Alyssa Corbett March 4, 2008 Assignment 5: Data Quality Assessment Project: Assessing the feasibility of installing an accessible and environmentally-friendly parking garage in the Harvard Square area Harvard Square is located in south-central Cambridge, Massachusetts (Fig 1), and is a hub for scholars from all over the world, college students, restaurant connoisseurs, and eccentric shoppers This diversity has led to increased pressure for affordable parking on top of the efficient public transportation system that is already in place This project will attempt to locate an parcel of land that could be engineered into a multi-use area: underground parking garage with a public urban space above (similar to the Boston Commons) This project will have to take into consideration infrastructure, land use, existing parcel patterns, and areas with a high frequency of restaurants, shops, etc For this conceptual stage of the project, it is necessary to obtain approximate locations of land use zones and commercial addresses (±30m) However, it will be preferable to have accurate data for infrastructure and parcel location (±5m) as to correctly determine a feasible parcel for this project Fig 1: Harvard Square 1:3,750 Orthophoto (2005), Contrast stretched, ½ meter resolution Road Centerline Data: Three road centerline datasets were used in the development of this project: the Massachusetts Department of Transportation (Executive Office of Transportation, or EOT) dataset, provided by MassGIS, the Census TIGER road data, and Streetmap USA The MassGIS EOT data layer (Fig 2) is the most accurate, determined by comparing the street arcs to the Cambridge orthophoto and parcel data The TIGER (Fig 3) and Streetmap USA datasets, being that they are national datasets, are less accurate; they seem to have less data points, which creates jointed (rather than smooth) lines, and there are numerous occasions where streets seemingly run through buildings Of particular interest is the fact that the TIGER data (similar to the Streetmap USA data, which has been excluded from the figures to avoid clutter) has road centerlines through the Charles River, which is a testament to the inaccuracy of the data Fig 2: Harvard Square and EOT road centerline data Fig 3: Harvard Square and TIGER road centerline data If we narrow our focus to center of Harvard Square, the difference in data quality between the EOT (yellow) and TIGER (green) road centerlines is most obvious In Figure you notice that the TIGER lines within the blue box cut across the building to join adjacent roads, while the EOT lines correctly dead end In the red box you notice that the TIGER lines create a simple polygon around an area with a more complex pattern which the EOT lines recognize Lastly, in the orange box you notice the positional inaccuracy of the TIGER lines compared to the EOT lines, which, according to the ArcToolbox measuring tool, are off by approximately 20 meters Fig 4: Closer view of Harvard Square – comparisons between TIGER (green) and EOT (yellow) road centerlines Hydrography Data: A similar case occurs between the state data (MassGIS 1:25,000 layer, based on USGS Digital Line Graphs, scanned mylar separates, and digitized hydrographic features from topographic quadrangle maps) and national data (Census TIGER hydrography) Figure shows the discrepancy between the two data sets depiction of the Charles River Fig 5: The Charles River, as indicated by MassGIS (light blue) and TIGER (royal blue) layers Positional Accuracy of Data Layers: Road Centerlines and Hydrography While there is no information in the metadata that claims a specific positional accuracy of the road centerline and hydrography layers, I was able to determine that, on average, the TIGER and Streetmap USA data is accurate to ±15 meters (compared to the orthophoto) while the EOT road centerline data is accurate to ±1 meter Additionally, the TIGER hydrography layer, which strays anywhere from 10-70 meters from the actual coastline, can be averaged to a positional accuracy of ±40 meters, while the MassGIS hydrography layer is accurate to ±3 meters One must keep in mind that the orthophoto may not be completely accurate, so in order to verify the exact positional accuracy of these layers, one would have to use a hand-held GPS unit on site Bars and Restaurants The positional accuracy for bars and restaurants is extremely poor The addresses were collected using Reference USA and then the addresses were located based on the TIGER road centerline data (because the more accurate EOT layer is not geocoded) The addresses were also offset 10 meters from road centerlines The inaccuracy of the restaurant data is due to both Reference USA and TIGER data For example, Tommy Doyle’s, represented by the pink dot in Figure 6, is located at 96 Winthrop St., according to it’s website (http://www.tommydoyles.com/harvard/), while Reference USA has the restaurant located at 76 Winthrop St This inaccuracy has landed the restaurant across the street from its actual location, approximately 65 meters away, represented by the pink arrow Alternatively, Grendel’s Den Restaurant and Bar, represented by the blue dot, has the correct listing of 89 Winthrop St., but the dot is located on the wrong side of the road, according to the orthophoto This is most likely due to the fact that the address location data in the TIGER road centerlines did not correctly clarify between the left and right side of the road The correct position of Grendel’s Den is indicated by the blue arrow Fig 6: Location of bars and restaurants in a section of Harvard Square (Winthrop St highlighted in white for reference) For the purposes of the project, it is not necessary to have the exact location of each restaurant, but rather to have a general idea of where the cluster of these attractions are focused (and thus where there will be a high demand for parking) However, it is good to take note of the positional inaccuracies for inclusion in a project report Parcels The parcel positional accuracy in the project region is accurate to ±5 meters, when compared to the orthophoto (Fig 7) However, one is unable to judge the exactly positional accuracy without a GPS unit at the parcel site It is convenient for the project that the data is accurate, as developing a plan for engineering a new parking garage and public open space requires a clear outline of parcel location Fig 7: Parcels in Harvard Square as compared to the orthophoto Land Use It is important to know the current uses for sections of parcels in order to determine a good place for a parking garage The goal is to find an area that is already used for public urban access (so the original land use can be restored after the completion of the project), but is also nearby commercial areas that are of interest to commuters The land use data (Fig 8), collected by the members of the Resource Mapping Project at the University of Massachusetts, Amherst, was originally digitized data from their 1971 interpretation, and have since been updating the information The data is estimated to be accurate to ±10 meters, using the ArcToolbox measuring tool Fig Landuse in Harvard Square (green = urban public, yellow = participation recreation, gray = commercial, dark red = ½ acre residential, red = multi-family residential, purple = transportation, dark gray = cemetery, blue = water body) Protected Open Space Lastly, it is always important for potential building projects to keep in mind the possibility that a potential site may be restricted due to a protected status The protected (and recreational) open space data layer contains boundaries of conservation lands and outdoor recreation facilities (Fig 9) Areas in red indicate zones that are legally protected by way of deed or other official document, while areas in yellow indicate zones that have limited access, and require a majority municipal vote for a change in status Although this layer was last updated in February of 2008, it is possible that protection status has changed, so it is important to always re-evaluate the accuracy of this data Fig 9: Protected Open Spaces in Harvard Square Currency: The data used in this project is all fairly recent The EOT road centerlines were from 2002, which is a bit dated, but there did not seem to be any changes, according to the orthophoto, which is from 2005 It would take a GPS at the location site to determine changes in the past three years The TIGER data was from the 2000 Census TIGER but was updated in 2003 The hydrography data from MassGIS was updated in 2005, while the less accurate TIGER data is from 2003, similar to the road centerlines The parcel data used in the project was updated in 2007, the land use data in 2002, and the protected spaces data in February of 2008 Despite the fact that the land use data is the oldest, it is unlikely to have changed in an area that has been similarly developed for many decades Attribute Accuracy: Much of the data that I found is sufficient for my project For example, while the restaurant data is poor, it is unnecessary to have data accurate to more than ±50 meters, because the project is looking for clusters of restaurants rather than exact locations However, the project accuracy could be improved if the restaurant address data was checked against actual locations, and if the data was geocoded to the EOT road centerlines, instead of the TIGER data The rest of the data (parcels, land use, protected open spaces) is all sufficient for this project – although it is important to keep looking for updated information Additionally, each layer was complete, except for the restaurant data which needed address re-matching in order to get a 100% match between USA Reference information and the TIGER road centerline data ... similar to the road centerlines The parcel data used in the project was updated in 2007, the land use data in 2002, and the protected spaces data in February of 2008 Despite the fact that the land use... from their 1971 interpretation, and have since been updating the information The data is estimated to be accurate to ±10 meters, using the ArcToolbox measuring tool Fig Landuse in Harvard Square. .. 2: Harvard Square and EOT road centerline data Fig 3: Harvard Square and TIGER road centerline data If we narrow our focus to center of Harvard Square, the difference in data quality between the