Geo Leah

In this week's lab we created and compared TINs and DEMs. We used a DEM to do a suitability analysis for a ski run based on elevation, slope, and aspect and displayed it as a 3D image, with a TIN derived from the DEM as the elevation surface. We also explored adjusting the symbology for the points, contours, edges, and surface in a TIN layer. We created a TIN from a set of elevation points and compared its contours to those from a DEM created from the same elevation points using the Spline tool. The DEM contours have a few more small pockets at higher elevations, and the contour lines are smoother than the contour lines from the TIN. The differences are the smallest where the elevation points are closer together and the change in elevation is the steepest. The differences are the greatest at the highest elevations, where there is less elevation change and the points are spaced further apart. Although the contours from the DEM use interpolation to fill in the smooth curves between t...

Continuing with our exploration of data quality, this lab focused on different methods of assessing the relative completeness of two different road networks for Jackson County, Oregon: a street centerline network, and the TIGER 2000 network. We followed a similar methodology to Haklay (2010). The first metric was comparing the total length of the roads in each network. Next, we divided the roads into segments for each 5 km x 5 km grid cell in order to assess which network had the greater length in each cell. My process involved projecting the TIGER network into the same spatial reference system as the Street Centerlines network, clipping both networks to the extent of the grid, and taking the intersection of each network with the grid in order to segment lines based on the grid cells. I then used Summary Statistics to calculate the sum of the road lengths for each grid cell, and Join Field to add these sums to my Grid table for each grid cell. I added a field to my Grid table and used ...

For this week's lab, we tested the positional accuracy of two road networks in Albuquerque, New Mexico by digitizing well-defined points (in this case, road intersections) from imagery and comparing their coordinates to the coordinates of the intersections in these two networks. We followed the protocol of the National Standard for Spatial Data Accuracy to choose 20 test points (shown on the City of Albuquerque road network below, with quadrants to show the distribution of points), and then calculated the appropriate accuracy statistics, which are given below.    Positional Accuracy for the City of Albuquerque road network: Tested 19.07529 feet horizontal accuracy at 95% confidence level. Positional accuracy for StreetMap USA road network: Tested 181.3366 feet horizontal accuracy at 95% confidence level. These data sets were tested according to the NSSDA.

For this lab, we explored different ways of assessing the quality of spatial data, including precision, accuracy, root mean square error (RMSE), and a cumulative distribution function (CDF). I created the map below to show the horizontal precision and accuracy of a dataset of 50 points recorded at the same location by a Garmin GPSMAP 76. Horizontal accuracy is the distance between the true value (the yellow point) and the mean of the recorded values (the red point). The horizontal accuracy is 3.2 m. Horizontal precision is how close a set of recorded results are to each other. A common metric for precision is the 68th percentile: the value that is greater than or equal to 68% of the observations. For this data, the horizontal precision is 6.7 m. On the map, the second buffer has a radius of 6.7 m, and includes 68% of the Garmin measurements.