How to use PDD with monthly input spanning several years

lnicola

Hi! I am using ISSM v4.24 (from https://github.com/ISSMteam/ISSM), with Matlab. I want to force my model with monthly temperatures and precipitation fields into the future (in the shape of (num_years*12,x,y) or (number_vertices, num_year*12) after having it interpolated to my mesh. I want to tune the pddfac_snow and pddfac_ice so the calculated SMB matches a reference period.

I was running in a bit of a problem when trying to use the PDD scheme (src/m/classes/SMBpdd.m) as it would only allow a timeseries of 12 months.
How do I use the PDD option correctly on a monthly timestep beyond a year? What is the correct syntax? How do I run it with monthly input? I also have yearly ocean forcing. If you give timesteps, how do they have to be given to the model?
While trying to fix my runme, I found also the following: between the older SVN code and the current Github version v4.24 there was a slight change in the SMBpdd.m code.

In the newest version the md.smb.monthlytemperatures need to be of the shape (number_vertices, 12):

if(self.isdelta18o==0 & self.ismungsm==0)
md = checkfield(md,'fieldname','smb.monthlytemperatures','size',[md.mesh.numberofvertices 12],'NaN',1,'Inf',1);
md = checkfield(md,'fieldname','smb.precipitation','size',[md.mesh.numberofvertices 12],'NaN',1,'Inf',1);

but in a previous version (e.g: SVN27714M) it says

md = checkfield(md,'fieldname','smb.monthlytemperatures','timeseries',1,'NaN',1,'Inf',1);
md = checkfield(md,'fieldname','smb.precipitation','timeseries',1,'NaN',1,'Inf',1);

The .cpp seems to be equal between the versions. So it was always only allowing 12 months?

Not sure what ‘timeseries’ means, but maybe that the last row (num_vertices +1) contains the timesteps 1 2 3 4 … ?
Here is a snippet of my runme.m:

md.smb   = SMBpdd();
md.smb.monthlytemperatures = tas; % 2m surface air temperatures [K]
md.smb.precipitation       = pr;  % monthly surface precipitation [m/yr water eq]
sdim       = size(tas,1);         % num of vertices
md.smb.s0t = 2 * ones(sdim, 1);   % elevation from temperature source [m]
md.smb.s0p = 2 * ones(sdim, 1);   % elevation from precip source [m]
md.smb.issetpddfac = 1; % is user passing in defined pdd factors md.smb.pddfac_snow = 3.3* (1000/rhoi); %Pdd factor for snow for all the domain md.smb.pddfac_ice  = 8.8* (1000/rhoi); %Pdd factor for ice for all the domain md.smb.rlaps = 6.5; % present day lapse rate [degree/km]
md.smb.desfac= 0.5; % desertification elevation factor [m]

years_of_simulation = length(2016:2100);
md.timestepping.interp_forcing  = 1;
md.timestepping.start_time      = 21; % tas contain values from 1995-2300; if we start in 2016
md.timestepping.final_time      = years_of_simulation+ 21; %months_of_simulation+(21*12);  % run until x
md.timestepping.time_step       = 1/12;  %  monthly time step?
md.settings.output_frequency    = 1;  % every month?

I try to find a solution online but I couldn’t. I am happy to contribute to the documentation of the PDD option on the web https://issm.jpl.nasa.gov/documentation/pdd/ or in the manual.

Also, what does ismungsm==0 stand for? I could not find it anywhere what mungsm means or what this option does.
If there is another recommended way to calculate SMB from only tas and pr monthly input, I would appreciate any hints, thanks! (I have checked the GEMB model but it may need too many input variables for my current model set up).

lnicola

Tagging here maybe @joshcuzz as I saw in one of the forum discussion items, that you had a PDD setup?

joshcuzz

Hello.
It would be best if you tried the SMBd18opdd class. This was developed for paleo applications to allow a user to adjust a modern climatology based upon variations in ice core derived d180, but we have adapted this to be more widely used, allowing the user to input their own temperature and precipitation timeseries.

The setup should follow similarly to:

   % Set up SMB D180 class and set PDD factors
   md.smb = SMBd18opdd(); % Turn on SMB Class
   md.smb.isd18opd=1;  % Deactivate Ice Core Scaling, allowing user to pass in Precipitation. and Temperature time series
   md.smb.delta18o = [-40;0]; % Needs to be set, but not used here.  Only used for ice core scaling of temp, and precipitation.

   md.smb.rlaps=6.0; % This is the lapse rate you want to use
   md.smb.desfac = 1;  %Desertification level (between 0-1)
   md.smb.rlapslgm = 5.5; % Only used in Ice Core scaling.  
   md.smb.s0p = (max(dem,0));  % Reference DEM elevation for precipitation source
   md.smb.s0t = (max(dem,0));   % Reference DEM elevation for temperature source
 
   % Setting Temperature and Precipitation

   % Set your base climatology: 
   md.smb.temperatures_presentday = NaN(md.mesh.numberofvertices,12); % Set to a monthly mean over some time period. Length is as shown.
   md.smb.precipitations_presentday = NaN(md.mesh.numberofvertices,12);
  
   % Now, we want to pass in our own temperature and Precipitation time series
   md.smb.isprecipscaled=0;  % These allow us to pass in our own time series called Temp/Precipitations_reconstructed
   md.smb.istemperaturescaled=0;

   md.smb.precipitations_reconstructed =   NaN(md.mesh.numberofvertices+1,length(time));
   md.smb.temperatures_reconstructed =   NaN(md.mesh.numberofvertices+1,length(time));

   		% ** Note:  You need to pass in *Monthly* Temperature and Precipitation.  The size of the matrix is numberofvertices+1
		% The last row should be your timestamps (in monthly intervals): 1/12
		% For example, if you are running from 2014 to 2020, then time would start at 2014.0833 (first month) and step by 1/12
		% until you reach 2020.
		% Make sure your time stepping start and end times match these.  If monthly outputs (SMB, etc) are needed, make sure 		% time step is also monthly (1/12)

   % Set PDD Factors
   md.smb.issetpddfac = 1; % Allows user to adjust PDD factor
   md.smb.pddfac_snow = 4*ones(md.mesh.numberofvertices,1); %PDD factors can be spatially varying, or spatially constant
   md.smb.pddfac_ice = 8*ones(md.mesh.numberofvertices,1);

Apologies for any confusion. We need to update the documentation to acknowledge how to set this properly. If you run into any issues please do not hesitate to contact me at: Joshua.K.Cuzzone@jpl.nasa.gov

Thanks,
Josh Cuzzone