Modelling Protocol

The WaterMIP 2009A protocol (used for e.g. the first WaterMIP paper, "Multi-Model Estimate of the Global Terrestrial Water Balance: Setup and First Results", can be found here.

Content

• Atmospheric forcings
• Spatial resolution and landmask
• Elevation, land use/cover, soil information
• Routing network
• Human impacts
• Simulation and reporting periods
• Requested variables
• Array structure and variable naming conventions
• File information, daily files
• File information, monthly files
• Submission of simulation results  

Atmospheric forcings
WATCH Forcing Data (WFD), 1958-2001: The WATCH forcing data (from WATCH WB1) covering the period 1958-2001 can be found on the WFD ftp site at IIASA (the folder this is located in has its own password). See README file for information. The main variables are in 6-hourly or 3-hourly time steps, and at daily time step. NB! For rainf and snowf use the GPCC versions!  The WFD_Interpolator fortran code is available under folder WFD_code. The main use of the WFD_Interpolator will be to, for example, convert the 6-hourly WFD variables (Tair, Wind, Qair, PSurf and LWdown) to 3-hourly so that, together with the 3-hourly WFD variables (SWdown, Rainf and Snowf), models can be run with a diurnal cycle.

WATCH Forcing Data, 1900-1957: The WFD for 1901-1957 is available from the WFD IIASA ftp site. This means all the same format 3-hourly/6-hourly netCDF files as provided before for WFD 1958-2001 plus the daily average files. NB! For rainf and snowf use the GPCC versions! It is VERY IMPORTANT that everyone using these new data understands how they are created so please read the README file! In particular, these data will NOT have the correct times of sub-monthly extremes.

Routing network
The provided DDM30 routing network, which is mapped to the CRU land mask, should be used by all modelling groups. The network includes flow direction, slope, and basin numbers, and is available on the WATCH ftp site, see folder WorkBlock6. See also documentation in pdf file on the ftp site. The routing network includes large lakes not included in the provided land mask. These cells should not be included when results are submitted, though, and there should be no runoff added to the river network from these cells. I.e. these cells are included only for transportation purposes (streamflow).  

Human impacts
The models that can take human impacts like dams and water uses into account, should perform simulations both with and without human impacts. Be aware that time periods and requested variables are different for the human impact runs than those requested for the naturalized runs. For the human impact rounds, the modellers should include all water uses appropriate for their model. For those interested, the global land and wetlands database (GLWD1), which also includes information on reservoirs, can be found on the WATCH ftp site, see folder WorkBlock6. 

Simulation and reporting periods
All simulations should include a spin-up period of at least 5 years! The periods listed here include all model simulations, - see priority list below.

Using WFD:

• Simulation period 1958-2001 
  • Reporting period, daily fields (WATCH WB4 Multi-Model Analyses of Extremes): 1963-2001, naturalized runs
  • Reporting period, monthly fields (WaterMIP): 1971-2000 (naturalized and human impact runs)
• Simulation period 1900-1957
  • Reporting period, daily fields (WATCH WB4 Multi-Model Analyses of Extremes: 1905-1957, naturalized runs
  • No simulation results requested for WaterMIP 

2. Using forcings from WATCH WB3:

• Naturalized runs, simulation period 1960-2100
  • Reporting period, daily fields (WATCH WB4 Multi-Model Analyses of Extremes): 1965-2100
  • Reporting period, monthly fields (WaterMIP): 1971-2100 
• Human impact runs (WaterMIP, monthly fields): Reporting periods 1971-2000 (cntrl) and 2071-2100 (proj). 

Remember that all simulations should include a spin-up period of at least 5 years!

Priority list, simulations:
1. ECHAM A2, naturalized (including control period)
2. WATCH Forcing Data 1958-2001 naturalized run (if you have not done this already)
3. IPSL A2, naturalized (including control period)
4. CNRM A2, naturalized (including control period)
5. ECHAM A2, including human impacts, time slices (cntrl and proj)
6. IPSL A2, including human impacts, time slices (cntrl and proj)
7. CNRM A2, including human impacts, time slices (cntrl and proj)
8. ECHAM B1, naturalized
9. IPSL B1, naturalized
10. CNRM B1, naturalized
11. ECHAM B1, including human impacts, time slice (proj)
12. IPSL B1, including human impacts, time slice (proj)
13. CNRM B1, including human impacts, time slice (proj)
14. WATCH Forcing Data, 1900-1957 run, naturalized
15. WATCH Forcing Data, 1958-2001 run, including human impacts

Requested variables
All the requested output variables (see Tables 1-3) should be submitted as fields (gridded values) for each simulation run for land points and ocean points from -56 to 84 degrees latitude, and from -180 to 180 degrees longitude, i.e. 280 rows and 720 cols, or 201600 grid cells totally. Ocean points should be included because it makes visualizing the NetCDF files easier. The variable names and units are to follow the ALMA protocol (see menu "Data Format" and http://www.lmd.jussieu.fr/~polcher/ALMA).

The long-term water balance in each grid cell can be calculated as follows:

Precip = Qs + Qsb + Evap + Δ(SWE, SoilMoist, GroundMoist, SurfStor)

Table 1: Requested model output variables (see also menu "Data format" and http://www.lmd.jussieu.fr/~polcher/ALMA. Please submit as many variables as possible! For the human impact runs these variables should represent the results of the "actual water use run", if you do both an "actual" and a "potential" run.

 ¹⁾ If your model subdivides Qsb (subsurface runoff, slow drainage component of runoff response, which includes gravity drainage and lateral base-flow) into a slow and a quick subsurface component, then WB4 would also would like to receive Qsb_slow: slow subsurface runoff (WB4 will compute Qsb_quick = Qsb – Qb_slow). Please note that Qsb_quick is subsurface flow and not surface runoff (Qs)! The variable names Qsb_slow and Qsb_quick are not included in the ALMA conventions.

Table 2: List of requested model output variables for the WaterMIP simulations with human impacts (in addition to those in Table 1). See also menu "Data format". Follow the links for additional information. 

Table 3: Requested model output data, blue/green water terms. This list is for those submitting results for the blue/green water initiative. If you do not know what the blue/green water initiative is, you can ignore this Table, or, if you are interested, contact Jens Heinke and/or Holger Hoff (see Contacts information). See also menu "Data Format".

Array structure and variable naming conventions, submitted files

One file should include only one variable when submitting the data. Use only lowercase letters in the file names, since some NetCDF viewers only accept lowercase letters in their filenames. Data are to be submitted for the area -56 to 84 degrees latitude, and from -180 to 180 degrees longitude, i.e. 280 rows and 720 cols, or 201600 grid cells totally. Write the outputs row-wise starting at -55.75 and -179.75, and ending at 83.75 and 179.75.

Naming convention in NetCDF:
x-dimension name: lon
x-dimension units: degrees_east
y-dimension name: lat
y-dimension units: degrees_north
t-dimension name: time
t-dimension units, daily files: days, i.e. numbered from 1 to 365/366 for the daily fields (0 to 364/365 is also acceptable). t-dimension units, monthly files: months since first month in file.
missing_value: 1.e+20f

Example of naming of variables (for more information on the ALMA specifics, see menu "Data Format"):
Long field name: Total evapotranspiration (Use the ALMA variable description)
Short field name: Evap (Use the ALMA variable names!)
Field units: kg/m²s (Use the ALMA variable units!)

File information, submitted files with daily fields (WATCH WB4, Multi-Model Analyses of Extremes) 

One file should include the time series of daily values for only one variable and for one year when submitting the data. Use only lowercase letters in the file names, since some NetCDF viewers only accept lowercase letters in their filenames. For the period 1963-2001 and 10 variables this means 39*10 = 390 files. The submitted time series of daily data for a particular year should use the following file naming convention, where forcing refers to input forcings, setup refers to model setup, variablename refers to the variable names used in the ALMA convention, and yyyy refers to the year:

modelname_forcingsetup_variablename_daily_yyyy.nc

For forcing the following abbreviations should be used:

wfd – Watch forcing data
echamcon – Echam control period
echama2 – Echam A2 scenario
echamb1 – Echam B1 scenario
ipslcon – IPSL control period
ipslca2 – IPSL A2 scenario
ipslb1 – IPSL B1 scenario
cnrmcon – CNRM control period
cnrma2 – CNRM A2 scenario
cnrmb1 – CNRM B1 scenario

For "setup" the following abbreviation should be used:
nat - naturalized simulations

Examples: