A user guide for the ACCESS land surface model Community
Contents
1. 2 872 2 872 2 872 2 872 2 872 2 872 2 872 2 872 2 872 Vegetation types 1 broad leaf evergreen trees tropical rainforest 2 broad leaf needle leaf and broad leaf deciduous trees 3 broad leaf and needle leaf trees 4 needle leaf evergreen trees 5 needle leaf deciduous trees 6 broad leaf trees with ground cover short vegetation C4 grassland savanna 7 perennial grasslands 8 broad leaf shrubs with grassland 9 broad leaf shrubs with bare soil 10 tundra 11 bare soil and desert 12 agricultural c3 grassland 13 ice VEG Type 1 Type 2 Type 3 Type 4 Type 5 Type 6 Type 7 Type 8 Type 9 Type Type Type Type 10 11 12 13 canstl 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 dleaf 0 075 0 12 0 07 0 028 0 02 0 16 0 16 0 155 0165 0 155 0 005 0 155 0 005 ejmax 170 0 170 0 160 0 130 4 140 0 20 2 18 0 70 106 78 0 76 106 34 04 240 0 2 0 10 frac4 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 he 35 0 20 0 20 0 17 0 17 0 1 0 0 5 0 6 0 5 4 0 0 05 1 0 0 01 iveg 1 2 3 4 5 6 7 8 9 10 11 12 13 meth 0 0 0 0 0 0 0 0 0 0 0 0 0 ratecp1 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 ratecp2 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 0 03 ratecp3 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 0 14 rp20 1 1342 1 4733 2 3704 3 3039 3 2879 1 0538 0 8037 12 032 1 2994 7 4175 22. CABLE s essential inputs and outputs in several panels In each a simple time series of the variables are shown from a specified start point to a specified end point Good for diagnosis model behavioural issues Output can be chosen to go to screen pdf postscript png or jpeg e generalTest R this script is intended as a general diagnostic test for CABLE if for example one were testing changes to CABLE In sequence it creates a new directory named by date and time runs CABLE for one site produces all four of the above plots for this site and saves them to the new directory repeats for other sites in a user defined list within the script By default generalTest will produce the timeseries plot for the entire data set for any given site and use png output for this plot vector format files can get very large otherwise regardless of the output format specified in the call to generalTest The other three plots will save 14 CABLE user guide v1 3 in the format specified in the argument to generalTest All five of these R scripts require that CABLEplots R also included in the distribution be present in the same directory This file contains the analysis and plotting commands for all of the scripts so any user modification to the functionality of these scripts will likely be in this file 15 CABLE user guide v1 3 References Bernhofer C M Aubinet R Clement A Grelle T Gr nwald A Ibrom P Jarvis C Rebmann E D S
3. H A and Klooster S A 1993 Terrestrial ecosystem production A process model based on global satellite and surface data Global Biogeochemical Cycles 7 4 811 841 Stainforth D A T Aina C Christensen M Collins N Faull D J Frame J A Kettleborough S Knight A Martin J M Murphy C Piani D Sexton L A Smith R A Spicer A J Thorpe amp M R Allen 2005 Uncertainty in predictions of the climate response to rising levels of greenhouse gases Nature 433 pp 403 406 Vrugt J H V Gupta L A Bastidas W Bouten and S Sorooshian 2003 Effective and efficient algorithm for multi objective optimization of hydrologic models Water Resources Research 39 8 1214 doi 10 1029 2002WRO001746 Wang Y P Leuning R Cleugh HA and Coppin PA 2001 Parameter estimation in surface exchange models using non linear inversion how many parameters can we estimate and which measurements are most useful Global Change Biology 7 495 510 Zobler L 1988 A world soil file for global climate modeling Technical Report 87802 NASA Goddard Institute for Space Studies 16 CABLE user guide v1 3 Appendix 1 CABLE s structure and coding CABLE s Fortran coding while not necessarily elegant is written to aid understanding of its structure and content Below is a figure which illustrates the flow of module dependency in CABLE Each dashed region represents a file and each coloured box represents a module An
4. Marine and Atmospheric Research 2 Tharandt Dresden Germany 140y o coniferous forest site half hourly time step four years 1997 2000 are provided See Bernhofer et al 2003 and Griinwald and Bernhofer 2007 for more detail Data were provided by Christian Bernhofer of the Technical University Dresden 3 Bondville Illinois USA crop site annual rotation between corn C4 and soyabean C3 half hourly time step three years 1997 1999 are provided Data were provided by Tilden Meyers of NOAA ARL ATDD These data are available in netcdf e g Tumbarumba nc or text e g Tumbarumba txt format and are gap filled datasets The netcdf versions additionally contain observations of latent heat flux Qle sensible heat flux Qh and net ecosystem echange of CO NEE 12 CABLE user guide v1 3 9 Addressing parameter uncertainty Calibrating CABLE This section provides a suggestive framework for addressing parameter uncertainty in CABLE Traditionally this has been done using an automated parameter estimation technique the approach we cover in the first subsection below Alternatively if one has reason to believe that CABLE is biased for the purpose of the simulation so that automated parameter estimation is likely simply to tune CABLE to a particular dataset details about using a perturbed parameter ensemble to address parameter uncertainty are given in the second subsection Calibration automated parameter estimation B
5. driver for CABLE non default parameters will need to be specified by the user in the Fortran code For a discussion on more advanced techniques of parameter selection such as automated calibration or parameter perturbed ensembles see section 9 below Using the default parameter sets from the default grid As mentioned above each of the provided driver routines for running CABLE offline is able to utilise a default_params subroutine which collects values for each of the parameters above point by point based on latitude and longitude For a given simulation grid point this subroutine first locates the closest grid point in a coarse 2 by 2 global map of nine soil based on Zobler 1998 and thirteen vegetation types based on Potter et al 1993 and then uses these types to infer the above parameters Note that these values are unlikely to appropriate in most circumstances particularly using different grid sizes Care should be taken for example using these coarse grid parameter values for single site applications there is no guarantee that the correct vegetation or soil type will be chosen for example Currently all sites have a default reference height of 40m The values for each soil and vegetation type are as follows Soil types 1 Coarse sand Loamy sand 2 Medium clay loam silty clay loam silt loam 3 Fine clay 4 Coarse medium sandy loam loam 5 Coarse fine sandy clay 6 Medium fine silty clay 7 Coars
6. have a netcdf installation available The netcdf package must be compiled with access to the same Fortran 90 compiler you are using to compile CABLE The netcdf based CABLE driver using make or make netcdf uses the netcdf Fortran 90 interface There must be a symbolic link to netcdf mod part of the netcdf installation in the directory in which you want to compile CABLE In addition you will need to change the netcdf path in the Makefile to reflect the location of your netcdf installation Compiling with the supplied Makefile linux unix To use the Makefile all that should be required is to modify the Makefile to reflect the command line Fortran compiler you wish to use and if you are using netcdf the path of the netcdf installation For example FC g95 e g g95 or 1fort or 1f95 etc FFLAGS NCDIR usr local netcdf 3 6 1 Compiling without the Makefile The order of file compilation for CABLE is as follows 1 cable_variables f90 2 cable_soilsnow f90 cable_carbon f90 cable_parameters f90 each of these depend only on cable_variables 3 cable_cbm f90 4 cable_checks f90 For netcdf i o For text i o 5 cable_input f90 5 cable_output_text f90 6 cable_output f90 6 cable_drivertxt f90 7 cable_driver f90 This information can be inferred from the supplied Makefile or the CABLE structure diagram in Appendix 1 The code has been tested to compile and run on Windows Compaq g95 Intel amp Lahey Fujitsu co
7. land land compress y x float nav_lat y x nav_lat units degrees_north nav_lat valid_min 90 f nav_lat valid_max 90 f nav_lat long name Latitude float nav_lon y x nav_lon units degrees_east nav_lon valid_min 180 f nav_lon valid_max 180 f nav_lon long_ name Longitude float time tstep time units seconds since 1949 01 01 00 00 00 time title Time time long_name Time axis time time_origin 1949 JAN 01 00 00 00 Most of the variables will be structured as SWdown is above The relationship between land and x and y inside the CABLE netcdf driver is y INT landGrid j 1 xdimsize x landGrid j y xdimsize y y l 4 Simulations of more than a single site gridpoint which do not use the compression by gathering structure described above must contain an integer mask variable dependent on the x and y dimensions only A 1 value implies land gridpoint anything else is assumed to be ocean 5 Meteorological variables in broadly conforming to the ALMA standard are of dimension 3 x y t or 4 x y z t if using the x y grid or 2 land t if using the compressed grid All must have a text units field Essential variables are SWdown Tair Qair Rainf Wind or Wind_N and Wind_E and optional variables are LWdown can be synthesised from AirT PSurf can be estimated at a fixed value based on temperature and an elevation vari
8. one or the other type make netcdf or make text It is recommended that netcdf driver is used especially if running at more than a single point for several reasons Firstly all spatial and temporal timing variables are guaranteed to be set as the netcdf version will not run if they are not present or consistent Next the ranges and units of all input met and output variables will be checked as they are read and written Lastly the netcdf version prints a more detailed log file which reports any discrepancies during execution as well as all model parameters and variable initialisations The text version by comparison uses a quick and dirty and much simpler driver and i o routine for confident users Somewhere between a one gridpoint and a many gridpoint simulation the netcdf routine will become the less time consuming of the two even for very experienced users Using CABLE in a coupled model To use CABLE as a coupled model keep in mind the five points in Section 4 In particular make sure your atmospheric variables are compatible with CABLE s grid CABLE uses a land only grid and its variables have only one spatial dimension Translating a land sea grid to CABLE s land only grid can be made easier with the Fortran commands 2 CABLE user guide v1 3 PACK and UNPACK Spatial and temporal variables are discussed in Section 5 below 3 Compiling CABLE Notes on using netcdf To use the netcdf driver for CABLE you must
9. the met file and set to be 2 Medium clay loam silty clay loam silt loam all those parameter values associated with soil type 2 will be set e g bch 7 1 If bch is also present in the met file say set as bch 6 5 then 6 5 will be used Note the danger this poses in terms of possible inconsistent parameter values e g setting clay 0 4 in addition to isoil 2 will result in clay silt sand not equal to 1 7 Initialisation Below is a complete list of variables that need to be initialised Both of the provided drivers utilise the subroutine default_params in cable_parameters f90 which will initialise these variables using values from a global simulation using CABLE coupled to CCAM They help CABLE avoid instability oscillations during the first few time steps but are very crude and of course not accurate initial states for any point in time making sure CABLE is properly initialised is your responsibility The list of variables to be intialised all of dimension mp unless stated otherwise is canopy cansto canopy water storage mm or kg m canopy sghflux canopy oghflux ssoil ssdnn overall snow density kg m ssoil snowd snow liquid water equivalent depth mm or kg m ssoil osnowd snowd from previous timestep mm or kg m ssoilYosnage snow age ssoil isflag snow layer scheme flag 0 no or little snow 1 snow ssoilY owbice soil ice dimension mp 6 ssoilYotggsn snow temperature per layer K dimension m
10. 8879 3 0 0 10 rpcoef 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 0 0832 shelrb 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 tminvj 5 0 5 0 5 0 2 0 5 0 10 0 10 0 0 0 5 0 5 0 5 0 10 0 5 0 tmaxvj 15 0 15 0 10 0 5 0 10 0 15 0 15 0 5 0 10 0 0 0 10 0 15 0 0 0 vbeta 1 0 1 0 1 0 1 0 1 0 4 0 4 0 4 0 4 0 1 0 4 0 4 0 1 0 vemax 85 0 85 0 80 0 65 2 70 0 10 1 9 0 35 053 39 0 38 053 17 02 120 0 1 0 10 xfang 0 1 0 25 0 125 0 01 0 01 0 3 0 3 0 2 0 01 0 2 0 01 0 3 0 01 froot 1 0 02 0 04 0 04 0 04 0 04 0 05 0 05 0 05 0 05 0 05 0 05 0 05 0 05 froot2 0 06 0 11 0 11 0 11 0 11 0 15 0 15 0 10 0 10 0 10 0 10 0 15 0 15 froot3 0 14 0 20 0 20 0 20 0 20 0 35 0 35 0 35 0 35 0 35 0 35 0 34 0 35 froot4 0 28 0 26 0 26 0 26 0 26 0 39 0 39 0 35 0 35 0 35 0 35 0 38 0 40 CABLE user guide v1 3 froot5 0 35 0 24 0 24 0 24 0 24 0 05 0 05 0 10 0 10 0 10 0 10 0 06 0 04 froot6 0 15 0 15 0 15 0 15 0 15 0 01 0 01 0 05 0 05 0 05 0 05 0 02 0 01 Note that the parameter values associated with a particular vegetation and or soil type will be used if that type is specified in the met forcing file while using the netcdf driver Any additional parameter specifications in the met file will overwrite these values For example if isoil is present in
11. CABLE user guide v1 3 A user guide for the ACCESS land surface model Community Atmosphere Biosphere Land Exchange CABLE Gab Abramowitz Yingping Wang and Bernard Pak University of New South Wales CSIRO Marine and Atmospheric Research gabsun gmail com For CABLE version 1 3 October 2007 This document is a user guide for the Community Atmosphere Biosphere Land Exchange CABLE land surface model It is likely to be changing regularly as CABLE itself is updated please check the ACCESS website to make sure you have the latest version It compliments CSIRO Marine and Atmospheric Research paper 013 Kowalczyk et al 2006 which provides a full technical description of CABLE as well as its development history It is intended to help users understand the workings of CABLE and speed its implementation The authors however assume no responsibility for results which arise from its use 0 Introduction 1 Files in the basic CABLE package 2 Driving CABLE 3 Compiling CABLE 4 General considerations when running CABLE 5 Temporal and spatial variables 6 CABLE parameters 7 Initialisation spin up and restart files 8 Input and output 9 Addressing parameter uncertainty calibrating CABLE 10 Viewing CABLE s output and judging performance References Al CABLE s structure and coding 0 Introduction Like most land surface models LSMs CABLE simulates the exchanges of radiation moisture heat and c
12. E s met arrays with the correct units Details are given in Section 8 d Temporal and spatial variables ensure that CABLE s time step size run length number of gridpoints and gridpoint details are compatible with your simulation Instructions for changing these are given in Section 5 e Output CABLE s output obviously needs to be of a nature and in a format that you can process and is useful for your application including variable units Options for output are discussed in Section 10 Making sure that these areas are appropriately configured for your application is your responsibility Plotting CABLE s output in a variety of measures to make sure it is behaving properly is essential Example plotting scripts are described 3 CABLE user guide v1 3 in Section 10 Log file and diagnostics There are two levels of diagnostics Permissible runtime comments are written to the log file name determined in the namelist file cable nml whilst serious and or fatal errors are printed to screen The log file contains details of timing locations initialisations and parameters used during simulation If using the netcdf driver with many sites being simulated writing of each site s initialisations and parameter values can be suppressed by setting verbose to FALSE in cable nml namelist file to avoid a very large log file 5 Temporal and spatial variables CABLE s variables use a single spatial dimension That is for regional glob
13. able Snowf assumed to be included in Rainf if not present and CO2air assumes a fixed value determined in the cable nml namelist file Note that for PSurf to be estimated a variable elevation dependent on x and y dimensions must be present 6 To ensure the ranges of input variables are appropriate make sure checkY oranges TRUE in the driving routine 10 CABLE user guide v1 3 7 Met input data must be continuous in time if date time variables are going to be correct Example input files are provided with CABLE Both single and multiple site examples are included Using text input In the text driver a simple read line gives an example of how one might input met data Note that there are no units changes here something you may need for your own input data CABLE s variables and units for meteorological data are Downward shortwave radiation met fsd Wim2 Downward longwave radiation met fld Wim2 Precipitation solidt liquid met precip mm time step e Surface air temperature met tk K Surface wind speed met ua m s e Surface specific humidity metY qv kg kg e Surface air pressure met pmb mbar or hPa e Surface air carbon dioxide concentration met ca mol mol Leaf area index In addition to parameters and met forcing CABLE requires leaf area index LAI input This can be provided in one of two ways Firstly and recommended is that the user provides LAI values in th
14. al simulations the two dimensional latitude and longitude based grid needs to be mapped to a single line of sites The provided netcdf driver automatically maps from latitude longitude references in the netcdf input file to a single dimension It can do so either with the use of a mask variable or through an ALMA style compressed land grid see Section 8 Fortran pack and unpack functions may be useful when doing this manually for coupled runs The single dimension allows CABLE to run for the entire region globe in a single subroutine call Netcdf driver The netcdf offline driver will set all temporal and spatial variables from data in the meteorological data input file Text driver In the supplied text based offline driver temporal and spatial variables are set manually either in the namelist file cabletxt nml which the driver reads or in the fortran code itself The user must set mp the number of land grid points dels time step size kend the number of time steps in the run latitude and longitude for each grid cell site simulated as well making sure all grids map consistently between input and output 6 CABLE parameters Each one of the parameters in the table below needs to be set to represent the site s region s being simulated CABLE has mp land points and ms soil layers Note that the ranges here are physically possible ranges for the entire globe they are not like
15. alibration Techniques Below is a brief list of automated parameter estimation techniques check the CABLE user site to see if drivers for these or others are available The Multi Objective Shuffled Complex Evolution Metropolis algorithm MOSCEM This is a monte carlo based 13 CABLE user guide v1 3 technique is described by Vrugt et al 2003 It uses a multiple criteria approach instead of weighting cost functions it considers them independently and so produces and optimal region of the parameter space for a given calibration data set rather than a single optimal parameter set A driving routine for CABLE that is compatible with this C based code cable_moscem f90 is available please contact Gab Abramowitz gabsun gmail com PEST This is a gradient descent based technique is described in Doherty 2001 A published example of calibration using CABLE s predecessor CBM is given in Wang et al 2001 Addressing parameter uncertainty using a perturbed parameter ensemble An alternative approach to addressing parameter uncertainty is using a perturbed parameter ensemble e g Stainforth et al 2005 The process is as follows First identify the ranges of uncertainty for each model parameter Then for each parameter select a finite number of values from within the range to adequately represent the qualitative spread of model behaviour associated with variation in the particular parameter This is usually just a matter of adequate dens
16. arbon at the land surface It is provided with meteorological conditions its inputs and based on these predicts fluxes its outputs e g latent heat flux upward long wave radiation net ecosystem exchange of CO or drainage through deep soil The characteristics of the soil and vegetation in the regions it simulates are for the most part constant throughout the simulation and described by CABLE s parameters Variables which CABLE stores in memory from one time step to the next known as model states include soil moisture content soil and vegetation temperatures and carbon stored in the vegetation and soil An abstract schematic representation of these model components is shown in Figure 1 PARAMETERS i INPUT md MODEL y OUTPUT t 1 STATES Figure 1 A schematic representation of a model with real number inputs outputs states and time independent parameters Ultimately CABLE is designed to work as a single component of a larger global climate model GCM In this situation CABLE receives meteorological data from and passes flux information to a boundary layer atmospheric model When coupled to such a model CABLE is said to be running online Alternatively meteorological data from CABLE user guide v1 3 observational sites or saved from an atmospheric model can be used to force CABLE and its output simply saved to file in which case it is operating offline Using CABLE involves agreeing to a license agreement largely
17. chulze and J D Tenhunen 2003 Spruce forests Norway and Sitka spruce including Douglas fir Carbon and water fluxes and balances ecological and ecophysiological determinants In Ecological Studies Vol 163 Fluxes of Carbon Water and Energy of European Forests R Valenitini ed 99 123 Doherty J 2001 PEST Model Independent Parameter Estimation Watermark Numerical Computing Falge E D Baldocchi R J Olson P Anthoni M Aubinet C Bernhofer G Burba R Ceulemans R Clement H Dolman A Granier P Gross T Griinwald D Hollinger N O Jensen G Katul P Keronen A Kowalski C Ta Lai B E Law T Meyers J Moncrieff E Moors J W Munger K Pilegaard U Rannik C Rebmann A Suyker J Tenhunen K Tu S Verma T Vesala K Wilson S Wofsy 2001 Gap filling strategies for longterm energy flux data sets Agricultural Forest Meteorology 107 71 77 Griinwald T C Bernhofer 2007 A decade of carbon water and energy flux measurements of an old spruce forest at the Anchor Station Tharandt Tellus B 59 387 396 Kowalczyk E A Y P Wang R M Law H L Davies J L McGregor and G Abramowitz 2006 The CSIRO Atmosphere Biosphere Land Exchange CABLE model for use in climate models and as an offline model CSIRO Marine and Atmospheric Research paper 013 www cmar csiro au e print open kowalczykea_2006a pdf Potter C S Randerson J T Field C B Matson P A Vitousek P M Mooney
18. e medium fine sandy clay loam 8 Organic peat 9 Permanent ice SOIL Type 1 Type 2 Type3 Type 4 Type 5 Type 6 Type 7 Type 8 Type 9 albsoil 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 bch 4 2 7 1 11 4 5 15 10 4 10 4 7 12 5 83 7 1 clay 0 09 0 30 0 67 0 20 0 42 0 48 0 27 0 17 0 30 css 850 850 850 850 850 850 850 1920 2100 hyds 10 166 0 4 0 1 0 21 0 2 0 1 0 6 0 800 0 1 0 isoilm 1 2 3 4 5 6 7 8 9 ratecs1 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 ratecs2 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 ratecs3 0 004 0 004 0 004 0 004 0 004 0 004 0 004 0 004 0 004 rhosoil 1600 1600 1600 1600 1600 1600 1600 1300 910 rs20 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 sand 0 83 0 37 0 16 0 60 0 52 0 27 0 58 0 13 0 37 sfc 0 143 0 301 0 367 0 218 0 31 0 37 0 255 0 45 0 301 silt 0 08 0 33 0 17 0 20 0 06 0 25 0 15 0 70 0 33 ssat 0 398 0 479 0 482 0 443 0 426 0 482 0 420 0 451 0 479 sucs 0 106 0 591 0 405 0 348 0 153 0 49 0 299 0 356 0 153 swilt 0 072 0 216 0 286 0 135 0 219 0 283 0 175 0 395 0 216 zse 0 022 0 022 0 022 0 022 0 022 0 022 0 022 0 022 0 022 zse 2 0 058 0 058 0 058 0 058 0 058 0 058 0 058 0 058 0 058 CABLE user guide v1 3 zse 3 0 154 0 154 0 154 0 154 0 154 0 154 0 154 0 154 0 154 zse 4 0 409 0 409 0 409 0 409 0 409 0 409 0 409 0 409 0 409 zse 5 1 085 1 085 1 085 1 085 1 085 1 085 1 085 1 085 1 085 zse 6
19. e must have a complete set of parameters as restart files written by the netcdf driver do whereas the met file may have only a subset of CABLE s parameters When prescribing only a subset of the parameters in the met file take care to avoid parameter inconsistencies such as those described above If no restart file is found for example the remaining subset of unprescribed parameter values will be loaded from the default grid described below which may give inappropriate vegetation and soil types for your site A specification of isoil and or iveg soil and vegetation type in the met file will load all parameters associated with that type any additional parameter specifications will overwrite these Both the meteorological forcing file and the restart file for the netcdf driver are netcdf The variable names for each of the parameters are as listed in the table above all lower case They of course being parameters have no time dimension Spatial representation in the met file is dependent on whether an x y grid with land sea mask or a land only vector essentially as shown in the table above with dimension mp is used Please see the Input and output section below for the distinctions between these two spatial representations Spatial representation in the restart file will always CABLE user guide v1 3 be land only so that parameters stored there will have the same dimensions as in the table above In the sample text based
20. e netcdf met file The LAT variable in this file can either have the same spatial and temporal dimensions as other variables e g tstep land in the case of compressed land grid file see the section in input output above or if appropriate be time invariant have only spatial dimensions in the netcdf file as a parameter The second option which is the default in the netcdf driver when no LAI variable is found in the met file is a provided coarse grid LAI file LAI Monthly Global nc containing monthly global MODIS LAI data using the subroutine get_default_lai inside cable_input f90 The provided text driver reads in LAI with met data LAI is therefore included in the example text met data files provided The LAI variable within CABLE is veg vlai which has dimension mp the number of land grid points Using output from the netcdf driver The netcdf driver writes output to a netcdf file of a similar format to that required in the met input file The default name for this file is out_cable nc which is set in the CABLE namelist file Dimensions mimic those whether based on a land sea mask or an ALMA style compress by gathering grid in the met data file The variables to be included in the output file can be set in one of two ways Firstly by setting the logical output switches in the namelist file e g output oflux TRUE The collection of switches and their corresponding output variables are output me
21. elow is a brief discussion about calibrating CABLE It covers the type of data required as well as details about a few automated parameter estimation techniques Information about several example single site datasets which we feel meet quality criteria is also provided Choosing which parameters to calibrate the ranges to use for these during calibration and the values for those parameters which will remain static can be a tricky business There is no unique or correct way to calibrate and often no unique result Making these choices will take knowledge about the site s at which one is calibrating which parameters is CABLE likely to be sensitive to in this environment What additional information is available to help narrow the ranges of uncertainty for parameters In the first instance the default values and ranges for parameters in Section 6 could be used More broadly automated parameter estimation techniques usually make very strong assumptions about parameter independent model bias and observational bias Using a technique that assumes the only barrier to a model s perfect match with observations is correct identification of parameters can be problematic This often becomes evident when during the calibration process parameters push against the boundaries of the ranges specified as realistic An experiment using a parameter estimation technique to infer true parameter values for a site is therefore unlikely to succeed Dataset gu
22. h netcdf input output and one with text input output It is recommend that users choose the netcdf driver where possible particularly if they are not familiar with CABLE In the netcdf driver all input data and running variables spatial and temporal details are retrieved from a netcdf file s the user does not then need to compile the code more than once It also produces netcdf output with a choice which output variables are to be written with variable units included in the file In the text driver the running variables must be set manually inside the driver code with meteorological data read from a formatted text file These processes are described in more detail in Sections 4 8 The two driving routines are named cable driver f90 and cable_drivertxt f90 respectively Both the netcdf and text driver use a namelist file cable nml and cabletxt nml respectively to read in some operating variables Also both may call the default_params subroutine in cable_variables f90 to generate default parameter and initial values for a run if no other source is found see Sections 6 and 7 These values may not be appropriate for your application Both will also produce a log file named log cable txt by default can be changed in the namelist file see Section 4 which lists the parameter values read for each grid cell in the simulation as well as the initial states By default the included Makefile will compile the netcdf version To compile
23. ible heat from vegetation W nr output balances Ebal Cumulative energy balance W m Wbal Cumulative water balance mm output carbon NEE Net Ecosystem Exchange of CO2 umol m 7 s GPP Gross Primary Production of CO2 umol m s NPP Net Primary Production of CO2 umol m s AutoResp plant respiration umol m s HeteroResp soil respiration umol m s output oparams All CABLE input soil and vegetation parameters listed in Section 6 Alternatively one can explicitly name the variables to be included Any of the variables in the above list for example NEE can be included by adding output NEE TRUE below the other output statements in the namelist file The output file will contain both variables specified through the group output variables above as well those specified individually Note that if the land sea mask grid type is used sea grid squares are given a default undefined value unique for each variable set in the output module Using text output A simple example text output subroutine is included with CABLE and is contained within the file cable_output_text f90 Sample data sets Provided with CABLE are several sample data sets from flux tower observations The three sites included are 1 Tumbarumba south eastern NSW Australia wet sclerophyll forest one hour time step two years 2002 2003 are provided Data were provided by Eva van Gorsel and Ray Leuning of CSIRO
24. idelines Datasets for calibration must have a complete gap filled set of CABLE s input met variables see Section 8 and also most likely require a gap filled time series of the variable s which one wishes to calibrate against e g NEE or latent heat flux depending on the calibration technique employed Some techniques may allow a missing time step flag for example Observations of model states particularly soil moisture and soil temperature or at very least enough data for a spin up period see Section 7 are essential We have identified some single site flux tower datasets which we feel adequately meet these criteria Each of these sites has more than a year of half hourly or hourly data These are Veg type latitude longitude country Annual rain cropland 40 0 N 88 18 W USA IL 760 Little Washita grassland 34 58 N 97 59 W USA OK 830 Metolius coniferous 44 30 N 121 37 W USA OR 710 coniferous 50 58 N 13 38 E Germany 820 eucalypt 35 39 S 148 09 E Aus NSW 1000 alker Branch mixed forest 35 58 N 84 17 W USA TN 1350 coniferous 50 09 N 11 52 E Germany 890 Note that data from three of these site are provided with the basic CABLE package More detail about the sites can be obtained from http public ornl gov ameriflux site select cfm http www unitus it dipartimenti disafri progetti eflux euromap2 html http www dar csiro au lai ozflux monitoringsites tumbarumba index htm Automated C
25. ile the value for single site grid cell 9 CABLE user guide v1 3 simulations is assumed to be local Single site simulation time values will be read as GMT if and only if a coordinate field is present for the time variable and set to be GMT This time coordinate system will be reported in the log file 2 The met data file name is set in the cable nml namelist file 3 Multiple site regional global simulations can use either an x y grid input file or a land only compressed grid type input file For the x y grid an x and a y dimension must be present even if the simulation is only a single site gridpoint Additionally single precision variables named latitude and longitude or nav_lat and nav_lon if using ALMA formatting both dependent on the x and y dimensions only must be present Both sea and land points may be included by using the mask variable discussed below For the single dimension land only compression by gathering grid see http www lmd jussieu fr polcher ALMA dataformats html a single spatial dimension is used An example of the netcdf header from such a file is below dimensions tstep UNLIMITED 1461 currently land 15238 y 180 x 360 variables float SWdown tstep land SWdown axis TYX SWdown units W m 2 SWdown long name Surface incident shortwave radiation SWdownzassociate time nav_lat nav_lon SWdown missing value 1 e 20f int land
26. imulation period for a single site data set Results are shown for net radiation latent heat sensible heat and net ecosystem exchange of CO with separate plots for each season Both CABLE output and observations in the met forcing file in the sample data sets will be plotted Currently assumes that the data set begins Jan 1 It will generate sixteen plots on a single page Output can be chosen to go to screen pdf postscript png or jpeg seasonal R plots the average seasonal cycle monthly averages over the entire simulation period for a single site data set Results are shown for net radiation latent heat sensible heat and net ecosystem exchange of CO2 Both CABLE output and observations in the met forcing file in the sample data sets will be plotted Currently assumes that the data set begins Jan 1 It will generate four plots on a single page Output can be chosen to go to screen pdf postscript png or jpeg averagingwindow R plots RMSE CABLE vs observed regression gradient and correlation coefficient for for a range of averaging window sizes e g RMSE for daily average values 1 on x axis weekly average values 7 on x axis and monthly average values 30 on x axis Results are shown for net radiation latent heat sensible heat and net ecosystem exchange of CO It will generate twelve plots on a single page Output can be chosen to go to screen pdf postscript png or jpeg e timeseries R plots many of
27. inup TRUE in the namelist file and the avPrecip variable is found precipitation values read from the meteorological data input file will be rescaled to match this value for the duration of the spin up only If information of rainfall patterns in the pre simulation period is available avPrecip should be set appropriately In addition CABLE like any other natural system model produces biases in its input state mapping and so the states are likely to drift from true values Restart files The offline netcdf driver allows for the use of netcdf restart files That is a file may be read in before model simulation which contains initial state values for the variables above as well as a complete set of CABLE s parameters The driver will by default look for a restart file whose name and path is set in the cable namelist file cable nml using the filename_restart_in variable If no such file exists the default and rather crude initial values read in through cable_parameters f90 will be used To create a restart file simply set the logical variable output restart to true then when a run finishes the final states and parameters used will be written to a restart file The name of the output restart file is set by the filename_restart_out variable in the namelist file This facility is useful in conjunction with the spin up facility above if output restart TRUE for the spin up simulation when states have converged a restart file wi
28. ity of sampling Using this sampling perform a simulation for every possible combination of parameter values This now means that analysis of model results must deal with density functions of simulated variables rather than a single value a trickier evaluation task and extremely expensive computationally The density functions do however represent a more honest exposition of parameter uncertainty An alternative netcdf driver is available which allows for a simple exploration of this approach It will read data from a single site netcdf file and subject to user modification generate and ensemble from many alternative parameter sets as described above Contact Gab Abramowitz gabsun gmail com 10 Viewing CABLE s output and judging performance Several example scripts for plotting CABLE s netcdf output are provided which use the R package free and multi platform In the first instance CABLE s netcdf output can be perused with ncBrowse This program is available free for most platforms and does not require any programming or scripting knowledge it s simply a matter of double clicking on the CABLE output file The netcdf output files are also formatted to allow reading by GrADS R and Matlab although to read netcdf in Matlab you ll need the Matlab netcdf interface installed The plotting scripts we provide for R are very simple and easy to modify e diurnal R plots the average diurnal cycle hourly averages over the entire s
29. ll be written saving the need to spin up again if a repeat run is required Restart files also store CABLE parameter information so that if a combination of prescribed and default parameters was used in the initial run see section on parameters above when the restart file is loaded for the new run the availability of these parameters may negate the need to load the default grid saving time 8 Input and output If you intend to use CABLE with the text based offline driver or coupled to an atmospheric model the mode of input and output is essentially your own choice Since the safest execution of CABLE involves using producing a netcdf meteorological data file we discuss this first Details about the netcdf met data file The netcdf format used here broadly conforms to the ALMA standard 1 It must contain a double precision real time variable dependent on a single time dimension name not important which is the time value in seconds of each time step since the starting time value The starting time value is obtained from the units field for the time variable and is of the form seconds since 2001 02 22 00 01 00 It is not essential that this start time be the start time of the simulation e g the first value of the time variable may be 86400 in which case the actual start time might be 2001 02 22 00 01 00 86400 seconds i e 2001 02 23 00 01 00 The time value for regional or global simulations is assumed to be GMT wh
30. ly to be appropriate for automated parameter estimation in most circumstances Ensuring the appropriate parameters are used for an individual application is the user s responsibility name type dimension units ranges description albsoil real mp 0 1 0 4 snow free shortwave soil reflectance fraction bch real mp 4 15 parameter b in Campbell equation clay real mp 0 1 fraction of soil which is clay CSS real mp kJ kg K 800 2500 soil specific heat capacity hyds real mp m s 10 10 hydraulic conductivity saturation isoilm int mp 1 9 Soil type ratecs real mp 2 l year 0 001 5 0 soil carbon pool rate constant rhosoil real mp kg Im 500 2500 soil bulk density rs20 real mp 5 0 01 10 NOT CURRENTLY USED soilrespiration scaler sand real mp 0 1 fraction of soil which is sand sfc real mp m m 0 1 0 5 vol H20 field capacity silt real mp 0 1 fraction of soil which is silt ssat real mp m m 0 1 0 6 vol H2O saturation sucs real mp m 0 7 0 1 suction at saturation swilt real mp m m 0 05 0 4 vol H2O wilting ZSe real mp ms m currently fixed thickness of each soil layer CABLE user guide v1 3 canst real mp mm LAI 0 08 0 12 max intercepted water by canopy dleaf real mp m 0 005 0 2 characteristic length of leaf ejmax real mp m
31. mpilers Mac OSX g95 ifort linux unix Intel ifort SUNws f95 Lahey Fujitsu 1f95 g95 and NEC SX6 4 General considerations when running CABLE Whether you intend to run CABLE offline or online there are five broad areas where problems may arise a Model parameters CABLE has a lengthy list of parameters which describe the soil and vegetation characteristics of the region s it simulates These parameters should be set to best represent the conditions of your simulation A full list of these including their definitions and variable names within CABLE s code is given in Section 6 below A subroutine is provided to read default values of CABLE s parameters from a coarse global grid based on latitude and longitude more detail in Section 6 These are NOT appropriate for all applications b Initialisation a collection of model states e g soil moisture and soil temperature need to be given initial values A full list of these and a discussion of the provided default initialisation routine is given in Section 7 It is recommended that in addition to prescribing initial values for model states users spin up the model states using at least one year of input data until they reach equilibrium The netcdf driver has a facility for setting spin up sensitivity and the use of restart file More detail is given in Section 7 c Meteorological input CABLE s input is a collection of meteorological variables These need to be read in to CABL
32. ntil its internal states stabilise To this end the netcdf driver namelist file has a logical variable which may be set to TRUE for a limited automatic spin up The driver will then run the specified filename_met file repeatedly until soil moisture and soil temperature stabilise Stabilisation is defined by the delsoilM and delsoilT variables set in the same namelist file They define the allowed variation in each of these two variables in any soil layer in the final time step between consecutive spin up runs of filename _met When these variables stabilise the data set will be run one more time and output written Note that in a regional or global simulation this may take some time There are several reasons why a model spin up may not produce reasonable initial states Firstly the input data set period may not be representative Ideally the model would spin up using observed meteorological forcing data from the period preceding the simulation period that the states be as realistic as possible Using the same data set for spin up and simulation could mean that the soil moisture initialisation for example is too dry when the data set represents drought years To try to avoid this problem the netcdf driver looks for an avPrecip variable in the meteorological input file This variable has no time dependence it has a single value for each grid cell and represents the average precipitation value for the particular grid cell site in mm If sp
33. ol m s le 5 3e 4 max pot electron transport rate top leaf frac4 real mp 0 1 fraction of c4 plants froot real mp ms 0 1 fraction of root in each soil layer he real mp m 0 100 height of canopy iveg int mp 1 13 Vegetation type meth int mp currently fixed method for calculation of canopy fluxes and temp ratecp real mp 3 1 year 0 02 2 0 plant carbon pool rate constant rp20 real mp 0 01 10 plant respiration scaler rpcoef real mp 1 C 0 03 0 2 NOT CURRENTLY USED temperature coef for non leaf plant respiration shelrb real mp 0 5 2 0 sheltering factor tminvj real mp C 5 10 min temperature of the start of photosynthesis tmaxvj real mp C 15 40 max temperature of the start of photosynthesis vbeta real mp 1 20 stomatal sensitivity to soil water vcmax real mp mol m s 5e 6 1 5e 4 maximum RuBP carboxylation rate top leaf xfang real mp 1 0 1 0 leaf angle parameter za real mp m 1 100 reference height lowest level of atmospheric model CABLE s parameters can be loaded from one of three sources the meteorological forcing file a restart file see Initialisation section or using values extracted from a coarse global grid based on soil and vegetation types provided It is recommended that users at least specify correct soil and vegetation types for their site s of simulation ideally derive their own parameter sets for all of the above variables for all but global simulation
34. p 3 3 soil layers ssoil ssdn snow density per layer kg m dimension mp 3 ssoil smass snow mass per layer kg m dimension mp 3 ssoilYorunoff runoff total subsurface surface runoff ssoilYornofl surface runoff mm timestepsize ssoilYornof2 deep drainage mm timestepsize ssoil ortsoil turbulent resistance for soil canopy ga ground heat flux W m canopy dgdtg derivative of ground heat flux wrt soil temp canopy fev latent heat flux from vegetation W m canopy ofes latent heat flux from soil W m canopy fhs sensible heat flux from soil W m ssoil albsoilsn albedo of soil snow dimension mp 3 3 radiation bands ssoilYowb soil moisture dimension mp 6 ssoilMtgg soil temperature dimension mp 6 bgc cplant plant carbon g C m dimension mp ncp bgc csoil soil carbon g C m dimension mp ncs bal wbtot0 bal osnowd0 and the following cumulative variables need to be initialised at 0 0 sum_flux sumpn 0 0 sum_flux sumrp 0 0 sum_flux sumrpw 0 0 sum_flux sumrpr 0 0 sum_flux sumrs 0 0 sum_flux sumrd 0 0 sum_flux dsumpn 0 0 sum_flux dsumrp 0 0 sum_flux dsumrd 0 0 CABLE user guide v1 3 bal precip_tot 0 0 bal rnoff tot 0 0 bal evap_tot 0 0 bal wbal_tot 0 0 bal ebal_tot 0 0 balY odrybal 0 0 bal wetbal 0 0 Model spin up For most purposes CABLE should have a spin up period forced with a year or several years of meteorological data u
35. s Ideally one would have enough information about the sites region being simulated to provide reasonable values for the parameters listed above This is often not the case however In deciding the parameter values for any given grid cell in the first instance we suggest using the appropriate soil and vegetation type shown below for that grid cell Then adjust those parameters for which site specific information is available Care needs to be taken to avoid inconsistencies e g in the relationship between wilting point field capacity and soil saturation values or soil silt and sand fractions Also note that the reference height za is not prescribed by either vegetation or soil types and must be set appropriately Once this best guess parameter set is defined we suggest adding it to the meteorological data file as described below Prescribing CABLE s parameters To prescribe CABLE s parameters they need to be added present in the meteorological forcing file or the restart file see Initialisation section As the restart file may be overwritten we suggest the best method is to include the parameters in the met file Any restart file generated using this met file will then inherit these parameter values When the netcdf driver loads parameters it gives first preference to any values found in the met file then values the restart file and lastly to the default values described below if no restart file is present Any restart fil
36. stating that its use will not be used for commercial purposes obtaining a password to the cable user site and downloading the code getting CABLE to compile using a Fortran 90 compiler and netcdf distribution on your own machine and running with default datasets to compare against operating benchmark runs Since the CABLE community is relatively small please pass on any bug fixes or genuine improvements to the code by contacting the core CABLE team through the CABLE website 1 Files in the basic CABLE package The set of files should look something like Core code cable_carbon f90 cable_soilsnow f90 cable_cbm f90 cable_variables f90 Drivers and i o cable_driver f90 cable_drivertxt f90 cable_parameters f90 cable_input f90 cable_output f90 cable_checks f90 cable_output_text f90 cable nml cabletxt nml Makefile Makefile Documentation CABLE _userguide pdf End_user_licence_terms pdf READMEv1 3 Surface data surface_data LAI Monthly Global nc surface data def_soil_params txt surface _data def_ veg params txt surface_data VegSoil_Type_Global txt Sample data sets sample_met Tumbarumba nc sample_met Bondville nc sample_met Tharandt nc sample_met Tumbarumba txt Plotting scripts testing diurnal R testing seasonal R testing timeseries R testing averagingwindow R testing CABLEplots R testing generalTest R 2 Driving CABLE Using CABLE offline There are two example offline drivers provided to help run CABLE one wit
37. t SWdown Downward shortwave W m LWdown Downward longwave W m Rainf mm s Tair K Qair spec humidity Wind m s Psurf hPa CO2air ppmv output flux Qle latent heat W m Qh sensible heat W m Qg ground heat W m Qs surface runoff mm s Qsb subsurface runoff mm s Evap total evapotranspiration mm s Ecanop Wet canopy evaporation mm s Tveg Vegetation transpiration mm s ESoil Evaporation from soil mm s HVeg Sensible heat from vegetation W nr HSoil Sensible heat from soil W m NEE net ecosystem exchange of CO2 umol m s 11 CABLE user guide v1 3 output soil SoilMoist average soil moisture per layer kg m SoilTemp average soil temperature per layer K BaresoilT bare soil temperature K ESoil Evaporation from soil mm s HSoil Sensible heat from soil W m outputYosnow SWE soil water equivalent mm or kg m SnowT snow surface temperature K SnowDepth m output oradiation SWnet Net absorbed shortwave radiation W m LWnet Net absorbed longwave radiation W m Rnet Net absorbed radiation W m Albedo RadT surface radiative temperature K output oveg VegT average vegetation temperature K CanopInt canopy water storage mm LAI Ecanop Wet canopy evaporation mm s Tveg Vegetation transpiration mm s HVeg Sens
38. y subroutines contained within a module are shown in bold type below the module name The main subroutine for CABLE subroutine cbm lies within the cbm_module shown in green roughness radiation canopy soil and snow routines are called from it cable_carbon f90 17
Download Pdf Manuals
Related Search