delineate() and plotDelineations() issues with addtopo and elev plot arguements

[dsilas]

Hello, thanks for this impressive package. I have been fiddling around with delineate() to identify reflectors and horizons. It seems that with a GPR object that has coordinates added, delineations made when the plot has addTopo = T or elev = T with coords on, cannot plot correctly with plotDelineations() if you replot the same object but with addTopo = F. The same is true visa versa. It would be convenient if the function did the math for the user so delineations made to a GPR object with coords would appear correctly regardless of the plotting arguments you decide to plot the GPR object with. I think the same thing happens with hyperbolas too.

library(RGPR)
gpr_topocorrect = gainAGC(interpPosFromXYZ(frenkeLine00,xyz = frenkeTopo00))

sample_reflector <- list(
  x= c(23.32410909,24.51200732,24.98716661,27.04619020,28.86763414,31.08504416,34.72793205,37.18292172,42.56806033),
  y=c(88.02512734,91.08566153,98.22690798,105.36815442,105.36815442,123.73135957,112.50940087,135.97349634,126.79189376))
delineated_gpr = delineate(gpr_topocorrect,values = sample_reflector)
# zrel column seems rounded?
print(delineations(delineated_gpr)[1])
plot(delineated_gpr)
lines(sample_reflector,lwd=2)
points(sample_reflector,pch=4,lwd=2)
# plots as expected
plotDelineations(delineated_gpr,lwd="3",col="green")
#delineation not appear, because the delineation was identified without topo/elevation on the plot, although the gpr object had coords already
plot(delineated_gpr,addTopo = T,elev=F)
plotDelineations(delineated_gpr,lwd="3",col="green")



# Adding deliniations with topo on but elev off.
plot(gpr_topocorrect,addTopo=T,elev=F)
sample_reflector_topo = list(
  x=c(23.38931038,24.74575641,28.52021492,30.58437192,34.29985451,37.95636120,43.38214530),
  y=c(2.467659703,3.023396342,3.393887435,4.181181007,3.718067141,5.200031511,4.505360713)
)
delineated_gpr_topo = delineate(gpr_topocorrect,values = sample_reflector_topo,addTopo=T,elev=F)
# zrel column seems rounded?
print(delineations(delineated_gpr_topo)[1])
lines(sample_reflector_topo,lwd=2)
points(sample_reflector_topo,pch=4,lwd=2)
# plots as expected
plotDelineations(delineated_gpr_topo,lwd="3",col="green")
# will not plot correctly with elev = T
plot(delineated_gpr_topo,addTopo=T,elev=T)
plotDelineations(delineated_gpr_topo,lwd="3",col="green")
# also will not plot correctly
plot(delineated_gpr_topo,addTopo=F)
plotDelineations(delineated_gpr_topo,lwd="3",col="green")

# adding delineations with the topo on and elevations turned on
plot(gpr_topocorrect,addTopo=T,elev=T)
sample_reflector_topo_elev = list(
  x=sample_reflector_topo$x,
  y=-(sample_reflector_topo$y) + 347
)
delineated_gpr_topo_elev = delineate(gpr_topocorrect,values = sample_reflector_topo_elev,addTopo=T,elev=T)
# zrel column seems rounded?
print(delineations(delineated_gpr_topo_elev)[1])
lines(sample_reflector_topo_elev,lwd=2)
points(sample_reflector_topo_elev,pch=4,lwd=2)
# plots as expected
plotDelineations(delineated_gpr_topo_elev,lwd="3",col="green")
# will not plot correctly with elev = F
plot(delineated_gpr_topo_elev,addTopo=T,elev=F)
plotDelineations(delineated_gpr_topo_elev,lwd="3",col="green")
# also will not plot correctly
plot(delineated_gpr_topo,addTopo=F)
plotDelineations(delineated_gpr_topo_elev,lwd="3",col="green")

[dsilas]

I would also add- the zrel values in delineations() seem oddly rounded to the nearest tenth,, and on a GPR object with coords added, trying to recreate the original delineations you hand delineated using e.g. delineations(gpr)[[1]][,c(4,5)] (this is the xrel and zrel) columns, if you replot those points on the plot they don't always precisely match your original delineation. This means the original delineations are lost if you wanted to use them in another thing outside RGPR to calculate slopes of dipping reflectors or something. (though i guess you could calculate the 3d slope using the x y z columns shown when you call delineations(gpr)[[n]].

is there another better way to get the 2d spatial x /y positions of delineations from the GPR object using the i and j values in GPR@delineations? ( are these traces and samples?)

[emanuelhuber]

Hello @dsilas
Thank you for rising this issue. The handling of delineation is tricky, because from the developer side you never know if the user will process the data after delineating some structures. So the trick is to fix the delineation to the pixel. That is why the delineation is transformed into the pixel coordinates (i, j). Now you can re-interpolate the coordinates of your data, apply a velocity model, the delineation stay fixed with the original pixel. That is the reason you observed a loss of precision.

Do you understand my logic? Do you have any suggestion from a user perspective?

I will (try) fix the problem with addTopo = FALSE in the next day.

[emanuelhuber]

Hello @dsilas

I checked RGPR and tried to fix the bug you discovered. It is unfortunately quite complex to fix that, because there are a lot of nested functions.
I am working on a new version of RGPR and I will implement correctly the plot of delineation (maybe I will add an argument directly to the plot function to plot the delineations and to ensure a correct plot). Thank you for you patience.

[dsilas]

Dear @emanuelhuber Thank you for the response and I sincerely apologize for my extremely delayed response. You deserve better given the effort you are doing. I think I do understand your explanation of the difficulty of handling delineations and the rationale for the choice you made. I will download the latest master and see what the latest developments are.

From a user perspective, one useful thing is if the reflectors people click on/delineate can later be accurately "restored" afterwards as real world xyz coordinates, for example sent to a text file using the exportDelineations() function, so that desired quantitative spatial analyses can be performed on these delineations. I think the way that the reflectors are slightly different when I tried to replot the ones I previously delineated (due to the pixel stuff you mentioned here #92 (comment) ) made me uncertain whether what I was able to export would be spatially identical to what I had delineated, or not.