Eugen Taso Assignment #5 1. Provide one-paragraph description of the project you are using as a benchmark to assess the data and what positional accuracy it will require (or what is good enough - think about how far off the position could be and still work for the project needs) For this assignment I am looked at the City of Cambridge. By analyzing orthophotos, building layers, hydro data, railroads and various street layers, I want to look at bike paths and sidewalks and see where they are positioned relative to their real position, in order to see whether a person who wants to bike or walk can rely on the data available in Mass GIS to accurately draw a path. I would also like to see the position of public buildings and historic landmarks in the City of Cambridge and make sure that I am able to identify them and position them in the right places on the map, without roads, bike paths, rivers or lakes, etc. passing through them. 2. Briefly discuss the three different road centerline data sets in terms of their positional relation to each other (look at how far apart they are at different points using the measure tool in ArcGIS, and if there is consistency in the differences. Include some graphic examples to illustrate your points. Which data set would be best for your project? I actually loaded four different road layers to my map. I used the EOT Roads layer, the Census TIGER layer, the USA Streets layer and the Cambridge GIS street layer, for comparison. I also loaded the orthophoto to the map, for visual accuracy. After overlaying all the roads layers, I noticed the EOT Roads layer is the most accurate (based on the orthophoto) in presenting the road position on the map. It’s very close to the City Of Cambridge road centerlines layer, but more complete, representing all the major roads and side streets, while the Cambridge layer only represents the main streets. Using the measuring tool, the EOT roads and the City of Cambridge roads are spot on the orthophoto. For my project (estimating the best bike paths and sidewalks), it would make most sense to use these layers. The TIGER data and the USA streets data on the other hand is off by 25 meters and 34 meters, respectively in a specific area around my house (Copley Street), which makes is less reliable. The USA streets layer goes right through residential buildings for most of the roads, as does the TIGER data. I zoomed into my neighborhood to find that both the TIGER and the USA streets go through my house. I included some of the images below, for reference. 1 Figure 1 - Street layers in Cambridge Figure 2 - Difference and distance between street layers in Cambridge 3. Do the same as above for the two hydrography layers. Done! Just kidding As with the roads layer, I loaded the TIGER data from the Census, the City of Cambridge hydro data as well as the orthophoto for visual comparison. What I noticed is that the TIGER data is more “choppy”, and there are approximations for where the body of water goes. The lines are cut at right angles, rather than following the natural river, lake or bond contours. For example, at Fresh Pond, the TIGER layer represents the reservoir smaller on the Cambridge 2 side than it actually is, putting the roads and implicitly the sidewalks and the bike paths right in the river. In Southwest Cambridge, the river is on Soldier’s Field Road in the TIGER data, 64m from the actual river bank, so a rather large difference, making it unreliable for this project. For this reason, the City of Cambridge data is more reliable for my project. After all, I would not want a biker or a pedestrian to walk right into a lake, or miss it by tens of meters. Please see below for examples of Fresh Pond and the Charles River in Southwest Cambridge. Figure 3 - Hydro data difference (Fresh Pond, Cambridge) Figure 4 – Hydro difference in layers (Charles River, Cambridge) 3 4. Can you provide a quantitative assessment of positional accuracy for each of your data layers (e.g., +/- 20 feet)? Why or why not? For roads, the TIGER data appears to be in a +/- 20m range, while the USA street data can venture even further, to +/- 30m (please see map above in question 2) For hydro data, the range is higher, but TIGER appears to be off the real location by a +/- 15 to 70m (please see map above in question 3) For buildings (optional layer from City of Cambridge), it appears that they coded each building right on the money, so it is within 1m of the right location (to small to show on a screenshot). However, in some instances, there is a +/- 5 meter distance from the outline to the building, which could be a problem for walking/bike path design in GIS mapping. For landmarks, they are coded very close to the actual site, as in the Memorial Hall example below. The landmark includes the land parcel as well as the building, which makes it even more useful to use, as it lets a potential biker/walker of the landmark and the adjacent area associated with it. The area seems to have been superimposed on the orthophoto. Railroads appear to be spot on as well, with little or no margin of error from the City data compared to the orthophoto. Figure 5 – Building outlines over Orthophoto match (Cambridge, Harvard Square) 4 Figure 6 – Building outline difference over orthophoto (Memorial Drive, Cambridge) Figure 7 – Historic landmarks outlines over orthophoto (Memorial Hall, Harvard University) 5 Figure 8 – Railway outline over orthophoto (North Cambridge) 5. Give a qualitative assessment of positional accuracy of each of the four optional layers relative to the other layers (e.g., do streets run through buildings? are schools in the correct location along a road?). TIGER and USA Street road outlines do indeed run through buildings as they are +/- 20 and 30m off the mark, respectively. An example is presented below: Figure 9 – TIGER and USA Street layers going through buildings (West Cambridge) A similar thing happens for the hydro layers. Unfortunately, this makes the TIGER data and the USA street data less reliable for this project. Having roads pass through buildings, not being able to identify the correct path for walking or biking could significantly hinder the success of a person trying to map the best route for a walk, or the perfect bike ride. The same is not true for the optional layers. Buildings are in the right place. However a +/5m deviation in some cases could be problematic, given the walking patterns of pedestrians. For biking, that’s rather insignificant. 6 Railroads are identified in the Cambridge data in the right places, as are landmarks, For the project, that is extremely useful, as a rail crossing over an actual rail is much better than a rail crossing that is off by a significant distance (both for bikers and pedestrians). However, there is no information on the particular railroad track (T, commuter rail, etc.), which for this project is not important, but for another project might make a difference. Figure 10 – Railway crossing on Sherman Street (North Cambridge) 6. Are these optional layers appropriate for your project in terms of their positional accuracy? Yes. Railroads appear to be in the right places, as do buildings and landmarks. For walking or biking, it is imperative to have accurate measurements, and not deviate more than a couple of meters at most from the real location. With buildings being within 5m and railroads being right where they are supposed to be, this problem is taken care of. 7. Completeness: Is each data set complete? (Does it cover the area question, are all relevant features present, and is the attribute information complete for all features?) Roads: The EOT roads attribute is accurate, but only at locating the roads. The information present in the attribute table is not sufficient for more in-depth analysis. The addresses are not present, and if they are, they’re inaccurate. However, the Cambridge main roads data contains the right information, from addresses to location, so it’s a better layer for use. The TIGER census data is more accurate and complete in terms of addresses, but does a bad job at location Hydro: Cambridge layer is complete and accurate at locating the right bodies of water in the right locations. TIGER layer is not great at locating or offering more information. 7 Buildings layer is not complete. While location is accurate, information about public buildings is not available. For example, the armory/National Guard building in North Cambridge is not labeled or available in the attribute table. Figure 11 – Attribute table for EOT roads layer Figure 12 – Attribute table for Cambridge Roads layer (Cambridge) 8 Figure 13 – Building layer completeness (North Cambridge) 8. Currency: Are the data up to date? How do you know the answer to this? You can find out by looking at the Metadata in Arc Catalog. For this particular case, I looked at my layers to get the following results: TIGER data: 2000 (census data) EOT Roads: 2007 USA Streets: 2000, 2002 Cambridge Roads: 2003 data, updated in 2004 Cambridge Hydro: 2003 data, updated in 2004 TIGER Hydro: 2000 (census data) Cambridge Buildings: 2003 data, updated in 2004 Cambridge Historical Landmarks: N/A Cambridge Railroads: 2003 data, updated in 2004 Based on this information, the data is fairly up to date. Buildings may have been built since 2003/2004, but roads, lakes/rivers, and historical landmarks are unlikely to change over a 5 year period. This makes the data good to use for a project looking into walking and biking, as well as identifying current public buildings. 9. Attribute accuracy: provide a qualitative assessment of attribute accuracy for critical attribute items (e.g., land use codes, street names and address ranges, school names, etc). How adequate is the attribute information for your project needs? The attribute information is very relevant for a walkability/bike path project. This is why the EOT data is not as useful as it could be. While the positioning is great, the lack of proper addresses and the lack of street names at times make it less usable. Fortunately, the Cambridge 9 roads layer (which I added to my map) solves that problem, at least for the major roads. Similarly, the Cambridge buildings data would not be very useful, as it fails to identify the major public buildings in the area, which would be useful for the public buildings project. Railroads are very accurately depicted in terms of location. However, there is no information on the particular railroad track (T, commuter rail, etc.), which for this project is not important, but for another project might make a difference. It seems there is always a trade-off between good data (TIGER, which has exact street addresses) and location/drawing (EOT data) which has the perfect location, but no real info on addresses. However, a combination of the two (perhaps with geocoding) may be useful in mitigating that problem (like the example in class today, 3/6). 10