The following method has been added in the GridCoverage class in the development branch:
However this method is inefficient if invoked for more than one point (which is a typical scenario). It may trig many readings of small areas, construct many times the same CoordinateOperation, etc. Furthermore it does not allow users to control the interpolatipn method.
An alternative approach could be to define a separated Evaluator class with methods for setting the interpolation types among other. All coordinates to evaluate could be specified in advance, allowing Evaluator to perform only one larger read operation instead than many small operations.
We may keep GridCoverage.evaluate as a convenience method when there is really only one point to evaluate. But we would need to clarify at least the policy regarding interpolation method.