Produce all pyramid levels from start and end zoom.
Produce all pyramid levels from start and end zoom.
The first entry in the result stream is the tuple of rdd
and startZoom
.
The RDDs of pyramid levels have a recursive dependency on their base RDD.
Because RDDs are lazy take care when consuming this stream.
Choose to either persist the base layer or trigger jobs
in order to maximize the caching provided by the Spark BlockManager.
RDD key type (ex: SpatialKey)
RDD value type (ex: Tile or MultibandTile)
Metadata associated with the RDD[(K,V)]
the base layer to be resampled
the scheme used to generate next pyramid level
the pyramid or zoom level of base layer
the pyramid or zoom level to stop pyramid process
the options for the pyramid process
[up]
Resample base layer to generate the next level "up" in the pyramid.
Resample base layer to generate the next level "up" in the pyramid.
Builds the pyramid level with a cell size twice that of the input level---the "next level up" in the pyramid. Each tile is resampled individually, without sampling pixels from neighboring tiles to speed up the process. The algorithm proceeds by reducing the input tiles by half using a resampling method over 2x2 pixel neighborhoods. We support all AggregateResampleMethods as well as NearestNeighbor and Bilinear resampling. Nearest neighbor resampling is, strictly speaking, non-deterministic in this setting, but is included to support categorical layers (e.g., NLCD). Given this method, input tile layers are obviously expected to comprise tiles with even pixel dimensions.
RDD key type (ex: SpatialKey)
RDD value type (ex: Tile or MultibandTile)
Metadata associated with the RDD[(K,V)]
the base layer to be resampled
the scheme used to generate next pyramid level
the pyramid or zoom level of base layer
the options for the pyramid process