topic Re: vertical exaggeration in 3D Questions

vertical exaggeration

SuzanneFaber — Fri, 18 Sep 2020 09:29:21 GMT

Hi,

I have a set of 3d point data (hydraulic conductivity according to performed drillings). I want to use this dataset for geostatistical interpolation. To be able to start with an investigation of my dataset I coloured the values with a gradient. I see my data as below. The datapoints are too close together in the vertical direction to be able to study them. When I zoom in by scrolling the mouse, the dots disappear from view (see bottom picture). With the sliding bar on the right I can't seem to do anything for the view.
I thought I use a vertical exaggeration, but I don't see how (there is no option for it under the appearance tab).

Does anyone have an idea how to solve this?

Thanks, Suzanne

Re: vertical exaggeration

DanPatterson — Fri, 18 Sep 2020 09:53:18 GMT

perhaps in a scene

Extrude features to 3D symbology—ArcGIS Pro | Documentation

Re: vertical exaggeration

SuzanneFaber — Fri, 18 Sep 2020 12:13:45 GMT

Hi Dan,
Thank you for the suggestion. I uploaded the data in Scene and I don't see any difference in visualisation options.
By the way I am using ArcGIS pro 2.6.

Re: vertical exaggeration

DanPatterson — Fri, 18 Sep 2020 18:57:46 GMT

So that is after extruding the features as in the link?

Re: vertical exaggeration

Anonymous User — Fri, 18 Sep 2020 19:29:39 GMT

Hi Suzanne- I love these use cases, and I have an example like this right in front of me.

I think what you may have happening here is simply a rendering issue with immediate subsurface 2D symbols. Even though you've removed the basemap in your scene view, the 2D point symbols are conflicting with the elevation surface and irregularly rendering. You can check this by navigating under the surface-

In the Contents pane, choose the Ground elevation surface.
On the Appearance tab, check the Navigate Underground checkbox.

Even though it is possible to vertically exaggerate these readings, I would advise against it before producing an analysis product with 3DEBK as it's expect real-world inputs and the results will differ with exaggeration. You're also able to set a vertical exaggeration for Voxels.

I'll tell you how to do it anyway

Exaggeration in Pro is a per-layer setting. Under Layer Properties->Elevation, you'll find a vertical exaggeration parameter.

If you intend to publish or share this layer outside of this Pro Project,you'll need to us the GP tool "Layer 3D to Feature Class" to "burn" the geometry exaggeration changes into the feature.

So- cool about the workflow afterwards. Definitely check out the 3D Empirical Bayesian Kriging tool, which can take numeric values and interpolate data. There's a recent tool with Pro 2.6 where you're able to export 3DEBK results to netCDF to be read as Voxels in Pro 2.6 Advanced.

I have a similar example here with soil density borehole values under a large piece of infrastructure:

Really interested in your results!

Phil

Re: vertical exaggeration

SuzanneFaber — Tue, 22 Sep 2020 09:58:58 GMT

Hi Dan,
silly I didn't see you posted a link there..
Meanwhile I managed with Philip's remark, simply by unchecking the ground (world elevation) box under the contents tab.
Thanks again for our help!
Suzanne

Re: vertical exaggeration

SuzanneFaber — Tue, 22 Sep 2020 10:23:00 GMT

Hey Philip,

thanks a lot for you elaborate reply! Actually, all I needed to do was uncheck this ground elevation surface box.. I cannot find the navigate underground checkbox you mention, but anyway problem is solved (see pic below).

I tried to use the 3D EBK tool before. However, it tells me I do not have enough data to perform the analysis (see 2nd pic). If you would have an idea for a work-around around this I would be very greatful.

For now, I have started my analysis in 2D: I have subdivided the data in layers and analyse each separately. The layer that I am looking at is a semi-permeable aquitard: it consists of a number of clay lenses overlaying each other. (I used indicator kriging in 2d, simply saying clay=1, sand=0) As it is a sedimentary basin, the clay lenses have been deposited in the x-y plane, which would justify my 2D analysis. However, the clay layers might be ever so slightly tilted, and by looking in 2D I might be missing continuities that will actually have a great influence on the groundwater flow. SO, if I could use 3D EBK, where one can separate the domain in smaller overlapping domains and adjust the elevation inflation factor, that would be great. 🙂 Having not enough data to do the analysis thoroughly doesn't bother me to much.., as I just want to use the analysis for finding some sort of range, a few 'this might be possible' scenarios which I can use for a few different conceptual models for the flow study.

Curious what you may think about this!

Suzanne

Re: vertical exaggeration

EricKrause — Tue, 22 Sep 2020 14:19:14 GMT

Hi Suzanne,

I can try to help you with the problems you're having in EBK3D. First, a quick clarification for something earlier. EBK3D uses the geometry of the features for the calculations, so it is not affected by visualization settings in the scene like vertical exaggeration.

As for why you are seeing the error about not having enough data, I can see in your picture that you have plenty of points (only 10 valid values are required). Only two things come to mind that could cause this error.

Accidentally leaving on a selection or definition query
Providing an input field or elevation field where almost all values are null (both fields must have non-null values for a point to be included in the model). Note that the elevation field is used to specify the z-coordinates of the features, and it defaults to Shape.Z if the points are z-enabled.

Please let me know if neither of these resolves the problem. A screenshot of the attribute table of the points and a screenshot of the first page of the Geostatistical Wizard with parameters filled out might help me identify the problem.

Thanks,

Eric

Re: vertical exaggeration

Anonymous User — Tue, 22 Sep 2020 14:51:50 GMT

Hi Suzanne-

Good advice from Eric above.

You can see those clay lenses really well as isosurfaces in the Voxel experience. Sounds like my scenario is identical to yours. I treated the numeric values for the data the same way (SPTN values).

Re: vertical exaggeration

SuzanneFaber — Wed, 23 Sep 2020 14:04:11 GMT

Dear Eric and Phil,

Again, thanks a lot for both of your inputs.

For the most bottom layers of my data I had fewer than 10 data points. I have now left these layers out and it works 🙂 I looked at the predication map and used the empirical transformation and an exponential semivariogram model type (I didn't see a big decrease in the error when I used K-Bessel.).
See pictures below (at 30, 50 and 59m-bg resp). I also included two screen shots of the EBK analysis and the histogram of my data. Looking at the prediction vs measurement map, the EBK results also don't seem amazing (to my unexperienced eye). I am wondering to what extend the locations of my data points have to do with that. As you can see from the top view, they are not very evenly distributed.

Prediction maps:

EBK 3D analysis:

Histogram of my data:

Of course now I have some new questions..

Question 1. I would love to create an isosurface in voxel space as Phil suggested, I saw the esri video on it and it looks really amazing. But how? According to the esri website I should find the option under the add data tab, but I only find the multidimensional raster layer.

I thought I start by trying out this multidimensional raster layer. I first created a netCDF file by exporting the 3D EBK results as point features, but I get an error when uploading. This might have to do with the fact that I don't know what to fill in under the 'feature to netcdf' menu.. Phil, I would be very curious to learn how you create isosurfaces in voxel experience.

Question 2: The EBK supplies us with an optimal fit to the data. If I would want to create different fields (to have best- and worst case scenarios) I understand I need to look at quantile surfaces, in which I could choose to study what my outcomes may be if my measurement values are either over- or under estimated. Condition for using quantile surface is that a multivariate normal distribution can be assumed. I am not sure what that means in my case, after all, I have used the empirical transformation to be able to work around the fact that my data is not normally distributed.

Very greatful for your support!

Suzanne

Re: vertical exaggeration

EricKrause — Wed, 23 Sep 2020 15:17:40 GMT

Hi again,

I have a lot to say but unfortunately not a lot of time right now. I'm going to try to cover the most important things, sorry for quickly jumping between topics.

The GA Layer 3D To NetCDF tool is used to make the source file for the Voxel layer. This tool and the voxel layer itself are both new in ArcGIS Pro 2.6, which has only been available for a few months. If you aren't seeing the option to add a voxel layer in Add Data in a scene view, you probably don't have the most recent version.

If you have ArcGIS Pro 2.5, you can use the GA Layer 3D To Multidimensional Raster tool to export the EBK3D layers directly to a multidimensional CRF raster.

I see in your image of cross-validation that many of the points have identical (or very close to identical) values. These are the horizontal lines of red points in the graph. Repeated values can be problematic for the Empirical transformation, especially with large gaps between the repeated values (this can be seen in the histogram). The empirical transformation is essentially trying to fit a smooth curve to the histogram, then uses this curve as a reference to the normal distribution. However, if the histogram isn't smooth, the curve won't fit the histogram well, and it will likely give strange results in the gaps between the peaks in the histogram.

Regarding multivariate normality for the quantile output, this is very hard to explain without giving a 2 hour statistics lecture, so I'll try to keep it short. Kriging of all types is designed to directly predict the mean and standard error of the true value at a new location. However, you need a full distribution to estimate quantiles, and the mean and the standard error are not enough to fully reconstruct a distribution. I could show you many different datasets that all have the same mean and standard error, but they would all have very different quantile values. So, to calculate quantiles, you have to make an assumption about the predictive distribution, which is almost always a normality assumption. The reason for this is that if your input data are normally distributed (or transformed to be normal), then the kriging predictions will also be normally distributed. This is why you check for normality in your input data first so that you don't have to worry about it later.

In practice, you do the best that you can to transform the data to be closer to normal, but that is going to be difficult to do for a histogram with repeated values and gaps. If it were me, I would probably try different EBK3D parameters like subset size and overlap factor and hope that you land on a particular subsetting where the model is most stable. I would also experiment with not using a transformation at all.

To judge how well a particular model is working, I would focus on cross-validation, specifically the RMSE and the Inside 90/95 Percent Interval. The RMSE is approximately equal to the average difference between the true and predicted value, so it's a quick test to see if a model is too inaccurate to be viable. The two Inside 90/95 statistics build confidence intervals using the normality assumption. If the values aren't too far off from 90 and 95, it may be safe to assume normality to estimate quantiles.

-Eric

Re: vertical exaggeration

Anonymous User — Wed, 23 Sep 2020 15:54:13 GMT

The only piece to add to this guidance is directions for activating the Voxel Exploration pane and creating Isosurfaces: Create a voxel layer isosurface—ArcGIS Pro | Documentation

Re: vertical exaggeration

SuzanneFaber — Thu, 24 Sep 2020 10:44:28 GMT

Hi again Eric and Phil,

regarding the voxel experience: indeed I needed to download the latest version of 2.6! I created the NetCDF file using the GA Layer 3D to NetCDF tool. I then added the voxel layer. However, when I then want to create the isosurface, nothing happens.. I think that it is trying to create the layer (at least I can 'pause the drawing' see pic) but doesnt get there.. I get no error messages or so. Any ideas what goes wrong?

Regarding Eric's comment about the cross validation: Indeed the data that I put in is very course. Most data points have the 'value': 'medium grained sand'. Then there are a few locations with course sand and some with clay. The bulk of course grained sand I have given a hydraulic conductivity of 15m/d, but of course in reality these values might locally vary between say 5 and 25m/d. I will play around with the other EBK3D parameters as you suggest and also add a measurement error column in my input file and see what that does for my results.

Cheers, Suzanne

Re: vertical exaggeration

EricKrause — Thu, 24 Sep 2020 13:55:01 GMT

Hi Suzanne,

The isosurface is a 3D contour, so you have to specify which value you want to contour. My guess is that the default value is producing a very small (or even empty) isosurface. You can change the isosurface value in the Voxel Exploration pane.

I recently wrote a LearnGIS lesson that performs EBK3D and ends with voxel visualization and exploration. You might find the entire thing useful, but here's the section specifically about creating isosurfaces:

Interpolate 3D oxygen measurements in Monterey Bay | Learn ArcGIS

-Eric

Re: vertical exaggeration

SuzanneFaber — Thu, 24 Sep 2020 14:17:34 GMT

Hi Eric,

I am not sure that is the problem, if I try for example to change the symbology, I get this message:

I wonder if that might also be the reason I cannot find the voxel exploration pane at all...

Suzanne

Re: vertical exaggeration

SuzanneFaber — Tue, 06 Oct 2020 08:43:08 GMT

Dear Eric (and Phil),

I have tried again to get the isosurface, but I do not manage. I follow the steps exactly as indicated in the lesson you sent me.

I created the .nc file, but when I want to add the multi-dimensional raster layer I get the following warning:

Despite this warning, it does create this menu for me

but it doesnt show me anything and tells me

I also tried right clicking on the Surfaces box and say 'create isosurface' but it doesnt do anything.

Any idea what might be going wrong?

Thanks...!

Kind regards,

Suzanne

Re: vertical exaggeration

Anonymous User — Tue, 06 Oct 2020 15:04:57 GMT

Hi Suzanne- This error occurs when you're attempting to add a Voxel Layer to a Global Scene. Voxels currently only work in Local Scenes.

Re: vertical exaggeration

SuzanneFaber — Wed, 07 Oct 2020 14:40:40 GMT

Hi Phil,

I am working in Local Scene..

Suzanne

Re: vertical exaggeration

Andrew--Johnson — Wed, 07 Oct 2020 16:56:28 GMT

Hi Suzanne,

Can you click on the notification? It will bring up the notification pane which will provide more details. The most common cause is that the voxel layer coordinate system must match the scene's coordinate system. If the scene already has data and a coordinate system set then you will need to change it to match the voxel layers coordinate system. The other reasons why this might appear is if you are using OpenGL as the rendering engine but once again the notification pane will have more details. Also check out the Voxel FAQ Frequently asked questions—ArcGIS Pro | Documentation if you have a chance.

Hope that resolves it!

thanks!

Andrew

Re: vertical exaggeration

EricKrause — Wed, 07 Oct 2020 17:04:25 GMT

I suspect the problem is that the coordinate system (both horizonal and vertical) of the scene have to be the same as the Voxel layer, and the coordinate systems of the Voxel layer are driven by the EBK3D layer.

Please try adding a brand new Local Scene and immediately try to add the netCDF file as a voxel layer before doing anything else with the scene (specifically, do not add any data to it). My bet is that it will render correctly. When you add data to a new map/scene, it will set the coordinate systems to match that data, so the voxel layer and the scene should automatically be put into the same coordinate systems.

If this ends up being the issue, you can change the coordinate systems of your other scene to match the coordinate systems of the EBK3D layer, and then you should be able to add the voxel layer.

[Edit: I see Andrew beat me to it]