September 17, 2021, 10:50:20 PM

News:

Due to SPAM attacks, new members must be approved before posting.  Please email jclough@warrenpinnacle.com when registering and your account will be approved.


Model calibration

Started by azarcos, June 02, 2020, 04:27:05 AM

previous topic - next topic
Go Down

azarcos

Hi Jonathon,

I am new to SLAMM. I want to conducted model calibration by running the "time zero" step, following the same methodology in Clough et al. (2016). I know already that no sea-level rise, accretion or erosion should be considered, but I am not sure about what to introduce in the last year of simulation. For how long should I run the model?
In the SLR scenarios to run' should I apply 0m by 2100? Can you please provide more details on the calibration procedure?
Thanks for your help

cheers
Rita

Jonathan S. Clough

The time-zero run is a very important step.  To the extent possible, it should not be any different than the model's initial conditions.  It tests that the conceptual model, the wetland coverage, the elevation data, and the tide range data are all consistent.  In the results the initial condition is shown as "0" and the "time zero" result is shown as the first date of the simulation.

From the tech doc:
SLAMM can also simulate a "time zero" step, in which the conceptual model can be validated against the data inputs for your site. The time-zero model predicts the changes in the landscape given specified model tide ranges, elevation data, and land-cover data. Any discrepancy in time-zero results can provide a partial sense of the uncertainty of the model. There will almost always be some minor changes predicted at time zero due to horizontal off-sets between the land-cover and elevation data-sets, general data uncertainty, or other local conditions that make a portion of your site not conform perfectly to the conceptual model. However, large discrepancies could reflect an error in model parameterization with regards to tide ranges or dike locations, for example, and should be closely investigated.

I would suggest setting up the model to run for a couple of years (photo date, photo date + 1, photo date + 1) with minimal or zero SLR.  Any changes predicted should be understood as much as possible in terms of data error (issues with the tide model, DEM, or wetland coverage), lack of dike or seawall accounting, etc.  You should also examine the inundation frequency maps to ensure that the wetlands are being regularly wetted.  You can consider using the results from the time-zero run as your initial condition for your projection runs to ensure that the effects that you are seeing in the simulation are from the SLR signal and not uncertainty or lack of precision in your input data sets.

Sorry about the long delay in response.  -- Jonathan

azarcos

June 18, 2020, 05:44:27 AM #2 Last Edit: June 19, 2020, 04:25:13 AM by azarcos
Dear Jonathan,

Thanks for your reply and advices, will do. The inundation files are the ... _Inund_Freq_GIS.SLB. How do I open these rasters?

At the end of the simulation a csv file is generated, named with the start and end date, and containing the summary of the modelled scenario. Are the values distributed in columns for each category in ha?

cheers

Jonathan S. Clough

Hi:  The SLB files are a SLAMM Binary format, not readable by GIS.  You need to either re-run and export to non-binary format (ASCII Rasters readable by ESRI and QGIS) or convert the binary files.

See the users manual, page 14-15 under "Data to Save" (GIS files options on the file execution screen) to do the former. 

To convert the binary files go to SLAMM File Setup and click the "Conv Binary Files to ASCII" button.  You can then drag the SLB file to the interface and it will convert to an ASCII Raster.

The output CSV is indeed in hectares,  (an embarrassing omission from the users manual!)

Best regards -- Jonathan


Go Up