notes on how to generate overview mosaics

custom_mosaic_tifs_w_gdal.md

How to generate an overview mosaic

NOTE: GDAL > v2.0.0 is needed. Assumes running from a unix-like system with prereqs installed.

1. put all the input files in a directory; name w/ extension .tif; cd to that dir
2. build the vrt gdalbuildvrt -a_srs EPSG:4326 my_overview_file.vrt *tif
3. edit the vrt xml to include your mosaic algorithm example:

       <VRTRasterBand dataType="Byte" band="1" subClass="VRTDerivedRasterBand">
           <PixelFunctionType>maximu</PixelFunctionType>
           <PixelFunctionLanguage>Python</PixelFunctionLanguage>
           <PixelFunctionCode><![CDATA[
   import numpy as np
   def maximu(in_ar, out_ar, xoff, yoff, xsize, ysize, raster_xsize,
                raster_ysize, buf_radius, gt, **kwargs):
        np.round_(np.clip(np.max(in_ar, axis = 0),0,11),
           out = out_ar)
           ]]></PixelFunctionCode>

   • NOTE: the indention on the python code will probably copy and paste wrong and you will need to fix it manually
   • NOTE: the "0" and "11" usied in np.clip should be set to match your output bounds; 15, 31, 63, 127, 255 are other common values.
4. enable python execution export GDAL_VRT_ENABLE_PYTHON=YES
5. make an "overview" at scale 1:1: gdaladdo --config BIGTIFF_OVERVIEW YES -ro my_overview_file.vrt 1
6. (optional) make additional changes to the result using gdal_edit
   • gdal_edit.py -a_srs EPSG:4326 my_overview_file.vrt.ovr

Running for swtx:

[tylar@tylar-pc circe_output]$ pwd
/mnt/ftp-ingest/circe_output

[tylar@tylar-pc circe_output]$ time gdalbuildvrt -a_srs EPSG:4326 swtx_2019-12-05.vrt swtx/*tif
0...10...20...30...40...50...60...70...80...90...100 - done.
real	0m0.672s
user	0m0.418s
sys	0m0.117s

[tylar@tylar-pc circe_output]$ atom swtx_2019-12-05.vrt 
[tylar@tylar-pc circe_output]$ atom swtx_2019-12-05.vrt 
[tylar@tylar-pc circe_output]$ export GDAL_VRT_ENABLE_PYTHON=YES
[tylar@tylar-pc circe_output]$ gdaladdo -ro swtx_2019-12-05.vrt 1
Warning: Overview with subsampling factor of 1 requested. This will copy the full resolution dataset in the overview!
0...10

Count() method for showing coverage available here