asymmetircal costs in network analysts

686
3
05-12-2020 03:21 PM
asevtsukSevtsuk
New Contributor

I would like to evaluate pedestrian walkshed / shortest paths using both distance and uplsope (elevation gain) as costs. The latter works differently depending on which direction each link is traversed. How to set up a network for asymmetrical costs such as elevation gain?

Tags (1)
0 Kudos
3 Replies
MelindaMorang
Esri Regular Contributor

When you set up your network dataset, you will need to configure a cost attribute that calculates the pedestrian travel time.  Here is some documentation about cost attributes in general: Cost attributes—ArcGIS Pro | Documentation 

When you configure a cost attribute, you have to set up the way it calculates that cost using "evaluators".  When you run a network analysis, the evaluator performs some sort of calculation for each network edge and returns the cost to traverse that edge.  You can configure these evaluators separately for each direction along each network edge.  So, you can have a different calculation going in the direction of digitization and against the direction of digitization.  There are several types of evaluators you can choose from, depending on your needs: Types of evaluators used by a network—ArcGIS Pro | Documentation 

Hopefully that's enough to get you started!

asevtsukSevtsuk
New Contributor

Thanks, i have worked with evaluators in the past and am broadly familiar with the concept of setting them up. It is still not clear, however, how to capture topographic ascent in the evaluator to me. I have computed the Start_Z and End_Z values for all network segments and added two columns that capture ascent from segment start to end (Rise_StoE) and from end to start (Rise_EtoS) which has inverse values of the former.

When assigning evaluators during network setup (or editing them later) I see that i can designate a field as a cost for "FROM TO", and separately for "TO FROM". Is using a "field" evaluator the way to go here? What confuses me is the relationship between the FROM-TO definition and the geometric definition of Start>End on the segments themselves. Do they mean the same thing? If I assign a field value to "FROM TO" movement, does this mean that when a segment is traversed from its geometric Start to End, a given value is assigned as cost? And the other way around if it is "TO FROM"?

On a related note, since elevation gain can be negative, can the evaluator handle negative costs?

thanks again, Andres.

0 Kudos
MelindaMorang
Esri Regular Contributor

Hello Andres.  You have understood the From-To and To-From thing correctly.  From-To refers to the direction of digitization from the start of the line to the end, and To-From is the reverse.

Evaluators treat negative costs as restrictions.  If the cost of an edge is negative, travel on that edge will be prohibited.  Consequently, you don't want to use negative costs to represent reverse slope.

You probably need to treat the slope as a scaling factor.  For example, pick a normal walking speed for flat ground and then scale it up or down based on the slope of the road.  You could pre-calculate this in a single field in your street features, or you could use a Function evaluator to calculate it within the network.  For example, you could have one cost attribute that simply calculates unscaled walk time based on the length of the road and assumed walking speed.  You could have a Descriptor attribute that calculates the scale factor for the road's slope.  You could have another cost attribute, like Scaled Walking Time, that uses a function of your unscaled walking time and the scale factor.

0 Kudos