nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Could I create a raster depicting cells that are visible from a road?

JesseKolar

I have a 30m DEM raster and a polyline shapefile for a road network. I would like to produce a 30m raster of I/0 values showing whether or not each cell is visible from a road up to 2000m. I'd like to use a height of 2m above the road to represent the height of a commercial vehicle.

Is this possible with the viewshed tool? The examples I have found work best for smaller point shapefiles, like whether a tower is visible, and do not help for a large, polyline feature.

Thank you for any ideas!

Finde mehr Beiträge mit dem Tag

Accepted answers

All comments

DarrenWiens2

Viewshed (and related tools) can take a polyline as input - there will just a be a trade-off between speed/storage and resolution depending on how dense you want the observation points along the line. For example, it may not be reasonable to calculate the viewshed every metre along a road, but every 100 m may work fine.

DanPatterson_Retired

Viewshed limits are 16 points per test... so you have to batch and choose wisely

DarrenWiens2

I can't test right now, but I only see that limitation for Observer Points, not Viewshed.

DanPatterson_Retired

That would be the limit I refer to... If the interest is in just to see if the road is visible, it just needs to be elevated, it is usually the observer points that is the limitation, not what is being looked at. Convert the road to raster, a con could be used ... ie, if road, then elevation + 2m, otherwise elevation, then there is no issue, unless the landscape needs to be viewed from multiple locations

DarrenWiens2

The way the problem is described, they want Viewshed in this case, so there shouldn't be an issue.

JesseKolar

I am looking at a large area approximately 80x80 miles with an extensive road network (see attached, roads are clipped by a study area buffer). So I cannot use the observer option because of the 32 point limit on observers.

I am currently running the Visibility tool with the DEM as the input raster, and roads as the input polyline. Analysis type: FREQUENCY, observer parameters-- Surface offset: 2, Outer radius: 1600.
I thought the Surface offset would account for the 2m vehicle height.

However, the tool has been running for 4+ hours, and I do not think it is working. With an outer radius of 1600, it seems like the calculation would just look at the center of each 30m pixel and calculate visibility from roads up to 1600m. However, I believe it is counting a density, which is taking too much resources.

Is there any way to calculate I's/0/s so that after confirming a pixel is visible from a road, the calculation could move on?

JesseKolar

And thank you very much for trying to help. If it clarifies my question at all, I'm looking at the influence of roads on deer movements. I hypothesize that deer will not be as affected by roads if the roads are out of sight. I have 600,000 (and counting) deer locations, so I was hoping to calculate the entire raster instead of calculating visibility for each deer location. This will also help to create predictive maps to highlight areas that are currently not visible from roads.

Thanks again!

DanPatterson_Retired

Interesting. Where are the sighting locations with respect to the roads? In my area, deer runs have preferential road crossing locations which are quite historic in nature and associated with forest coverage meeting the road edge. Sadly, adjacent fields which would provide a sight line for both animals and vehicles are rarely traversed. Similarily the interior network of runs is also well established and deviations from existing ones are usually only seen when a new predator enters the area.

RebeccaStrauch__GISP

And not being a biologist myself, but is noise a factor for deer? I am not sure of eyesight vs hearing for deer, but based on (personal) general observations, it seems like their ears perk up pretty fast due to noise..even when head down and eating. So, for roads, do you have to consider traffic into the equation? ....again, I am not a biologist (but I work with many...but totally personal comments)

JesseKolar

Thank you both for your feedback. We will be considering habitat type, road surface type (as a proxy for traffic levels), development densities, ruggedness, slope, aspect, and a whole suite of predictor variables that have been shown to influence deer distributions in the literature. The idea to incorporate viewshed is from anecdotal observations of deer using miniature "wilderness" areas among heavily developed portions of the study area. These areas may be very close to roads, but on the opposite side of a ridge, so on the ground, they feel much more remote than areas at the same distance to roads without any topography to break the viewshed.

There is a publication by Robert Montgomery et al. 2012. Importance of visibility when evaluating animal responses to roads. Wildlife Biology 18:393-405. I'm trying to replicate their methods, but without adding the height of forest cover, since I'm working in fairly open habitats.

JesseKolar

I got access granted to a much more powerful computer that can be left to run for a few days. I am going to try this using the visibility tool with 30m DEM raster data and a roads shapefile broken into 100m line segments (i.e. the tool will examine vertices at 100m). If I do not update my progress within a few weeks, feel free to contact me.

search-color.png

XuguangWang

Yes, the Visibility tool is the right tool for this kind of application. However, it seems there is a minor issue in your workflow – Since the roads are the observers, to model the vehicle height on the road, please set the observer offset to 2m. There is no need to set the surface offset to model your application.

The viewshed calculation is a global operation, which means that the tool needs to check more than the immediate neighboring cells of the observer in order to determine the visibility. So it is a computationally intensive process. When a polyline is used as input observers, each vertex of the polyline is treated as one observer. Given the large number of polylines in your road network, it is understandable that the visibility tool will take some time to finish.

Regarding the output of the Visibility tool (with Frequency option), each cell on the output raster stores the count of observers to whom it is visible. To learn more about how it works please visit
http://desktop.arcgis.com/en/arcmap/latest/tools/spatial-analyst-toolbox/using-viewshed-and-observer-points-for-visibility.htm

BethanyArndt

I am hoping to do a similar analysis with grizzly movement data. Wondering how this worked out for you and whether there are any publications as a result of your work?

JesseKolar

I ended up running this analysis on a university supercomputer, and it took

_{2 months to run for a subsample of my study area (}500 km^2 subsample)

with a rural road density. The road densities were higher in some areas

because they followed legal section lines every 1-mile. We deemed the

analysis unfeasible for our study. The best citation for the methods is

Montgomery et al. 2012 (http://www.bioone.org/doi/abs/10.2981/11-123)

Importance of visibility when evaluating animal response to roads, but

they're a bit light on the analytical methods to try to replicate the GIS

steps.

Good luck!

Jesse

ChrisDonohue__GISP

I did an analysis similar to what the original poster stated for a municipality about ten years ago with ArcGIS 9. We were evaluating what areas of a California City that could be seen from major roads as part of their Zoning update, as concerns had been raised about unusual and outlandish construction affecting the historic Gold Rush character of the old town section, which had quite a tourist draw. To get a handle of this, the City first wanted to know what areas were even possible to be spotted from the major roads.

To accomplish this, the major roads were broken down into points every several feet. Via Modelbuilder, a Viewshed was run for each point, then all the resultant several thousand viewsheds were combined to come up with a count for each cell. Thresholds were chosen to differentiate counts in each visible cell so as to divide off the results with no or just incidental results from those area that had major exposure. If I remember correctly, runs took about 6 hours to do (the project area was about a dozen miles across, so far smaller than what others are considering).

It was a fun project to do and I ended up putting up a poster on it in the Map Gallery.

Chris Donohue, GISP

Nimm an der Diskussion teil

Angemeldete Mitglieder können Beiträge verfassen, Updates folgen und mehr. Neu hier? Registriere ein kostenloses Konto.

Schnellzugriffe

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.