Coastal vulnerability - Sea Level Rise


I've modeled two scenarios for SLR in my study area, but I'm not sure about the CVI results... I've read the thread by irondaffodill (title: Population Vulnerability Index) that Jess_Silver gave lots of input, and it was great help to start with.

I'm working on a small Brazilian coastal area, which would not experience differences in RSLR across the region.
SLR are estimated to be between 2 and 5 mm/year, so I created three scenarios: no rise, 2 mm rise, and 5 mm rise.

My "problem" is that the Exposure Index of the 2 mm and 5 mm scenarios seem identical (and the "no rise" scenario isn't that different either).

Could there have been some problem with the data used? Or is this result acceptable? 



  • Hi Carla,

    It sounds like what you're trying to have the model do is test between a current (no rise) and two future SLR scenarios.  So, in total you will have three different model runs, and in each you'll have a different rank for the SLR variable.  Since you don't have any spatial heterogeneity w/r/t SLR you will have the same rank for all shore points in a given scenario.  How are you currently ranking SLR for your scenarios?  I assume you have a 1 for the no rise, and then a different rank corresponding to 2mm and 5mm rise.  I've included a little text from a report that we recently put together where we wanted to take an approach similar to what I think you're trying to do.  This was based on time, rather than rise, but you could take a similar approach for your application as well.  I've also attached a paper from a few years ago that might be helpful (I would download the supplemental from this and take a look - it was too big for me to attach here).  It's a little different because here we used local tide gauge information to generate spatially explicit information, but nevertheless might be useful for you to look over.

    The last thing I would mention is that there are lots of different ways to compare across scenarios.  The way we generally do it is by calculating any categorical hazard breakdown (i.e. quartiles, high/med/low, etc) from the full distribution of values.  This means that you'll be pooling the CVI from your three scenarios and using that as your resultant distribution to distinguish cutoffs.  The NCA paper attached below walks you through that....


    Let me know if you have additional questions.


    analysis did not include any spatially explicit information about sea-level
    rise due to a lack of local tide gauge data. 
    Instead, we looked at the relative increase in risk assuming uniform
    sea-level rise across the entirity of the study area.  To do this, a planning horizon of 2040 was
    assumed as this agreed with some of the ongoing planning efforts underway.  Using the projected SLR curves depicted in
    Parris et al. (2012), we assigned projected sea-level for 2100 a rank of 5 and
    current SLR (2014) a rank of 1.  Looking
    at the shape of the curve a rank of 2 was estimated for 2040 to reflect the
    relative change in SLR from current to this time step.  To summarize, current SLR was represented by
    a rank of 1 and future SLR (in 2040) by a rank of 2.

  • Thanks for such a detailed response, Jess!

    Yes, my idea is to have three model runs to represent a current (no rise), 2 mm rise and 5 mm rise scenarios.

    Maybe I didn't understand the User's Guide well. I thought the "Trend" attribute would in fact represent the value of mm/year of SLR.

    This sentence in Appendix B "Once all polygons or points for specific regions are created, you must create an attribute field called “Trend” and populate it with values indicating net sea level change in mm/year according to Table 4.1." got me a bit confused.

    In Table 4.1, sea level change is ranked according to percentiles, but how do I calculate which percentile my increase in mm/year represents?

    In any case, my Trend attribute for the 2 mm scenario was "2", and for the 5 mm scenario was "5".
  • Hi Carla, 

    I can see how that would throw you off.  We should probably put some more detail in the UGuide with regards to SLR because there are a few different ways you could do it and I get a lot of questions about it.  That sentence refers to a layer that you might create if you knew that one part of your AOI had a trend of 2mm/yr and another part had a trend of 5mm/yr, and you wanted to capture that spatial heterogeneity.

    Right now the way you've run the model with all 2 for one scenario and all 5 for one scenario it make sense why side by side they would not have different spatial patterns - because you're not capturing anything spatial there.  I would assume though, that these two scenarios have different ranges of min/max for the resultant exposure index (CVI).  So, if you calculated your thresholds based on the full distribution of values from both scenarios (and the no rise one) you would then capture the influence of increased SLR. 

    Again, I'd encourage you to check out the text and paper (and the supplement!!) that I attached to the last email.

  • Right. I've gone through the supplementary material of the paper (which really inspired my MSc and PhD projects!), but I still have some questions...

    When I first ran the model, I uploaded my SLR .shp to the "Sea Level Rise (Vector)" and "Additional Layer (Vector)" areas.
    From what you've explained, that is in fact redundant, since there's no regional change? 
    I've ran the model again uploading the SLR .shp only to the "Sea Level Rise (Vector)" area, but results were the same - so redundant, but not a source of error.

    The problem I'm finding here is exactly the ranges of the CVI:
    For the no rise scenario it ranges from 1.48 to 3.62
    The 2 mm scenario ranges from 1.67 to 3.51
    And the 5 mm scenario ranges from 1.67 to 3.51

    So, in addition to the CVI values of the 2 mm and 5 mm scenarios being the same, the no rise scenario actually reaches higher CVI values...

    What could be causing this?

    I have some limitations regarding local data, so I'm using the default land polygon, bathymetry layer, relief, climatic forcing (including a large area to encompass enough WWIII points), and population layer.
    For this study I'm also only considering coral reefs for the habitat layer.

    Let me know if you'd like to see any of my data, I could e-mail it to you.
Sign In or Register to comment.