readme.txt

超强的大尺度水文模拟工具

Arc/info soil parameter files and ARNO soil parameters

	Files affected:
	read_soilparam_arc.c

	Description:
	VIC 4.0.4 does not handle ARNO soil parameters in arc/info soil files.
	VIC 4.0.5 now handles them in all soil files (assuming ARC_SOIL is
	TRUE).


Soil thermal node calculations

	Files affected:
	soil_conduction.c

	Description:
	set_node_parameters(): Modified to correct differences between
	calculations to determine maximum node moisture and node moisture, so
	that nodes on the boundary between soil layers are computed the same
	way for both. (port from 4.1.0 beta)
	distribute_node_moisture_properties(): Modified to check that node soil
	moisture is less than or equal to maximum node soil moisture, otherwise
	an error is printed to the screen and the model exits. (port from 4.1.0
	beta)


Canopy evaporation and distributed precipitation

	Files affected:
	surface_fluxes.c

	Description:
	Fixed initialization of canopyevap to initialize for every value of
	dist, rather than just dist 0. (found by Justin Sheffield at Princeton)


Output debug file error

	Files affected:
	write_atmosdata.c

	Description:
	No longer close the debug file, since the next cell must write to it.
	(found by Justin Sheffield at Princeton)


Calculation of deltaH when FS_ACTIVE is FALSE

	Files affected:
	func_surf_energy_bal.c

	Description:
	Added check that both FS_ACTIVE and FROZEN_SOIL are true before
	adjusting *deltaH.  This is just a safety measure; ice and ice0 should
	both be 0 when FS_ACTIVE is FALSE.  (found by Justin Sheffield at
	Princeton)


Root_brent error message clarification

	Files affected:
	calc_surf_energy_bal.c, frozen_soil.c, root_brent.c, snow_melt.c,
	vicNl.h

	Description:
	Instead of printing error messages, root_brent.c passes error messages
	to the functions that called it, leaving it to them to describe the
	specific error and its consequences.  In 4.0.4, root_brent's messages
	sometimes were wrong.


Statefile name change

	Files affected:
	open_state_file.c

	Description:
	Modified the statefile name to contain year, month, day rather than
	day, month, year.  This makes it consistent with the planned release
	of 4.1.0. (port from 4.1.0 beta)


Display current grid cell number for arc/info soil files

	Files affected:
	read_soilparam_arc.c

	Description:
	In 4.0.4, for arc/info-format soil files, the current grid cell is
	not displayed (for regular-format soil files, it is displayed).  Now
	the current grid cell number is always displayed.


Fluxes file output format (ascii)

	Files affected:
	write_data.c

	Description:
	In 4.0.4, some fields in the ascii-format fluxes output file were not
	separated by white space when values became large.  This has been
	fixed.


Binary file opening message

	Files affected:
	open_file.c

	Description:
	In 4.0.4, the message announcing the opening of a binary input file
	for reading was truncated.  This has been fixed.


--------------------------------------------------------------------------------

December 1, 2003: VIC release 4.0.4

This release covers fixes for a small number of additional bugs discovered
since the release of 4.0.4 beta r3, plus 2 new features.

New Features:

      - Typing 'vicNl -v' on the command line now displays the version
	of VIC you are running.  This provides a fast, simple way of
	determining the version of VIC without running a simulation.

      - The parameters {Ds, Dsmax, Ws and c} in the soil parameter file
	may now be replaced by the set of ARNO baseflow parameters
	{d1, d2, d3, d4}.  To enable VIC to read the ARNO baseflow
	parameters, you must add the following line to the global
	parameter file:
		ARNO_PARAMS TRUE
	and replace {Ds, Dsmax, Ws, and c} with {d1, d2, d3, and d4}
	in the soil parameter file.

	To disable this behavior, your soil file should contain
	{Ds, Dsmax, Ws, and c} and your global parameter file should
	contain:
		ARNO_PARAMS FALSE
	or you may omit the ARNO_PARAMS line entirely.

Modifications:

	read_vegparam.c:
	    Applied Alan Hamlet's fix for COMPUTE_TREELINE option,
	    which fixed a segmentation fault when COMPUTE_TREELINE=TRUE.
	    This consisted of removing the call to realloc and
	    instead allocating an extra veg class to begin with,
	    as well as assigning this extra veg class a very small
	    fraction of the grid cell's area to avoid changing the
	    results for areas below the treeline.		TJB
	runoff.c:
	    This fixed strange behavior (spikes) in bottom-layer
	    soil moisture and baseflow for low liquid moisture
	    conditions (e.g.  frozen soils).  Changed calculation
	    of dt_baseflow to go to zero when soil liquid moisture
	    <= residual moisture.  Changed block that handles case
	    of total soil moisture < residual moisture to not allow
	    dt_baseflow to go negative.				TJB
	write_model_state.c, read_initial_model_state.c:
	    (bugs found by Lifeng Luo at Princeton)
	    Read_initial_model_state was not parsing ASCII state
	    files according to the format written by
	    write_model_state.  Modified write_model_state.c to
	    add a "\n" after writing mu.  Modified
	    read_initial_model_state to loop over tmp_Nveg and
	    tmp_Nband when searching for the desired cellnum, and
	    to read the correct number of lines per veg class
	    once the desired cellnum has been found.		TJB
	vicNl_def.h, get_global_param.c, read_soilparam.c:
	    Added the ARNO_PARAMS option.  If TRUE, read_soilparam
	    reads the ARNO baseflow parameters d1, d2, d3, and d4
	    and converts them to Ds, Dsmax, Ws, and c.		TJB
	snow_utility.c:
	    In the snow_density function, modified the checks on
	    delta_depth so that the condition is
	    delta_depth > MAX_CHANGE*depth
	    Modified compression due to aging to only be calculated
	    if depth > 0.					TJB
	global.h, vicNl.c, cmd_proc.c:
	    Modified the version information banner to display the
	    version string defined in global.h.  Added -v option
	    to display version information as well.		TJB

--------------------------------------------------------------------------------

September 5, 2003: VIC release 4.0.4beta r3

	snow_utility.c: modification by KAC
	    Added check to keep compression from aging from exceeding
	    the actual depth of the snowpack.
	get_global_param.c, initialize_atmos.c, read_vegparam.c,
	vicNl_def.h: modification by KAC
	    Moved COMPUTE_TREELINE flag from user_def.h to the
	    options structure.  Now when not set to FALSE, the
	    value indicates the default above treeline vegetation
	    if no usable vegetation types are in the grid cell
	    (i.e. everything has a canopy).  A negative value
	    will cause the model to use bare soil.  Make sure that
	    positive index value refer to a non-canopied vegetation
	    type in the vegetation library.

	    To activate treeline calculation, add the following to
	    your global control file:
		COMPUTE_TREELINE n
	    If n < 0, the default above treeline vegetation is
	    set to bare soil.  If n >= 0, the default vegetation type
	    is set to type n in the vegetation library.
	    To deactivate treeline calculation, add the following or remove
	    the line from the global control file:
		COMPUTE_TREELINE FALSE

	    WARNING #1: Since the model does not store default root zone
	    distributions, the default vegetation type will use the
	    values from the last defined vegetation type for the current
	    grid cell (i.e. veg type N-1, prior to the addition of the
	    new vegetation type).

	    WARNING #2: If you are using GLOBAL_LAI, than the LAI and
	    dew storage terms for the default vegetation type will be set
	    to the values used by the last occurrence of the default
	    vegetation type.

	write_model_state.c: bug found by AWW and KAC
	    Modified to print space before dz_node for ASCII state
	    file, this corrects a problem with state files created
	    for models using the Cherkauer and Lettenmaier (1999) heat
	    flux formulation.

--------------------------------------------------------------------------------

June 4, 2003: VIC release 4.0.4beta r2

This covers bugs found during tests with the snow algorithm.

	solve_snow.c: Bug found by KAC
	    Counter for number of days since last snowfall was
	    incremented twice in the MELTING update.  This has been
	    fixed.
	solve_snow.c: modification by KAC
	    Added check so that MELTING flag is only TRUE if melt
	    occurs in the melt season - currently this is defined
	    between March 1 and October 1.  Otherwise the MELTING
	    flag can trigger rapid very early season melt
	write_model_state.c, read_initial_model_state.c, open_state_file.c,
	check_state_file.c: Modification by KAC
	    Modified to handle both ASCII and BINARY state files.
	    NOTE: ASCII mode is designed to make it easier to create
	    a state file for initializing a point model.  It includes
	    all features of the Binary state file, but values are
	    truncated so the resulting restart will not be identical.
	    If you need an exact restart, use the Binary files.  Also
	    removed ice content from the state file as it is computed
	    at the begining of each time step, so storing its value
	    is unnecessary.

--------------------------------------------------------------------------------

April 23, 2003: VIC release 4.0.4beta r1

This covers bug fixes found by beta testers and fixed in the version of
the code bundled with this file.

	surface_fluxes.c: (found by Ingjerd Haddeland)
	    Indexing fix sent SNOW_STEP to calc_surf_energy_bal rather
	    than the model time step, meaning that without snow the
	    evaporation was computed for SNOW_STEP hours rather than a
	    full day.  This was fixed by introducing step_inc to
	    index the arrays, while step_dt keeps track of the correct
	    time step.

--------------------------------------------------------------------------------

March 27, 2003: VIC release 4.0.4beta

This release covers patches for several bugs found in 4.0.3, which were 
never formally released (i.e. the downloadable source code was modified, 
but no major announcement was made).  It also includes other fixes and 
modifications that have been identified as being needed prior to releasing 
version 4.1.0, which will involve several significant changes (including 
lakes & wetlands, spatial snow & frost, and a closed canopy energy balance).

Modifications:

	Snow albedo update: (found by Keith)
	     In previous releases, the snow albedo function has been hyper 
	     sensitive to trace amounts of snowfall during the melt period.  
	     Whenever new snow falls or the cold content falls below 0, the 
	     albedo is switched from the ablation curve to the accumulation 
	     curve.  This curve is then followed until the cold content 
	     exceeds 0, indicating it is in the spring melt season.  This is 
	     fine when accounting for thin early season snowpacks or mid-
	     season melt events, however, a cold snap or light dusting of 
	     snow should not reset the snowpack albedo to much higher winter 
	     values for days or weeks at a time.  This release of the model 
	     monitors the state of pack with the variable MELTING.  This 
	     flag keeps the snowpack on the ablation curve unless new snow-
	     fall exceeds a threshold (TraceSnow in user_def.h) indicating 
	     that the top of the snowpack should be represented by the albedo 
	     of the new snow.
	Model initialization: (found by Ulysses and Keith)
	     In previous releases, the initialization of soil layer moist[] 
	     and ice[] was within a second set of loops for band and 
	     vegetation, using the same counters.  Because of the dual use
	     of counters, initialization was not completed correctly.  The
	     primary effect of this was that thermal node values beyond the
	     first vegetation type were not correctly initialized, which 
	     caused the model to crash during some simulations with full
             energy or frozen soils active.  Without frozen soil, most 
	     simulations would compensate for the problem within their
	     spin-up time, so it is unlikely that this bug impacts any 
	     simulations not employing frozen soil.
	Snow time step: (found by Andy, et al.)
	     The snow algorithm needs to run sub-daily for the energy balance
	     components to function properly.  This means that for daily 
	     simulations, the snow model must be solved at a finer (sub-daily)
	     time step.  In the previous release, initialize_atmos.c stored
	     sub-daily met data in each days variable using positions (e.g. 
             0,1,..8 for 3 hour data).  In surface_fluxes.c the model indexed 
	     the sub-daily time steps used by the snow algorithm with hours 
	     (e.g. 0,3,6,...21 for three hour data).  This means the arrays
	     were incorrectly indexed and the resulting model simulations 
	     would be wrong.  The fix implemented here has been tested under 
	     several model configurations and is deemed the official version.
	     WARNING: There are several versions of this fix circulating,
	     please update your code to this version - the previous fixes
	     may not work in all circumstances!
	FROZEN_SOIL active flag: (found by Ed and Justin)
	     The cause of the problem is a bug in the code that occurs when 
	     the global frozen soils flag (FROZEN_SOILS) is set to true but 
	     the individual cell frozen soil flag (FS_ACTIVE) is set to 
	     false. This causes the change in soil heat storage to be