Spinup using IHISTClm50Sp

[Shreya Bhowmick]

Hello everyone,

1) I want to start a simulation using compset FHIST having satellite phenology from the year 1950. Can compset IHistClm50Sp be used for land spinup from 1901 to 1950? Or is it better to use I2000Clm50Sp/I1850Clm50Sp for 50 years for the land spinup?

2) I have simulated IHistClm50Sp forced with GSWP3V1 from 1901 to 1960, 20 years at a time. The annual mean was calculated from the monthly output files. I have plot the spinup for the variable H2OSOI over an Indian region which is attached. There seems to be a jump after every 20 years when the model was restarted. Also this looks different from the one given as example in the CLM5.0 userguide. Is there any problem or mistake I am committing?

Thank you for your time,
Shreya.

 

[oleson]

Typically, we would spinup using I1850Clm50Sp and then run an IHIST transient from 1850 to whatever year we want to start from. Using an IHIST compset to spinup introduces some transient behavior (e.g., CO2, ndep, etc) into the spinup.
Your plot probably looks different because the userguide example plots the same year from each cycle of atmospheric data. E.g., if the atmospheric forcing is 1901-1920, model output from each cycle of 1901 data is used, whereas you are plotting every year.

 

[Shreya Bhowmick]

Thank you for your reply. Since GSWP3 starts from 1901, with what data should the IHIST transient run be forced if we start from 1850? Also doing a 50 year spinup and then a 100 years simulation from 1850 to 1950 will computationally expensive for us. Is the IHIST land spinup from 1901-1950 forced with GSWP3 sufficient for starting a FHIST run from 1951?

With regards,
Shreya

 

[oleson]

We use GSWP3 data for 1901-1920 to spinup an 1850 run, by necessity. Then we use 1901-1920 data to run the transient from 1850 through 1900. Then 1901-2014 data to run 1901-2014 transient.
The main purpose of the spinup is to get state variables such as soil moisture and temperature, carbon stocks, etc. into equilibrium before starting a transient simulation. The model comes with spunup initial files for certain starting points, e.g, 1850, 2000. If you change either the model code or the surface datasets then some spinup is likely required. The duration/type of the spinup required will depend on the magnitude of your changes and how they might affect the equilibrium of the state variables. You'll have to asses that yourself. Sorry I can't be more specific.