Hi J.O. Williams
thank you very much for your ideas and detailed explanation. I will try them out and let you know how it works.
Eva
Gesendet: Freitag, 06. November 2020 um 06:52 Uhr
Von: "J.O.Williams" <[email protected]>
An: [email protected]
Betreff: Re: [Qgis-user] How to get the thickness of a geol. layer from elevation isoline features
Von: "J.O.Williams" <[email protected]>
An: [email protected]
Betreff: Re: [Qgis-user] How to get the thickness of a geol. layer from elevation isoline features
Eva
I only saw the end of the third reply. I added my thoughts to it.
I hope it helps.
J.O. Williams
>
> I have the following task:
>
> I need to create a map with contour lines showing the thickness of
> one geological layer/horizon (actually not the true thickness, but
> the vertical extent is required, so if you drill down at one point,
> what is the distance between the upper and lower edge of the horizon).
>
> The only data I have are two line features (shape files) of isolines
> with elevation values (in m NHN) of each the upper edge and lower
> edge of the horizon.
This implies you only have contour (isolines) of the top and bottom of
the formation. Turn the isolines into points with a new attributes of
x, y and of elevation or use a 3-d shapefile point If the isolines x and
y values are in Lat-Long degrees, turn them into meters.
If the isolines are steeply dipping, consider using Universal kriging.
> My approach would be something like this, but I can’t figure out how
> exactly to do it:
> • Create a raster from each line feature (did that, but only the
> lines are rastered, so would need a polygon to get a plane? How do
> that? How can I include the elevation values of the lines into the
> polygon?)
A problem with contour lines, is that there are lots of values along the
elevation contours. You might need to remove every other value. The
search routines of kriging follow the contour lines and don't find
values between each contour line elevation.
>From the x,y and elevation values of the contour just run a inverse
distance (ID) or Ordinary Kriging (OK) of the contour lines. I would
suggest OK. You didn't mention if you needed the quality of estimation
that OK can also give you, so just running kriging run using a linear
variogram with a sill of 1.0 and a range of about 1/3 of the distance
across you area of of interest.
The only reason to calculate a variogram is to calculate the estimation
variance. If you don't need the estimation variance, a variogram that
doesn't have a sill and keeps increasing will work just fine. Linear
variograms are cheap to calculate. If you don't have a linear
variogram, the first 2/3 of the spherical variogram is linear.
Don't use nugget value. Nugget values are usually some sort of
measurement error, such as the Geologist picking the wrong value of the
top and bottom elevation of the formation.
However, with the holes you have in the data and the faults you will
have to constrain the input data. Do not let the data search cross the
faults lines. When the kriging takes place, you want any trends in
elevation to continue until up or down until you hit a fault. Likewise
with other areas between the faults, they too will have to be kriged
separately so the data points outside the area don't influence the
calculations. Lots of clipping will be necessary.
When you are done you should have two raster surfaces that have
elevation calculations that show each of the sub faulted areas. There
will be ugly elevation changes between the areas. Hide the ugliness
with very broad lines!
> • Interpolate the elevation values in between the isolines to receive
> a plane of elevation values (in my opinion would need a polygon to
> get a plane, but for interpolation input Points are needed? And which
> interpolation method? Also, some of the isolines are offset by
> faults, so need sharp edges in these areas, how to integrate that in
> the interpolation?)
The contour lines for each of these subareas will have to have an area
attribute added. The raster points will have to have a attribute value
that identifies each area. When kriging just use the rasters from each
area at a time and the elevation contour lines for each sub area.
Each subarea will have to be done separately and checked against the
original contour lines.
> • Then calculate the vertical distance at each point in the area
> between the two planes to get the thickness (how, which tool does
> that?)
> • Create a raster showing the thickness throughout the area in colour
> range
> • Add contour lines showing the thickness of the horizon
Be sure you enter known contour line values during contouring. It is
very easy to get negative elevations. Negative elevations are VERY
embarrassing on the final The one attached example has a zero elevation
line in the north-west edge. You can draw a zero line, if you need, by
drawing one contour line at an elevation of say 0.1 meters but labeling
the line as 1 meter.
If raster areas go negative, set the raster area to zero.
Thick contour lines are your friend and cover up possible problems with
the data (and your presentation).
>
> Maybe there is even a better and probably much simpler way to do
> this. However, I am only a beginner in QGIS, python and statistics.
> So any help and easy to understand tips will be very much
> appreciated. I am using QGIS 3.10.11 by the way.
>
> Best regards,
>
> Eva
>
I only saw the end of the third reply. I added my thoughts to it.
I hope it helps.
J.O. Williams
>
> I have the following task:
>
> I need to create a map with contour lines showing the thickness of
> one geological layer/horizon (actually not the true thickness, but
> the vertical extent is required, so if you drill down at one point,
> what is the distance between the upper and lower edge of the horizon).
>
> The only data I have are two line features (shape files) of isolines
> with elevation values (in m NHN) of each the upper edge and lower
> edge of the horizon.
This implies you only have contour (isolines) of the top and bottom of
the formation. Turn the isolines into points with a new attributes of
x, y and of elevation or use a 3-d shapefile point If the isolines x and
y values are in Lat-Long degrees, turn them into meters.
If the isolines are steeply dipping, consider using Universal kriging.
> My approach would be something like this, but I can’t figure out how
> exactly to do it:
> • Create a raster from each line feature (did that, but only the
> lines are rastered, so would need a polygon to get a plane? How do
> that? How can I include the elevation values of the lines into the
> polygon?)
A problem with contour lines, is that there are lots of values along the
elevation contours. You might need to remove every other value. The
search routines of kriging follow the contour lines and don't find
values between each contour line elevation.
>From the x,y and elevation values of the contour just run a inverse
distance (ID) or Ordinary Kriging (OK) of the contour lines. I would
suggest OK. You didn't mention if you needed the quality of estimation
that OK can also give you, so just running kriging run using a linear
variogram with a sill of 1.0 and a range of about 1/3 of the distance
across you area of of interest.
The only reason to calculate a variogram is to calculate the estimation
variance. If you don't need the estimation variance, a variogram that
doesn't have a sill and keeps increasing will work just fine. Linear
variograms are cheap to calculate. If you don't have a linear
variogram, the first 2/3 of the spherical variogram is linear.
Don't use nugget value. Nugget values are usually some sort of
measurement error, such as the Geologist picking the wrong value of the
top and bottom elevation of the formation.
However, with the holes you have in the data and the faults you will
have to constrain the input data. Do not let the data search cross the
faults lines. When the kriging takes place, you want any trends in
elevation to continue until up or down until you hit a fault. Likewise
with other areas between the faults, they too will have to be kriged
separately so the data points outside the area don't influence the
calculations. Lots of clipping will be necessary.
When you are done you should have two raster surfaces that have
elevation calculations that show each of the sub faulted areas. There
will be ugly elevation changes between the areas. Hide the ugliness
with very broad lines!
> • Interpolate the elevation values in between the isolines to receive
> a plane of elevation values (in my opinion would need a polygon to
> get a plane, but for interpolation input Points are needed? And which
> interpolation method? Also, some of the isolines are offset by
> faults, so need sharp edges in these areas, how to integrate that in
> the interpolation?)
The contour lines for each of these subareas will have to have an area
attribute added. The raster points will have to have a attribute value
that identifies each area. When kriging just use the rasters from each
area at a time and the elevation contour lines for each sub area.
Each subarea will have to be done separately and checked against the
original contour lines.
> • Then calculate the vertical distance at each point in the area
> between the two planes to get the thickness (how, which tool does
> that?)
> • Create a raster showing the thickness throughout the area in colour
> range
> • Add contour lines showing the thickness of the horizon
Be sure you enter known contour line values during contouring. It is
very easy to get negative elevations. Negative elevations are VERY
embarrassing on the final The one attached example has a zero elevation
line in the north-west edge. You can draw a zero line, if you need, by
drawing one contour line at an elevation of say 0.1 meters but labeling
the line as 1 meter.
If raster areas go negative, set the raster area to zero.
Thick contour lines are your friend and cover up possible problems with
the data (and your presentation).
>
> Maybe there is even a better and probably much simpler way to do
> this. However, I am only a beginner in QGIS, python and statistics.
> So any help and easy to understand tips will be very much
> appreciated. I am using QGIS 3.10.11 by the way.
>
> Best regards,
>
> Eva
>
_______________________________________________ Qgis-user mailing list [email protected] List info: https://lists.osgeo.org/mailman/listinfo/qgis-user Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
