used for formatting Well Known Text (WKT) and error messages.
      Overrides:
      getContextualParameters in class AbstractMathTransform
      Returns:
      the parameter values for the sequence of normalizethisdenormalize transforms.
    • normalizedToGridX

      final double normalizedToGridX(double x)
      Converts the given normalized x coordinate to grid index. "Normalized coordinates" are coordinates in the unit of measurement given by Unit.getSystemUnit(). For angular coordinates, this is radians. For linear coordinates, this is metres.
      Parameters:
      x - the "real world" coordinate (often longitude in radians) of the point for which to get the translation.
      Returns:
      the grid index for the given coordinate. May be out of bounds.
    • normalizedToGridY

      final double normalizedToGridY(double y)
      Converts the given normalized x coordinate to grid index. "Normalized coordinates" are coordinates in the unit of measurement given by Unit.getSystemUnit(). For angular coordinates, this is radians. For linear coordinates, this is metres.
      Parameters:
      y - the "real world" coordinate (often latitude in radians) of the point for which to get the translation.
      Returns:
      the grid index for the given coordinate. May be out of bounds.
    • computeHashCode

      protected int computeHashCode()
      Computes a hash value for this transform. This method is invoked by AbstractMathTransform.hashCode() when first needed.
      Overrides:
      computeHashCode in class AbstractMathTransform
      Returns:
      the hash code value. This value may change between different execution of the Apache SIS library.
    • equals

      public boolean equals(Object object, ComparisonMode mode)
      Compares the specified object with this math transform for equality.
      Specified by:
      equals in interface LenientComparable
      Overrides:
      equals in class AbstractMathTransform
      Parameters:
      object - the object to compare with this transform.
      mode - the strictness level of the comparison. Default to STRICT.
      Returns:
      true if the given object is considered equals to this math transform.
      See Also: