Methodology:
What led us into the realm of 3D GIS was the decision to do a project based on shade analysis. Westbank and Peterson Group has proposed to the city of Vancouver to build at Comox and Broughton St in the Downtown West End. The building’s plan includes 184 rental units and a community facility. It is designed to be a 22 story building. This is a residential neighborhood that is free of sky rises. This new building is more than twice as tall as all the surrounding buildings. We wanted to aid the neighboring people by conducting a shade analysis on the building to determine the amount of sunlight the current residences may lose. This analysis will hopefully help the community take a stand on the approval of this building.
Based off of the methodology from the Philadelphia model we were able to build a 3D model of the city block that contains the building site.
We started with the building footprints that we acquired from the city of Vancouver’s data catalogue. We created a new layer solely of the footprints we are conducting analysis on. The footprints lack a Z value (for height) so we used google earth to obtain the correct building heights. We assumed these numbers were relatively accurate based off of a test on the Shangri-La. From here we edited the attribute table of the footprints by adding a new field and entering in the heights (making sure to subtract the base elevation of 28m from each).
Moving over to ArcScene now; we added our building footprints layer in, as well as a DEM of the area obtained from the g drive. To make the buildings 3D you have to extrude them based on the height field. Then make their base heights floating on a custom surface, the DEM. This yields a 3D image of the buildings and the terrain.
Our next step is to use the python workflow that was successfully used in the Philidaelphia Model, Pyephem. It’s website states that, “Pyephem is an algorithm that provides scientific-grade astronomical computations” (Rhodes, 2010). Simply stated, it is a python script that will determine where the sun is located for your scene at any time and day you specify. It is critical for our analysis because it allows us to move on and create skyline barriers and ultimately project volumetric shadows behind our buildings.
Now
to complete the shadow analysis we
need to convert the vector building models into a multipatch. This is
because the multipatch
model creates vertices making it a solid entity
rather than a projection like entity. The conversion tool
that we used can be found here: 3D Analyst toolbox/Conversion/Layer 3D
to Feature Class in ArcScene. What led us into the realm of 3D GIS was the decision to do a project based on shade analysis. Westbank and Peterson Group has proposed to the city of Vancouver to build at Comox and Broughton St in the Downtown West End. The building’s plan includes 184 rental units and a community facility. It is designed to be a 22 story building. This is a residential neighborhood that is free of sky rises. This new building is more than twice as tall as all the surrounding buildings. We wanted to aid the neighboring people by conducting a shade analysis on the building to determine the amount of sunlight the current residences may lose. This analysis will hopefully help the community take a stand on the approval of this building.
Based off of the methodology from the Philadelphia model we were able to build a 3D model of the city block that contains the building site.
We started with the building footprints that we acquired from the city of Vancouver’s data catalogue. We created a new layer solely of the footprints we are conducting analysis on. The footprints lack a Z value (for height) so we used google earth to obtain the correct building heights. We assumed these numbers were relatively accurate based off of a test on the Shangri-La. From here we edited the attribute table of the footprints by adding a new field and entering in the heights (making sure to subtract the base elevation of 28m from each).
Moving over to ArcScene now; we added our building footprints layer in, as well as a DEM of the area obtained from the g drive. To make the buildings 3D you have to extrude them based on the height field. Then make their base heights floating on a custom surface, the DEM. This yields a 3D image of the buildings and the terrain.
Our next step is to use the python workflow that was successfully used in the Philidaelphia Model, Pyephem. It’s website states that, “Pyephem is an algorithm that provides scientific-grade astronomical computations” (Rhodes, 2010). Simply stated, it is a python script that will determine where the sun is located for your scene at any time and day you specify. It is critical for our analysis because it allows us to move on and create skyline barriers and ultimately project volumetric shadows behind our buildings.
We
were faced with some problems regarding the PyEphem work flow (these
are outlined below) and thus we had to be creative in order to finalize
our analysis on these buildings.
We did however learn a lot about how to build a 3D model of Vancouver in ArcScene based primarily off of the 25m DEM. A lot of our time was spent playing around with the new program and learning how to make a visually pleasing scene. We created a fly through video to better display the third dimension online.
We did however learn a lot about how to build a 3D model of Vancouver in ArcScene based primarily off of the 25m DEM. A lot of our time was spent playing around with the new program and learning how to make a visually pleasing scene. We created a fly through video to better display the third dimension online.
Moving on, to complete our shadow
analysis we decided to use Google Sketch Up.
This program was new to us as well so
it was a learning curve. We were able to import our multipatch
buildings after converting them to the google recognized “collada” type
in ArcScene (Conversion toolbox/To Collada/Multipatch to Collada). From
here we added a geo-location so that the program knows the latitude of
the buildings and thus can correctly project the shadows. The process
of positioning the individual buildings in the correct spots was the
hardest part because of the complex move and orbit tools.
This program was new to us as well so
it was a learning curve. We were able to import our multipatch
buildings after converting them to the google recognized “collada” type
in ArcScene (Conversion toolbox/To Collada/Multipatch to Collada). From
here we added a geo-location so that the program knows the latitude of
the buildings and thus can correctly project the shadows. The process
of positioning the individual buildings in the correct spots was the
hardest part because of the complex move and orbit tools.
Once our building block was visually
acceptable we created an animation of the shadows on August 1st from
eight am to three pm. This animation will be used in conjunction with
the following solar radiation analysis.
Back in ArcMap we used the
Solar Radiation tools to make a map of the watts per square meter of
the terrain. This involved clipping the DEM raster to a polygon of
Vancouver to reduce the time it would take to complete the computation
and then conducting: Spatial Analyst Tools/Solar
Radiation/Area Solar Radiation. The input is the clipped raster DEM and
the latitude will be set by the DEM automatically. We made it within
the day of August 1st to match the Google sketch up shadow analysis.
This concluded the process we went through with our Vancouver Model. Please see the results below. For a more conclusive step by step procedure please see this.
Limitations and Error:
In the end we could not complete the volumetric shadow analyses for our Vancouver model. We ran into a number of problems at the start, just building our 3D projections was difficult, once they were built we again had trouble turning them into a multipatch. The biggest problem that caused hours of painstaking trial and error to eventually heed no results, was PyEphem. This Python based code was the engine that was to drive the sun for our model. The only problem was we could not get it to function. We loaded the necessary files, however the instruction included a lot of programming that flew over our heads. No errors occurred when we used the CreateSunPoints tool. However when we looked in the attribute table it was blank. Because the CreateSunPoints tool is an interface for PyEphem we deducted that the PyEphem model was not running in ArcScene.
There are several pages of instruction on how to get PyEphem to run properly. However, no one in our team has experience with the Python programming language, and the instructions require knowledge of this. Our lack of programming experience, made it impossible to figure out the proper technique needed to run the download in ArcScene. The security measures on the school computers had an effect on PyEphem. In order to download the full version we needed to be an administrator. We employed Jose are lab technician to do this for us but even with the full download we could not find a way to make it run in ArcScene. Later on we learned that we could run the code from a virtual environment thus bypassing the need to be an administrator. It is my belief that if we kept trying, and learned our Python we would be able to get the file to run. This would then allow for us to complete our desired volumetric shadow analyses.
Results:
The video below shows a depiction of shadows cast on August 1, 2010, an arbitrary day chosen to display results for our model. However, Google Sketch Up provides a tool in which manipulation of time and date is quite accessible. As seen in the video clip the shadows created by this building on this date primarily falls on the buildings due west (j) (i) (h) and (g) and north (b) and (c) of the proposed building (a). Buildings (f) and (e) escape the shadow for this chosen time period. Building (d) is significant to our results as the shadow looks as if it will cast upon it. However, due to building (d)’s substantial height relative to the adjacent building and compared to height of the proposed building no shadow is cast on building (d). These results represent a small unit of potential results as further analysis could show difference between daily and hourly shadows cast. Furthermore since our research was limited to the Sketch Up programming volumetric shadow analysis was unobtainable, which would have aided us in a furture research of how the proposed building could affect solar radiation.
Since our shadow analysis was inconclusive we sought to comprehend if this site was a potential ‘hotspot’ for solar radiation using the solar radiation tool on a DEM of Vancouver. The tool was used for the same time period August 1, 2010 and takes into account slope and aspect. This shows that buildings (j) (i) and (h) which were mentioned previously as being effected by a shadow on this date are in an area with a lower watts/m2 output. The cell which represents the highest solar radiation output in the figure below looks to narrowly escape a shadow cast by the proposed building (a). However, a shadow by building (d) is cast in this area. Buildings (b) (c) (g) (f) and (e) are in an area which has a fairly high rate of potential solar radiation. Both (b) (g) and (c) fall within the shadow, yet (f) and (e) do not. One error related to these finding is that at such a close scale the resolution of the DEM does not precisely display data on solar radiation.
Finally, an understanding of the visual reality of this building can be seen by the model represented in Arc Scene. To get a better idea of the scale, orientation, and direction we have provided you with a fly by video clip of the proposed region: Here.