Project

General

Profile

TileSettings - Tiling functionality for high resolution datasets

See https://github.com/KNMI/adaguc-datasets/blob/master/adaguc.tiled.xml for example configuration

Mandatory settings:
  • tilewidthpx="720
  • tileheightpx="720"
  • tilecellsizex="0.0125"
  • tilecellsizey="0.0125"
  • tileprojection="+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs"
  • minlevel="1"
  • maxlevel="4"
  • debug="false"
  • optimizeextent="true" <-- Increases performance of tilecreation with --createtiles command. It will detect which region from the base grid is needed for creating a tile, the result is that only a fraction of the big base netcdf file is read into memory. Optimizeextent will also detect if the region has any data in it, if not a tile will not be created.
  • maxtilesinimage="24" <!-- limit on how many tiles are allowed to appear in a single getmap request or are used during a getfeatureinfo request.
  • tilepath="/data/adaguc-autowms/tiles"
Optional settings:
  • left="50" right="180" bottom="-90" top="90" <-- Note these are optional and when left out they will be calculated automatically.
  • tilemode="avg_rgba" or "rgba" <!-- only useful when using rgba netcdf's, dont use if you have standard byte/short/int/float/double data in your variable
  • threads="1" <!-- Optional, default is "1". Experimental setting to render a GetMap request with multiple threads
# To recreate all tiles, do:
rm -rf /data/adaguc-autowm/tiles/*
./adagucserver --updatedb --config /config/tilesconfig.xml --recreate 
./adagucserver --createtiles --config /config/tilesconfig.xml 
./adagucserver --updatedb --config /config/tilesconfig.xml

Some datasets in CLIPC are in such a high resolution that they cannot be stored in a single file. For example the flooding indicator from CLIPC/PIK is at 25 meter resolution along the coast in Europe. The data is organized in 6000 different NetCDF files, each covering its own area in 600x600 grid cells, which need to be “stitched” together. This can be compared with composing a large panoramic picture from multiple pictures taken along the horizon. The challenge here is to provide interactive visualization at several cartographic projections and zoom levels.

To achieve this, ADAGUC has been extended to allow building image pyramids at several zoom levels on coarser resolutions (see Figure ). For CLIPC 7 different pyramid levels are used for the flooding indicator, the top level consists of three tiles, the bottom level of 6000 tiles. When zooming to a specific distortion of the map, the level with the best matching resolution is used to render the requested image. That means that ADAGUC projects and combines several tiles into a single image suitable for WMS. This type of tiled datasets in ADAGUC behave like any other WMS, e.g. interactive zooming and panning at several cartographic map projections remain possible.For CLIPC the flooding indicator in the cartographic projection lambert equal area is remapped on request to the Mercator projection used in the CLIPC portal.

Image pyramids for tiling

Adaguc-server-TileSettings-Pyramid.png View - Image pyramids for tiling (153 KB) Maarten Plieger, 06/05/2017 07:40 PM