Lab: Field Survey 1 - Benchmarks, DEMs, and Landscape Modeling

Geospatial Field Methods GEOG 336.002

Lab: Field Survey 1 – Benchmarks, DEMs, and Landscape Modeling

Introduction/Background

This lab is centered around establishing familiarity with various modeling techniques to generate a digital elevation model, or DEM. By using previous classes’ data, we were able to get right into transforming sampled data to forming a process to display the data. Through use of benchmarks and differential leveling techniques, we can accurately model the data into DEMs and digitally formed landscape models for more intensive analysis that would otherwise be impossible to generate out in the field with a pencil and paper.

Problem Statement/Project Need

Field Survey 1 tasked us with answering two objective questions:

How does each generated model compare with the actual landscape?

Which statistical method generates the most accurate landscape model?

Data Collection Procedures and Techniques (the process in which was created for this lab)

The data used in this lab is a series of grid XYZ data points previously collected by former students and provided for our use by Professor Mike Bergervoet for conceptual analysis and to fit given time restraints. The purpose of this lab was to be familiarized with the basics of collecting surveying measurements and techniques below datum with a level and measuring rod based off a known benchmark location. A level and rod are used to establish elevations to locations within eyesight of the level. Using the fire hydrant on the corner of Park Avenue and Roosevelt Avenue as the known benchmark elevation, or BM from the location of the level in the vacant parking lot, the combination of the BM and measurement of the scope’s line from the distance of the level has been noted as the height of instrument, or HI.

Data Collection… The Numbers

Benchmark (BM): fire hydrant

recorded to be: 238.619 meters

Backsight (BS): rod at hydrant (add to BM)

recorded to be: 1.410 m

Foresight (FS): any other rod measurement (depth below datum)

formula of HI - FS

Height of Instrument (HI): height of level in vacant parking lot (subtract from BM)

formula: HI = BM + BS

Thus, 240.029 m = 238.619 m + 1.410 m

The elevation of the level is 240.029 meters above sea level

To find the elevation of the fire hydrant at Phillips Hall (PHH) we would use the formula:

PHH = HI – FS

Thus, 238.339 m = 240.029 m – 1.690 m

The elevation of the fire hydrant at Phillips Hall is 238.399 meters above sea level

Data Processing Procedures and Techniques

Modeling Methods/ArcMap tools used in Field Survey 1 Lab

Inverse Distance Weighting (IDW)- points estimated from averaging values of input data points and their nearby processing cells. Weighted according to shorter/longer distance causing more/less influence on the average values.

Natural Neighbors- input points steal data from nearby data points to create a weight system that proportionately interpolates a value and ‘neighborhood.’

Kriging- estimation of xyz points scattered over surface. The inclusion of the z-values requires additional spatial operations to be best estimated for creating the output surface.

TIN- the triangulation of vertices connected in a network of triangles. The lines connecting each point make up contiguous, nonoverlapping triangular facets with their edges. Nodes can help TINs maintain a higher resolution with their ability to be placed irregularly over a surface. This can create more detail or more variability.

Spline- points estimated through a function to minimize overall surface curvature. This gives a surface that is smooth enough to pass exactly through the input points.

Landscape Modeling Results in ArcMap

How does each generated model compare with the actual landscape?

a) Original DEM provided

b) Inverse Distance Weighting (IDW)- This tool generated a model that closely resembled the landscape and highlighted a key variation in topography. Several areas were slightly rounded but showed the topography more accurately than other models. Data points are splotchy.

c) TIN- in the instance of using a TIN in ArcMap, the TIN model that was generated resembled the differences in topography better than other models in this exercise. Different elevations and slope jaggedness were more accurately represented than all other models.

d) Kriging- Had smooth transitions between elevations, more rounded and smoother than in real life landscape but showed the contrast of different elevation steps more than natural neighbors, spline, and IDW modeling methods. TIN model showed more accurate topology than the Kriging model.

e) Natural Neighbors- Showed only slight differences in topography where it should have noted more differentiation in terms of slope/elevation. However, the natural neighbors model shows more accurate shaping of the tributaries than in the Kriging, IDW, and TIN models.

f) Spline- Had some falsely raised areas in the dark green valley portion of the landscape. Not as well-represented as compared to TIN.

Landscape Modeling Results in ArcScene

How does each generated model compare with the actual landscape? (Image right of each description) Green coloration indicates high elevations and red indicates low elevation. Color scheme continues through each model.

A) Original DEM provided- Valleys and peaks clear and distinct. Riverbed shows varying depths and floodplains. Elevation values are well blended, and contrasts indicate steep slopes.

B) Inverse Distance Weighting (IDW)- The IDW in ArcScene has more pixelated data points than the original DEM. Smaller details of the differences in topography are generalized and not defined clearly. Elevation values are moderately blended into the landscape. Slopes are difficult to evaluate.

C) TIN- The TIN ArcScene provides a blockier pixel format with clear cutoffs regarding differentiation between topography levels. Higher elevations are rounded off and not well defined in relation to actual landscape. Elevation values are not blended whatsoever. Slopes are harsh and difficult to evaluate.

D) Kriging- The Kriging model in ArcScene shows both low and high points more clearly than both the TIN and IDW models. However, it is lacking in some of the higher elevations to the west of the landscape. Elevation features are well blended into landscape. Slope is clearer to distinguish than TIN and IDW.

E) Natural Neighbors- The Natural Neighbors model doesn’t show the valleys as clear as other models. Peaks are more defined than TIN. However, the natural neighbors model does render the peaks and higher elevations more clearly than the kriging model.

F) Spline- The Spline model displays the landscape better than a most of the other models in this exercise. The detail in the upper center tributary is finer tuned than others.

Which statistical method generates the most accurate landscape model?

In the ArcMap portion of this lab, I found that the model that compared best to the original DEM was the spline statistical model. It demonstrated both the highs and lows with a wider range (more accurate to the original data set) than the other models.

Based on this specific dataset, I found that either the natural neighbors and spline statistical methods generate the most accurate landscapes of the five modeling methods in ArcScene.

Each modeling tool has a formula that works better or worse for different types of problems. In the case of the landscape, ArcMap didn’t suit the landscape elevation and DEM information better than in ArcScene.

Sources

Esri. Comparing Interpolation Methods. n.d. <Esri. “ArcGIS Pro.” Select Features by Location-ArcGIS Pro | ArcGIS Desktop, pro.arcgis.com/en/pro-app/tool-reference/3d-analyst/comparing-interpolation-methods.htm.>.

—. TIN in ArcGIS Pro. n.d. <http://pro.arcgis.com/en/pro-app/help/data/tin/tin-in-arcgis-pro.htm>.

Bergervoet, Michael. Other data collection techniques provided for completion of lab exercise. 2018