Water/Blend Masking using shapefile

[Kelvin Richardson]

hi guys, me again. Just have one question about adjoining tiles. See my screenshot, any ideas what could be causing this?

settings
[Destination]
DestDir = "F:\dev\working\resamplestaging"
DestBaseFileName = "Area_Lp1_SnapOff_S034232731_S034330088_E172500987_E173022804"
DestFileType = BGL
UseSourceDimensions = 1
CompressionQuality = 85
LOD=Auto

[Source]
Type = MultiSource
NumberOfSources = 3

[Source1]
Type = BMP
Layer = Imagery
SourceDir = "F:\dev\working\resamplestaging"
SourceFile = "Area_Lp1_SnapOff_S034232731_S034330088_E172500987_E173022804.bmp"
Lon = 172.8360700
Lat = -34.3909192
xDim = 1.43006627136388e-005
yDim = 1.07736852737187e-005
PixelIsPoint = 0
SamplingMethod = Point
Channel_BlendMask = 2.0
Channel_LandWaterMask = 3.0

[Source2]
Type = TIFF
Layer = None
SourceDir = "F:\dev\working\resamplestaging"
SourceFile = "_0000_blendmask.tif"
Lon = 172.8360700
Lat = -34.3909192
xDim = 1.07736852737187e-005
yDim = 1.07736852737187e-005
PixelIsPoint = 0
SamplingMethod = Gaussian

[Source3]
Type = TIFF
Layer = None
SourceDir = "F:\dev\working\resamplestaging"
SourceFile = "_0001_watermask.tif"
Lon = 172.8360700
Lat = -34.3909192
xDim = 1.07736852737187e-005
yDim = 1.07736852737187e-005
PixelIsPoint = 0
SamplingMethod = Gaussian

 

[arno]

Looks like the georeferencing of the images is not perfect, else the coast would align in both. How did you calculate your lat, lon, xDim and yDim values for them?

 

[Kelvin Richardson]

I entered the lon and lat in the resample file from the fsearthtiles areamask inf file. I wonder if not rounding to 3 decimal places caused this. I'll start the process from scratch and see what stage I can pinpoint the issue.

 

[arno]

Rounding the values would give offsets normally, more digits is better for the accuracy

Never use FSEarthTiles myself, so can't help with that bit. I usually work with GeoTIFF files.

 

[Kelvin Richardson]

Hm OK thanks. I actually convert the bmp aerial image to geotiff and import the coordinates from the fsearthtiles inf file so my mask files align correctly as I need that info to split the raster from shape file.

 

[Kelvin Richardson]

ok, I have sorted all my issues out, it seems that using Proj4 interpretation for the projection causes an offset in the below, by a small margin but big enough to cause an issue with a gap between my tiles.
CellXdimensionDeg = 1.43051147460938E-05
CellYdimensionDeg = 1.07288360595703E-05

So the process is as follows and I have found it must be in this order to avoid messing up the files.

gdal_rasterize -burn 255 -burn 255 -burn 255 -init 0 -ot Byte -ts 14334 14851 -te 172.631020545959 -34.3909192085266 172.836084365845 -34.5502531528473 -l land land.shp temp.tif

gdalwarp -t_srs EPSG:4326 -ts 14335 14851 -co INTERLEAVE=PIXEL -r cubic -of GTiff temp.tif wgs84_gdalwarp.tif

gdal_translate -a_srs EPSG:4326 -of GTiff -co "INTERLEAVE=PIXEL" temp.tif wgs84.tif

listgeo -d wgs84_gdalwarp.tif > wgs84.gtf

Edit wgs84.tif in Photoshop...

gdal_translate -a_srs EPSG:4326 -of GTiff -co "INTERLEAVE=PIXEL" wgs84_gdalwarp_0000_blendmask.tif blendmask.tif
gdal_translate -a_srs EPSG:4326 -of GTiff -co "INTERLEAVE=PIXEL" wgs84_gdalwarp_0000_blendmask.tif watermask.tif

geotifcp -g wgs84.gtf wgs84_gdalwarp_0000_blendmask.tif blendmask_gtf.tif
geotifcp -g wgs84.gtf wgs84_gdalwarp_0001_watermask.tif watermask_gtf.tif

 

[Kelvin Richardson]

Some latest screenshots with seamless tiles. So far only 2 tiles but the outlook is good!