There is a plenty of publications about this topic but no universal solution. Dependent on the data you have and methodology you use different approaches can be taken into account. In general, temporal factor is the frequency of landslide occurence. It can be modelled in terms of annual probability (for quantitative analysis) linked to the triggers (rainfalls or earthquakes) based on threshold values or alternatively from multitemporal landslide inventory (if availiable, but often not the case!). Let it be said that for regional assessment this task is not trivial.

Here some usefull links to get familiar with the topic:

Crovelli, R.A. 2000. Probabilistic models for estimation of number and cost of landlsides. U.S. Geological Survey Open File Report 00-249, 23 pp. http://pubs.usgs.gov/of/2000/ofr-00-0249/ProbModels.html

Beside the temporal factor you should have a look on magnitude-frequency relations e.g:

Malamud, B.D.; Turcotte, D.L.; Guzzetti, F. And Reichenbach, P. 2004. Landslide inventories and their statistical properties.Earth Surface Processes and Landforms, 29: 687-711

Chapter from Guzzetti (2005): http://hss.ulb.uni-bonn.de/2006/0817/0817-5.pdf

Great question. To "add the temporal dimension to the susceptibility maps," does this mean that maps must be generated for each time step? I think so.

We can use precipitation as an example to demonstrate this, but we need to know the precise timing of the landslide, which is often difficult to obtain. Combining "static" geospatial inputs such as slope, land cover, anthropogenic factors, etc. with temporally-dynamic satellite-based rainfall is shown in this example: http://trmm.gsfc.nasa.gov/publications_dir/potential_landslide.html

For a good example of a national-level decision support system, El Salvador combines automated rain gauge data with maps of steep slopes to identify low, medium, and high, landslide potential based on prior knowledge of recent and accumulated rainfall: http://mapas.snet.gob.sv/geologia/deslizamientos2010.php

Both of these examples, at very different scales, depict the most imminent landslide hazards based on rainfall as a triggering factor. And so, a new hazard map is created every day.