| 144 | | Vectorizing all the table would have been way too long and the huge amount of geometry would have benn impossible to load in OpenJUMP (and most GIS software). |
| 145 | | |
| 146 | | You will notice that a small area in the upper left corner of the vectorized tile is missing. This is an area with nodata value pixels. Any analysis function in PostGIS WKT Raster takes the nodata value into account. This means that those areas will be omitted from the operation. For example in the upcoming intersection operation, if a buffer completely overlaps a nodata value area of one tile, st_intersects(rast, geom) will return false for this tile/buffer couple and st_intersection(rast, geom) will return an "EMPTY GEOMETRY". If the buffer partially overlap a nodata value area and intersect other parts of the tile, st_intersects(rast, geom) will return true for this tile/buffer couple and st_intersection(rast, geom) will only returns parts of the tile with values, omiting parts with nodata values. |
| | 144 | Vectorizing all the table would have been way too long and the huge amount of geometry would have been impossible to load in OpenJUMP (and most GIS software). |
| | 145 | |
| | 146 | First notice: some pixels are grouped together into complex polygons. ST_DumpAsPolygons() not only blindly dump each pixels, it actually vectorize the raster grouping areas with shared values together. In a raster with an low range of value this would result in significantly less polygones than there is pixels. Higher number of different value (like continuous floating point value rasters) would result in many, many little polygons, most of them square corresponding to only one pixel. A future ST_Reclass() unction should help reducing the number of value in a raster and give more analysis flexibility. For now you can always reclassify your raster BEFORE loading them into PostGIS WKT Raster. |
| | 147 | |
| | 148 | This query includes the value associated each group of pixel. You can display those values by right-clicking on the OpenJUMP layer, selecting "Change Styles...", selecting the "Labels" tab, "Enabling labelling" and selecting the "val" Label attribute. |
| | 149 | |
| | 150 | You will also notice that a small area in the upper left corner of the vectorized tile is missing. This is an area with nodata value pixels. Any analysis function in PostGIS WKT Raster takes the nodata value into account. This means that those areas will be omitted from the operation. For example in the upcoming intersection operation, if a buffer completely overlaps a nodata value area of one tile, st_intersects(rast, geom) will return false for this tile/buffer couple and st_intersection(rast, geom) will return an "EMPTY GEOMETRY". If the buffer partially overlap a nodata value area and intersect other parts of the tile, st_intersects(rast, geom) will return true for this tile/buffer couple and st_intersection(rast, geom) will only returns parts of the tile with values, omiting parts with nodata values. |
