r.agropast.adaptive

#!/usr/bin/python3
#
############################################################################
#
# MODULE:               r.agropast.semiadaptive.7.0.5
#
# AUTHOR(S):            Isaac Ullah, San Diego State University
#
# PURPOSE:              Simulates agricultural and pastoral landuse and tracks 
#                       yields and environmental impacts. Farming and grazing
#                       strategies and yield goals are predetermined by the 
#                       researcher, and do not change (adapt) during the 
#                       simulation. However, catchment sizes can be adapted 
#                       over time to meet these goals. This version implments 
#                       a land tenuring alogrithm. Requires r.landscape.evol.
#
# ACKNOWLEDGEMENTS:     National Science Foundation Grant #BCS0410269, Center 
#                       for Comparative Archaeology at the University of 
#                       Pittsburgh, Center for Social Dynamics and Complexity 
#                       at Arizona State University, San Diego State University
#
# COPYRIGHT:            (C) 2016 by Isaac Ullah, San Diego State University
#
#                   This program is free software under the GNU General Public
#                   License (>=v2). Read the file COPYING that comes with GRASS
#                   for details.
#
#############################################################################


#%Module
#%  description: Simulates agricultural and pastoral landuse and tracks yields and environmental impacts. Basic farming and grazing strategies and yield goals are predetermined by the researcher, and do not change (adapt) during the simulation, but it is possible for population to change based on returns. This version implments a land tenuring alogrithm. Requires r.landscape.evol. Note that some stats files will be written to current mapset, and will be appended to if you run the simulation again with the same prefix.
#%END

##################################
#Simulation Control
##################################
#%option
#% key: years
#% type: integer
#% description: Number of iterations ("years") to run
#% answer: 200
#% required: yes
#% guisection: Simulation Control
#%END
#%option
#% key: prfx
#% type: string
#% description: Prefix for all output maps
#% answer: sim_
#% required: yes
#% guisection: Simulation Control
#%END

##################################
#Agent Properties
##################################
#%option
#% key: numpeople
#% type: double
#% description: Number of people in the village(s) (starting population size with flag -p, otherwise stays constant throughout the simulation)
#% answer: 120
#% guisection: Agent Properties
#%END
#%option
#% key: birthrate
#% type: double
#% description: Per-capita human fecundity (only active with flag -p)
#% answer: 0.054
#% guisection: Agent Properties
#%END
#%option
#% key: deathrate
#% type: double
#% description: Per-capita human mortality hazard (only active with flag -p)
#% answer: 0.04
#% guisection: Agent Properties
#%END
#%option
#% key: starvthresh
#% type: double
#% description: Starvation threshold. If returns are below this percentage of the normal subsistence needs, people bcome "resource-starved." No births will occur, but deaths will still happen. (only active with flag -p)
#% answer: 0.6
#% guisection: Agent Properties
#%END
#%option
#% key: agentmem
#% type: integer
#% description: Length of the "memory" of the agent in years. The agent will use the mean surplus/defict information from this many of the most recent previous years when making a subsistence plan for the current year.
#% answer: 5
#% guisection: Agent Properties
#%END
#%option
#% key: aglabor
#% type: double
#% description: The amount of agricultural labor an average person of the village can do in a year (in "person-days")
#% answer: 300
#% guisection: Agent Properties
#%END
#%option
#% key: cerealreq
#% type: double
#% description: Amount of cereals that would be required per person per year if cereals were the ONLY food item (Kg)
#% answer: 370
#% guisection: Agent Properties
#%END
#%option
#% key: animals
#% type: double
#% description: Number of herd animals that would be needed per person per year if pastoral products were the ONLY food item
#% answer: 60
#% guisection: Agent Properties
#%END
#%option
#% key: fodderreq
#% type: double
#% description: Amount of fodder required per herd animal per year (Kg)
#% answer: 680
#% guisection: Agent Properties
#%END
#%option
#% key: a_p_ratio
#% type: double
#% description: Actual ratio of agricultural to pastoral foods in the diet (0.01-0.99, where smaller numbers are mostly agricultural, and larger numbers are mostly pastoral)
#% answer: 0.20
#% options: 0.01-0.99
#% guisection: Agent Properties
#%END
#%option G_OPT_R_MAP
#% key: costsurf
#% description: Map of movement costs from the center of the agricultural/grazing catchments (from r.walk or r.cost).
#% guisection: Agent Properties
#%END
#%flag
#% key: p
#% description: -p Allow the population to vary over time, according to subsistence returns
#% guisection: Agent Properties
#%end


##################################
#Farming Options
##################################
#%option G_OPT_R_MAP
#% key: agcatch
#% description: Map of the largest possible agricultural catchment map (From r.catchment or r.buffer, where catchment is a single integer value, and all else is NULL)
#% guisection: Farming
#%END
#%option
#% key: agmix
#% type: double
#% description: The wheat/barley ratio (e.g., 0.0 for all wheat, 1.0 for all barley, 0.5 for an equal mix)
#% answer: 0.25
#% options: 0.0-1.0
#% guisection: Farming
#%END
#%option
#% key: nsfieldsize
#% type: double
#% description: North-South dimension of fields in map units (Large field sizes may lead to some overshoot of catchment boundaries)
#% answer: 20
#% guisection: Farming
#%END
#%option
#% key: ewfieldsize
#% type: double
#% description: East-West dimension of fields in map units (Large field sizes may lead to some overshoot of catchment boundaries)
#% answer: 50
#% guisection: Farming
#%END
#%option
#% key: fieldlabor
#% type: double
#% description: Number of person-days required to till, sow, weed, and harvest one farm field in a year.
#% answer: 50
#% guisection: Farming
#%END
#%option
#% key: farmval
#% type: double
#% description: The landcover value for farmed fields (Should correspond to an appropriate value from your landcover rules file.)
#% answer: 5
#% guisection: Farming
#%END
#%option
#% key: farmimpact
#% type: double
#% description: The mean and standard deviation of the amount to which farming a patch decreases its fertility (in percentage points of maximum fertility, entered as: "mean,std_dev"). Fertility impact values of indvidual farm plots will be randomly chosen from a gaussian distribution that has this mean and std_dev.
#% answer: 3,2
#% multiple: yes
#% guisection: Farming
#%END
#%option
#% key: maxwheat
#% type: double
#% description: Maximum amount of wheat that can be grown (kg/ha)
#% answer: 1750
#% guisection: Farming
#%END
#%option
#% key: maxbarley
#% type: double
#% description: Maximum amount of barley that can be grown (kg/ha)
#% answer: 1250
#% guisection: Farming
#%END
#%option
#% key: tenuretype
#% type: string
#% description: Choose the land tenuring strategy: "None" (Fields are chosen at random each year), "Satisficing" (First year's fields are chosen at random. Fields are tenured, but some may be randomly dropped or added to meet the minimum need), "Maximizing" (Same as "satisficing", but tenured fields are only dropped if production falls below the threshold defined by "tenuredrop," not according to minimum need.)
#% answer: Maximize
#% options: None,Maximize,Satisfice
#% guisection: Farming
#%END
#%option
#% key: tenuredrop
#% type: double
#% description: Threshold for dropping land out of tenure (with tenuretype as "Maximize"). If the value is 0, fields that yield less than the mean of all fields are dropped. If the value is greater than 0, then fields that are lower than that percentage of the maximum yield of all fields will be dropped.
#% answer: 0.1
#% options: 0.0-1.0
#% guisection: Farming
#%END

##################################
#Grazing Options
##################################
#%option G_OPT_R_MAP
#% key: grazecatch
#% description: Map of the largest possible grazing catchment (From r.catchment or r.buffer, where catchment is a single integer value, and all else is NULL).
#% guisection: Grazing
#%END
#%option
#% key: mingraze
#% type: double
#% description: Minimum amount of vegetation on a cell for it to be considered grazeable (in value of succession stages, matching landcover rules file)
#% answer: 2
#% guisection: Grazing
#%END
#%option
#% key: grazespatial
#% type: double
#% description: Spatial dependency of the grazing pattern in map units. This value determines how "clumped" grazing patches will be. A value close to 0 will produce a perfectly randomized grazing pattern with patch size=resolution, and larger values will produce increasingly clumped grazing patterns, with the size of the patches corresponding to the value of grazespatial (in map units).
#% answer: 50
#% guisection: Grazing
#%END
#%option
#% key: grazepatchy
#% type: double
#% description: Coefficient that determines the patchiness of the grazing catchemnt. Value must be non-zero, and usually will be <= 1.0. Values close to 0 will create a patchy grazing pattern, values close to 1 will create a "smooth" grazing pattern. Actual grazing patches will be sized to the resolution of the input landcover map.
#% answer: 1.0
#% guisection: Grazing
#%END
#%option
#% key: maxgrazeimpact
#% type: integer
#% description: Maximum impact of grazing in units of "landcover.succession". Grazing impact values of individual patches will be chosen from a gaussian distribution between 1 and this maximum value (i.e., most values will be between 1 and this max). Value must be >= 1.
#% answer: 3
#% guisection: Grazing
#%END
#%option
#% key: manurerate
#% type: double
#% description: Base rate that animal dung contributes to fertility increase on a grazed patch in units of percentage of maximum fertility regained per increment of grazing impact. Actual fertility regain values are thus calculated as "manure_rate x grazing_impact", so be aware that this variable interacts with the grazing impact settings you have chosen.
#% answer: 0.03
#% options: 0.0-100.0
#% guisection: Grazing
#%END
#%option G_OPT_F_INPUT
#% key: fodder_rules
#% description: Path to foddering rules file for calculation of fodder gained by grazing
#% answer: /home/iullah/Dropbox/Scripts_Working_Dir/rules/fodder_rules.txt
#% guisection: Grazing
#%END
#%flag
#% key: f
#% description: -f Do not graze in unused portions of the agricultural catchment (i.e., do not graze on "fallowed" fields, and thus no "manuring" of those fields will occur)
#% guisection: Grazing
#%end
#%flag
#% key: g
#% description: -g Do not allow "stubble grazing" on harvested fields (and thus no "manuring" of fields)
#% guisection: Grazing
#%end

##################################
#Landcover Dynamics Options
##################################
#%option
#% key: fertilrate
#% type: double
#% description: Comma separated list of the mean and standard deviation of the natural fertility recovery rate (percentage by which soil fertility increases per year if not farmed, entered as: "mean,stdev".) Fertility recovery values of individual landscape patches will be randomly chosen from a gaussian distribution that has this mean and std_dev.
#% answer: 2,0.5
#% multiple: yes
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_R_MAP
#% key: inlcov
#% description: Input landcover map (Coded according to landcover rules file below)
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_R_MAP
#% key: infert
#% description: Input fertility map (Coded according to percentage retained fertility, with 100 being full fertility)
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_R_MAP
#% key: maxlcov
#% description: Maximum value of landcover for the landscape (Can be a single numerical value or a cover map of different maximum values. Shouldn't exceed maximum value in rules file')
#% answer: 50
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_R_MAP
#% key: maxfert
#% description: Maximum value of fertility for the landscape (Can be a single numerical value or a cover map of different maximum values. Shouldn't exceed 100)
#% answer: 100
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_F_INPUT
#% key: lc_rules
#% description: Path to reclass rules file for landcover map
#% answer: /home/iullah/Dropbox/Scripts_Working_Dir/rules/luse_reclass_rules.txt
#% guisection: Landcover Dynamics
#%END
#%option G_OPT_F_INPUT
#% key: cfact_rules
#% description: Path to recode rules file for c-factor map
#% answer: /home/iullah/Dropbox/Scripts_Working_Dir/rules/cfactor_recode_rules.txt
#% guisection: Landcover Dynamics
#%END
#%flag
#% key: c
#% description: -c Keep C-factor (and rainfall excess) maps for each year
#% guisection: Landcover Dynamics
#%end

##################################
#Landscape Evolution Options
##################################
# %option G_OPT_R_ELEV
# % key: elev
# % description: Input elevation map (DEM of surface)
# % required : no
# % guisection: Landscape Evolution
# %end
# %option G_OPT_R_MAP
# % key: initbdrk
# % description: Bedrock elevations map (DEM of bedrock)
# % answer:
# % required : no
# % guisection: Landscape Evolution
# %end
# %option G_OPT_R_MAP
# % key: k
# % description: Soil erodability index (K factor) map or constant (values <= 0.09 [t.ha.h /ha.MJ.mm])
# % answer: 0.05
# % required : no
# % guisection: Landscape Evolution
# %end
# %option G_OPT_R_MAP
# % key: sdensity
# % description: Soil density map or constant for conversion from mass to volume (values typically >=1000 [kg/m3])
# % answer: 1218.4
# % required : no
# % guisection: Landscape Evolution
# %end
# %option
# % key: transp_eq
# % type: string
# % description: The sediment transport equation to use (USPED: Tc=R*K*C*P*A^m*B^n, Stream power: Tc=Kt*gw*1/N*h^m*B^n, or Shear stress: Tc=Kt*tau^m ).
# % answer: StreamPower
# % options: StreamPower,ShearStress,USPED
# % guisection: Landscape Evolution
# %end
# %option
# % key: exp_m
# % type: string
# % description: Exponent m relates to the influence of upslope area (and thus flow depth, discharge) on transport capacity. Values generally thought to scale inversely with increasing depth of flow between the two cutoff thresholds specified: "thresh1,m1,thresh2,m2"
# % answer: 10,2,100,1
# % required : no
# % guisection: Landscape Evolution
# %end
# %option
# % key: exp_n
# % type: string
# % description: Exponent n relates to the influence of local topographic slope on transport capacity. Default values set to scale inversely with increasing local slope between the two slope cutoff thresholds specified: "thresh1,n1,thresh2,n2"
# % answer: 10,2,45,0.5
# % required : no
# % guisection: Landscape Evolution
# %end
# %flag
# % key: m
# % description: -m Apply smoothing (useful to mitigate possible unstable conditions in streams)
# % guisection: Landscape Evolution
# %end
# %option G_OPT_R_MAP
# % key: r
# % description: Rainfall (R factor) map or constant (Employed only in the USPED equation) (values typically between 500 and 10000 [MJ.mm/ha.h.yr])
# % answer: 720
# % guisection: Climate
# %end
# %option G_OPT_R_MAP
# % key: rain
# % description: Precip total for the average erosion-causing storm map (Employed in stream power and shear stress equations) (values typically >=30.0 [mm])
# % answer: 30
# % guisection: Climate
# %end
# %option G_OPT_R_MAP
# % key: storms
# % description: Number of erosion-causing storms per year map or constant (Employed in stream power and shear stress equations) (values >=0 [integer])
# % answer: 2
# % guisection: Climate
# %end
# %option G_OPT_R_MAP
# % key: stormlength
# % description: Average length of the storm map or constant (Employed in stream power and shear stress equations) (values >=0.0 [h])
# % answer: 24.0
# % guisection: Climate
# %end
# %option G_OPT_R_MAP
# % key: stormi
# % description: Proportion of the length of the storm where the storm is at peak intensity map or constant (Employed in stream power and shear stress equations) (values typically ~0.05 [unitless proportion])
# % answer: 0.05
# % guisection: Climate
# %end
# %option  G_OPT_F_INPUT
# % key: climfile
# % required: no
# % description: Path to climate file of comma separated values of "rain,R,storms,stormlength,stormi", with a new line for each year of the simulation. This option will override values or maps entered above.
# % guisection: Climate
# %end
# %option G_OPT_R_MAP
# % key: manningn
# % description: Map or constant of the value of Manning's "N" value for channelized flow. (Employed in stream power and shear stress equations) (0.03 = clean/straight stream channel, 0.035 = major river, 0.04 = sluggish stream with pools, 0.06 = very clogged streams [unitless])
# % answer: 0.03
# % required : no
# % guisection: Hydrology
# %end
# %option
# % key: convergence
# % type: integer
# % description: Value for the flow convergence variable in r.watershed. Small values make water spread out, high values make it converge in narrower channels.
# % answer: 5
# % options: 1,2,3,4,5,6,7,8,9,10
# % required : no
# % guisection: Hydrology
# %end
# %flag
# % key: d
# % description: -d Don't output yearly soil depth maps
# % guisection: Landscape Evolution
# %end
# %flag
# % key: r
# % description: -r Don't output yearly maps of the erosion/deposition rates ("ED_rate" map, in vertical meters)
# % guisection: Landscape Evolution
# %end
# %flag
# % key: s
# % description: -s Keep all slope maps
# % guisection: Landscape Evolution
# %end
# %flag
# % key: t
# % description: -t Keep yearly maps of the Transport Capacity at each cell ("Qs" maps)
# % guisection: Landscape Evolution
# %end
# %flag
# % key: e
# % description: -e Keep yearly maps of the Excess Transport Capacity (divergence) at each cell ("DeltaQs" maps)
# % guisection: Landscape Evolution
# %end

import sys
import os
import tempfile
import random
import numpy
import grass.script as grass

#new random-poisson babymaker
def babymaker(p, n): #p is the per capita birth rate, n is the population size
    babys = (numpy.random.poisson(p*100)/100.)*n
    return(int(babys))

# old random-normal babymaker
#def babymaker(p, n): #p is the per capita birth rate, n is the population size
#    babys = 0
#    for m in range(int(n)):
#        x = numpy.random.random()
#        if x < float(p):
#            babys = babys + 1
#    return(babys)

#new random-poisson deathdealer
def deathdealer(p, n): #p is the per capita death rate, n is the population size
    deaths = (numpy.random.poisson(p*100)/100.)*n
    return(int(deaths))

#old random-normal deathdealer
#def deathdealer(p, n): #p is the per capita death rate, n is the population size
#    deaths = 0
#    for m in range(int(n)):
#        x = numpy.random.random()
#        if x < float(p):
#            deaths = deaths + 1
#    return(deaths)

def climfile(d, y, years):
    """
    Check a climate variable and read in from text if needed.
    Takes a variable name to check and returns as a parsed list of variables
    If there is a climate file, values will be read in for each year.
    If not, the value entered will be applied to all years
    d = climate variable to be parsed
    y = column containing that climate variable in the input file
    years = number of years of the simulation
    """
    l = []
    try:
        d1 = float(d)
        for year in range(int(years)):
            l.append(d1)
        return l
    except:
        with open(d, "rU") as f:
            for line in f:
                l.append(line.split(",")[y])

        # Check for text header and remove if present
        try:
            float(l[0])
        except:
            del l[0]

        # Throw a warning if there aren't enough values in the column
        if len(l) != int(years):
            grass.fatal(
                "Number of rows of rainfall data in your climate file\ndo not match the number of iterations you wish to run.\n Please ensure that these numbers match and try again"
            )
            sys.exit(1)
        return l

#main block of code starts here
def main():
    grass.message("Setting up Simulation........")
    #setting up Land Use variables for use later on
    agcatch = options['agcatch']
    nsfieldsize = options['nsfieldsize']
    ewfieldsize = options['ewfieldsize']
    grazecatch = options['grazecatch']
    grazespatial = options['grazespatial']
    grazepatchy = options['grazepatchy']
    maxgrazeimpact = options['maxgrazeimpact']
    manurerate = options['manurerate']
    inlcov = options['inlcov']
    years = int(options['years'])
    farmval = options['farmval']
    maxlcov = options['maxlcov']
    prfx = options['prfx']
    lc_rules = options['lc_rules']
    cfact_rules = options['cfact_rules']
    fodder_rules = options['fodder_rules']
    infert = options['infert']
    maxfert = options['maxfert']
    maxwheat = options['maxwheat']
    maxbarley = options['maxbarley']
    mingraze = options['mingraze']
    tenuretype = options['tenuretype']
    tenuredrop = options['tenuredrop']
    costsurf = options['costsurf']
    agmix = options['agmix']
    numpeople = float(options['numpeople'])
    birthrate = float(options['birthrate'])
    deathrate = float(options['birthrate'])
    starvthresh = float(options['starvthresh'])
    agentmem = int(options['agentmem'])
    animals = float(options['animals'])
    agratio = 1 - float(options['a_p_ratio'])
    pratio =  float(options['a_p_ratio'])
    indcerreq = float(options['cerealreq'])
    indfodreq = float(options['fodderreq'])
    aglabor = float(options['aglabor'])
    fieldlabor = float(options['fieldlabor'])
    cereal_pers = numpeople * agratio
    fodder_anim = animals * numpeople * pratio
    cerealreq = indcerreq * cereal_pers
    fodderreq = indfodreq * fodder_anim
    totlabor = numpeople * aglabor
    maxfields = int(round(totlabor / fieldlabor))
    #these are various rates with min and max values entered in the gui that we need to parse
    farmimpact = list(map(float, options['farmimpact'].split(',')))
    fertilrate = list(map(float, options['fertilrate'].split(',')))
    #Setting up Landscape Evol variables to write the r.landscape.evol command later
    elev = options["elev"]
    initbdrk = options["initbdrk"]
    k = options["k"]
    sdensity = options["sdensity"]
    transp_eq = options["transp_eq"]
    exp_m = options["exp_m"]
    exp_n =  options["exp_n"]
    manningn = options["manningn"]
    convergence = options["convergence"]
    # These values could be read in from a climate file, so check that, and
    # act accordingly. Either way, the result will be some lists with the same
    # number of entries as there are iterations.
    if options["climfile"]:
        R2 = climfile(options["climfile"], 0, years)
        rain2 = climfile(options["climfile"], 1, years)
        stormlength2 = climfile(options["climfile"], 2, years)
        storms2 = climfile(options["climfile"], 3, years)
        stormi2 = climfile(options["climfile"], 4, years)
    else:
        R2 = climfile(options["r"], 0, years)
        rain2 = climfile(options["rain"], 1, years)
        stormlength2 = climfile(options["stormlength"], 2, years)
        storms2 = climfile(options["storms"], 3, years)
        stormi2 = climfile(options["stormi"], 4, years)
    #get the process id to tag any temporary maps we make for easy clean up in the loop
    pid = os.getpid()
    #we need to separate out flags used by this script, and those meant to be sent to r.landscape.evol. We will do this by popping them out of the default "flags" dictionary, and making a new dictionary called "use_flags"
    use_flags = {}
    use_flags.update({'g': flags.pop('g'), 'f': flags.pop('f'), 'c': flags.pop('c'), 'p': flags.pop('p')})
    #now assemble the flag string for r.landscape.evol'
    levol_flags = []
    for flag in flags:
        if flags[flag] is True:
            levol_flags.append(flag)
    #check if maxlcov is a map or a number, and grab the actual max value for the stats file
    try:
        maxval = int(float(maxlcov))
    except:
        maxlcovdict = grass.parse_command('r.univar', flags = 'ge', map = maxlcov)
        maxval = int(float(maxlcovdict['max']))
    #check if maxfert is a map or a number, and grab the actual max value for the stats file
    try:
        maxfertval = int(float(maxfert))
    except:
        maxfertdict = grass.parse_command('r.univar', flags = 'ge', map = maxfert)
        maxfertval = int(float(maxfertdict['max']))
    #set up the stats files names
    env = grass.gisenv()
    statsdir = os.path.join(env['GISDBASE'], env['LOCATION_NAME'], env['MAPSET'])
    textout = statsdir + os.sep + prfx + 'landcover_temporal_matrix.txt'
    textout2 = statsdir + os.sep + prfx + 'fertility_temporal_matrix.txt'
    textout3 = statsdir + os.sep + prfx + 'yields_stats.txt'
    textout4 = statsdir + os.sep + prfx + 'landcover_and_fertility_stats.txt'
    statsout = statsdir + os.sep + prfx + 'erdep_stats.txt'
    # Make color rules for landcover, cfactor, and soil fertilty maps
    lccolors = tempfile.NamedTemporaryFile(mode = "w")
    lccolors.write('0 grey\n10 red\n20 orange\n30 brown\n40 yellow\n%s green'% maxval)
    lccolors.flush()
    cfcolors = tempfile.NamedTemporaryFile(mode = "w")
    cfcolors.write('0.1 grey\n0.05 red\n0.03 orange\n0.01 brown\n0.008 yellow\n0.005 green')
    cfcolors.flush()
    fertcolors = tempfile.NamedTemporaryFile(mode = "w")
    fertcolors.write('0 white\n20 grey\n40 yellow\n60 orange\n80 brown\n100 black')
    fertcolors.flush()
    #Figure out the number of cells per hectare and how many square meters per cell to use as conversion factors for yields
    region = grass.region()
    cellperhectare = 10000 / (float(region['nsres']) * float(region['ewres']))
    #sqmeterpercell = (float(region['nsres']) * float(region['ewres']))
    #do same for farm field size
    fieldsperhectare = 10000 / (float(nsfieldsize) * float(ewfieldsize))
    #find conversion from field size to cell size
    #cellsperfield = (float(nsfieldsize) * float(ewfieldsize)) / (float(region['nsres']) * float(region['ewres']))
    #find out the maximum amount of cereals per field according to the proper mix
    #maxyield = (((1-float(agmix))*float(maxwheat))+(float(agmix)*float(maxbarley)))/fieldsperhectare
    #find out number of digits in 'years' for zero padding
    digits = len(str(abs(years)))
    #set up the agent memory
    farmingmemory = []
    farmyieldmemory = []
    grazingmemory = []
    grazeyieldmemory = []
    grass.message('Simulation will run for %s iterations.\n\n............................STARTING SIMULATION...............................' % years)
    # Before we get going on the loop, write out some basic information about the run. These can be used to remeber what the settings were for this particular run, as well as to provide some interpretation for the other stats files that will be made.
    f = open(statsdir + os.sep + prfx + 'run_info.txt', 'a')
    f.write("Variables used in the model:\ncell resolution (grazing patch size),%s\nagcatch,%s\nnsfieldsize,%s\newfieldsize,%s\ngrazecatch,%s\ngrazespatial,%s\ngrazepatchy,%s\nmaxgrazeimpact,%s\nmanurerate,%s\ninlcov,%s\nyears,%s\nfarmval,%s\nmaxfert,%s\nmaxwheat,%s\nmaxbarley,%s\nagmix,%s\nagentmem,%s\nnumpeople,%s\nanimals,%s\ncalculated agricultural ratio,%s\ncalculated pastoral ratio,%s\ncalculated cereal required per person,%s\ncalculated fodder required per animal,%s\ncalculated total cereal required,%s\ncalculated total number of animals required,%s\ncalculated total fodder required,%s\n\nFarming stats in Kg wheat and/or barley seeds per farmplot.\nGrazing stats in Kg of digestable matter per grazing plot. Note that this may also include stubble grazing if enabled." % (region['nsres'],agcatch,nsfieldsize,ewfieldsize,grazecatch,grazespatial,grazepatchy,maxgrazeimpact,manurerate,inlcov,years,farmval,maxfert,maxwheat,maxbarley,agmix,agentmem,numpeople,animals,agratio,pratio,indcerreq,fodder_anim,indfodreq,cerealreq,fodderreq)) 
    f.close()
    #Set up loop
    for year in range(int(years)):
        now = year + 1
        then = year
        if numpeople == 0:
            grass.fatal("Everybody is dead. \nSimulation stopped at year %s." % then)
        #grab the current climate vars from the lists
        rain = rain2[year]
        r = R2[year]
        storms = storms2[year]
        stormlength = stormlength2[year]
        stormi = stormi2[year]
        #figure out total precip (in meters) for the year for use in the veg growth and farm yields formulae
        precip = 0.001 * (float(rain) * float(storms))
        grass.message('_____________________________\nSIMULATION YEAR: %s\n--------------------------' % now)
        #make some map names
        fields = "%s%04d_Farming_Impacts" % (prfx, now)
        outlcov = "%s%04d_Landcover" % (prfx, now)
        outfert = "%s%04d_Soil_Fertilty" % (prfx, now)
        outcfact = "%s%04d_Cfactor" % (prfx, now)
        grazeimpacts = "%s%04d_Gazing_Impacts" % (prfx, now)
        outxs = "%s%04d_Rainfall_Excess" % (prfx, now)
        #check if this is year one, use the starting landcover and soilfertily and calculate soildepths
        if now == 1:
            oldlcov = inlcov
            oldfert = infert
            oldsdepth = "%s%04d_Soil_Depth" % (prfx, then)
            grass.mapcalc("${sdepth}=(${elev}-${bdrk})", quiet ="True", sdepth = oldsdepth, elev = elev, bdrk = initbdrk)
        else:
            oldlcov = "%s%04d_Landcover" % (prfx, then)
            oldfert = "%s%04d_Soil_Fertilty" % (prfx, then)
            oldsdepth = "%s%04d_Soil_Depth" % (prfx, then)
        #GENERATE FARM IMPACTS
        #create some temp map names
        tempfields = "%stemporary_fields_map" % pid
        tempimpacta = "%stemporary_farming_fertility_impact" % pid
        #temporarily change region resolution to align to farm field size
        grass.use_temp_region()
        grass.run_command('g.region', quiet = 'True',nsres = nsfieldsize, ewres = ewfieldsize)
        #generate the yields
        grass.message("Calculating potential farming yields.....")
        #Calculate the wheat yield map (kg/cell)
        tempwheatreturn = "%stemporary_wheat_yields_map" % pid
        grass.mapcalc("${tempwheatreturn}=eval(x=if(${precip} > 0, (0.51*log(${precip}))+1.03, 0), y=if(${sfertil} > 0, (0.28*log(${sfertil}))+0.87, 0), z=if(${sdepth} > 0, (0.19*log(${sdepth}))+1, 0), a=if(x <= 0 || z <= 0, 0, ((((x*y*z)/3)*${maxwheat})/${fieldsperhectare})), if(a < 0, 0, a))", quiet = "True", tempwheatreturn = tempwheatreturn, precip = precip, sfertil = oldfert, sdepth = oldsdepth, maxwheat = maxwheat, fieldsperhectare = fieldsperhectare)
        #Calculate barley yield map (kg/cell)
        tempbarleyreturn = "%stemporary_barley_yields_map" % pid
        grass.mapcalc("${tempbarleyreturn}=eval(x=if(${precip} > 0, (0.48*log(${precip}))+1.51, 0), y=if(${sfertil} > 0, (0.34*log(${sfertil}))+1.09, 0), z=if(${sdepth} > 0, (0.18*log(${sdepth}))+0.98, 0), a=if(x <= 0 || z <= 0, 0, ((((x*y*z)/3)*${maxbarley})/${fieldsperhectare})), if(a < 0, 0, a))", quiet = "True", tempbarleyreturn = tempbarleyreturn, precip = precip, sfertil = oldfert, sdepth = oldsdepth, maxbarley = maxbarley, fieldsperhectare = fieldsperhectare)
        #Create the desired cereal mix
        tempcerealreturn = "%stemporary_cereal_yields_map" %pid
        grass.mapcalc("${tempcerealreturn}=if(isnull(${agcatch}), null(),  (((1-${agmix})*${tempwheatreturn})+(${agmix}*${tempbarleyreturn})) )", quiet = "True", tempcerealreturn = tempcerealreturn, agmix = agmix, tempwheatreturn = tempwheatreturn, tempbarleyreturn = tempbarleyreturn, agcatch = agcatch)
        grass.message("Figuring out the farming plan for this year...")
        #gather some stats from yields maps in order to make an estimate of number of farm plots...
        cerealstats2 = grass.parse_command('r.univar', flags = 'ge', map = tempcerealreturn)
        # Grab the agent's current memory of farming yields to see what they think they need to do this year
        if len(farmyieldmemory) == 0:
            fuzzyyieldmemory = random.gauss(float(cerealstats2["mean"]), (float(cerealstats2['mean']) * 0.0333))
            fuzzydeficitmemory = -1
        else:
            if agentmem > (year -1):
                slicer = year - 1
            else:
                slicer = agentmem
            grass.debug("slicer: %s" % slicer)
            fuzzyyieldmemory = numpy.mean(farmyieldmemory[slicer:])
            fuzzydeficitmemory = numpy.mean(farmingmemory[slicer:])
        #Figure out how many fields the agent thinks it needs based on current average yield
        if fuzzyyieldmemory == 0:
            numfields = 0
        else:
            numfields = int(round(float(cerealreq) / fuzzyyieldmemory))
        grass.debug("total fields should be %s" % numfields)
        #discover how many cells to add or subtract based on agent memory of surplus or deficit
        if fuzzyyieldmemory == 0:
            fieldpad = 0
        else:
            fieldpad = int(round(fuzzydeficitmemory / fuzzyyieldmemory))
        grass.debug("fieldpad (sign reversed) %s" % fieldpad)
        #add or remove the number of extra cells calculated from agent memory surplus or deficit
        numfields = numfields - fieldpad
        grass.debug("did numfields change? %s" % numfields)
        #Then, make sure we don't exceed our catchment size or labor capacity (and reduce number of fields if necessary)
        if numfields > int(round((float(cerealstats2['cells']) - float(cerealstats2["null_cells"])))):
            numfields = int(round((float(cerealstats2['cells']) - float(cerealstats2["null_cells"]))))
        if numfields > maxfields:
            numfields = maxfields
        grass.debug("did numfields hit the max and be curtailed? %s" % numfields)
        #check for tenure, and do the appropriate type of tenure if asked
        if tenuretype == "Maximize":
            grass.message("Land Tenure is ON, with MAXIMIZING strategy")
            #check for first year, and zero out tenure if so
            if now == 1:
                tenuredfields = "%s%04d_Tenured_Fields" % (prfx, now)
                if numfields == 0:
                    grass.mapcalc("${tempfields}=0", quiet = True, tempfields = tempfields)
                else:
                    grass.run_command('r.random', quiet = 'True', flags="s", input = agcatch, npoints = numfields, raster = tempfields)
                grass.run_command('g.copy', quiet = True, raster = "%s,%s" % (tempfields,tenuredfields))
                grass.message('First year, so %s fields randomly assigned' % numfields)
                tenuredcells = numfields
                droppedcells = 0
                newcells = 0
            else:
                #make some map names
                oldfields = "%s%04d_Farming_Impacts" % (prfx, then)
                oldtenure = "%s%04d_Tenured_Fields" % (prfx, then)
                tenuredfields = "%s%04d_Tenured_Fields" % (prfx, now)
                tempcomparefields = "%stemporary_Old_Fields_yield_map" % pid
                tempagcatch = "%stemporary_agricultural_catchment_map" % pid
                grass.message('Performing yearly land tenure evaluation')
                #make a map of the current potential yields from the old fields
                grass.mapcalc("${tempcomparefields}=if(isnull(${oldfields}), null(), ${tempcerealreturn} )", quiet = "True", tempcomparefields = tempcomparefields, oldfields = oldfields, tempcerealreturn = tempcerealreturn)
                #grab stats from the old fields
                oldfieldstats = grass.parse_command('r.univar', flags = 'ge', map = tempcomparefields)
                #temporarily update the agricultural catchment by withholding last year's fields. All other cells in the catchment will be fair game to be chosen for the new fields.
                grass.mapcalc("${tempagcatch}=if(isnull(${agcatch}), null(), if(isnull(${oldfields}), ${agcatch}, null() ))", quiet = "True", tempagcatch = tempagcatch, agcatch = agcatch, oldfields = oldfields)
                newcells = numfields-tenuredcells #find out the difference between what we have and what we need
                if newcells <= 0: #negative number, so we drop any underperforming fields)
                    #check the comparison metric and drop underperforming fields compared to average yield...
                    if tenuredrop == 0: #drop threshold is 0, so compare to mean yield of all fields
                        grass.mapcalc("${tenuredfields}=if(${tempcomparefields} < (${meanyield}), null(), 1)", quiet = "True", tenuredfields=tenuredfields, tempcomparefields = tempcomparefields, meanyield = oldfieldstats['mean'])
                    else: # drop threshold is above zero, so use it to compare to the maximum yeild.
                        grass.mapcalc("${tenuredfields}=if(${tempcomparefields} < ${maxyield}-(${maxyield}*${tenuredrop}), null(), 1)", quiet = "True", tenuredfields=tenuredfields, tempcomparefields = tempcomparefields, maxyield = oldfieldstats['max'], tenuredrop = tenuredrop)
                    #pull stats out to see what we did (how many old fields kept, how many dropped, and how many new ones we need this year
                    tenuredstats = grass.parse_command('r.univar', flags = 'ge', map = tenuredfields)
                    tenuredcells = int(float(tenuredstats['cells']) - float(tenuredstats['null_cells']))
                    droppedcells = int(float(oldfieldstats['cells']) - float(oldfieldstats['null_cells'])) - tenuredcells
                    tempfields = tenuredfields
                    grass.message("Keeping %s fields in tenure list, dropping %s underperforming fields" % (tenuredcells, droppedcells))
                else: #positive number, so add fields
                    grass.mapcalc("${tenuredfields}=if(isnull(${oldfields}), null(), 1)", quiet = True, oldfields = oldfields, tenuredfields = tenuredfields) # copy last year's fields forward as tenured
                    tenuredstats = grass.parse_command('r.univar', flags = 'ge', map = oldtenure)
                    tenuredcells = int(float(tenuredstats['cells']) - float(tenuredstats['null_cells']))
                    #Now run r.random to get the required number of additional fields
                    tempfields1 = "%stemporary_extra_fields_map" % pid
                    grass.run_command('r.random', quiet = 'True', flags="s", input = tempagcatch, npoints = newcells, raster = tempfields1)
                    #patch the new fields into the old fields
                    grass.run_command('r.patch', quiet = "True", input = "%s,%s" % (tempfields1,oldtenure), output = tempfields)
                    grass.message("Keeping %s fields in tenure list, adding %s new fields" % (tenuredcells, newcells))
        elif tenuretype == "Satisfice":
            grass.message("Land Tenure is ON, with SATSFICING strategy")
            #check for first year, and zero out tenure if so
            if now == 1:
                if numfields == 0:
                    grass.mapcalc("${tempfields}=0", quiet = True, tempfields = tempfields)
                else:
                    grass.run_command('r.random', quiet = 'True', flags="s", input = agcatch, npoints = numfields, raster = tempfields)
                grass.message('First year, so %s fields randomly assigned' % numfields)
                tenuredcells = numfields
                droppedcells = 0
                newcells = 0
            else:
                #make some map names
                oldfields = "%s%04d_Farming_Impacts" % (prfx, then)
                tenuredfields = "%s%04d_Tenured_Fields" % (prfx, now)
                tempcomparefields = "%stemporary_Old_Fields_yield_map" % pid
                tempagcatch = "%stemporary_agricultural_catchment_map" % pid
                grass.message('Performing yearly land tenure evaluation')
                #make a map of the current potential yields from the old fields
                grass.mapcalc("${tempcomparefields}=if(isnull(${oldfields}), null(), ${tempcerealreturn} )", quiet = "True", tempcomparefields = tempcomparefields, oldfields = oldfields, tempcerealreturn = tempcerealreturn)
                #grab stats from the old fields
                oldfieldstats = grass.parse_command('r.univar', flags = 'ge', map = tempcomparefields)
                #make map of tenured fields for this year
                grass.mapcalc("${tenuredfields}=if(isnull(${oldfields}), null(), 1)", quiet = "True", tenuredfields=tenuredfields, oldfields = oldfields)
                #temporarily update the agricultural catchment by withholding tenured cells. All other cells in the catchment will be fair game to be chosen for the new fields.
                grass.mapcalc("${tempagcatch}=if(isnull(${agcatch}), null(), if(isnull(${tenuredfields}), ${agcatch}, null() ))", quiet = "True", tempagcatch = tempagcatch, agcatch = agcatch, tenuredfields = tenuredfields)
                #pull stats out to see what we did (how many old fields kept, how many dropped, and how many new ones we need this year
                tenuredstats = grass.parse_command('r.univar', flags = 'ge', map = tenuredfields)
                tenuredcells = int(float(tenuredstats['cells']) - float(tenuredstats['null_cells']))
                droppedcells = 0
                newcells = numfields-tenuredcells #find out the difference between what we have and what we need
                if newcells <= 0: #negative number, so we need to drop some fields.
                    #Now run r.random to randomly select fields to drop
                    tempfields1 = "%stemporary_dropped_fields_map" % pid
                    grass.run_command('r.random', quiet = 'True', flags="s", input = tenuredfields, npoints = 1-newcells, raster = tempfields1)
                    #Remove the new fields from the tenured fields map
                    grass.mapcalc("${tempfields}=if(isnull(${tenuredfields}), null(), if(isnull(${tempfields1}), 1, null()))", quiet = "True", tempfields = tempfields, tempfields1 = tempfields1, tenuredfields = tenuredfields)
                    grass.message("Dropping %s excess fields" % (1-newcells))
                else: #positive number, so add fields
                    #Now run r.random to get the required number of additional fields
                    tempfields1 = "%stemporary_extra_fields_map" % pid
                    grass.run_command('r.random', quiet = 'True', flags="s", input = tempagcatch, npoints = newcells, raster = tempfields1)
                    #patch the new fields into the old fields
                    grass.run_command('r.patch', quiet = "True", input = "%s,%s" % (tempfields1,tenuredfields), output = tempfields)
                    grass.message("Adding %s new fields" % (newcells))
        else:
            grass.message("Land Tenure is OFF")
            tenuredcells = 0
            droppedcells = 0
            newcells = 0
            #Now run r.random to get the required number of fields
            if numfields == 0:
                    grass.mapcalc("${tempfields}=0", quiet = True, tempfields = tempfields)
            else:
                grass.run_command('r.random', quiet = 'True', flags="s", input = agcatch, npoints = numfields, raster = tempfields)
        #use r.surf.gaussian to cacluate fertily impacts in the farmed areas
        grass.run_command('r.surf.gauss', quiet = "True", output = tempimpacta, mean = farmimpact[0], sigma = farmimpact[1])
        grass.mapcalc("${fields}=if(isnull(${tempfields}), null(), ${tempimpacta})", quiet = "True", fields = fields, tempfields = tempfields, tempimpacta = tempimpacta)
        #grab some yieled stats while region is still aligned to desired field size
        #first make a temporary "zone" map for the farmed areas in which to run r.univar
        tempfarmzone = "%stemporary_farming_zones_map" % pid
        grass.mapcalc("${tempfarmzone}=if(isnull(${fields}), null(), ${tempcerealreturn})", quiet = "True", tempfarmzone = tempfarmzone, fields = fields, tempcerealreturn = tempcerealreturn)
        #now use this zone map to grab some stats about the actual yields for this year's farmed fields
        cerealstats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = tempfarmzone)
        #calculate some useful stats
        cerealdif = float(cerealstats['sum']) - float(cerealreq)
        numfarmcells = int(float(cerealstats['cells']) - float(cerealstats['null_cells']))
        areafarmed = numfarmcells * float(nsfieldsize) * float(ewfieldsize)
        #update agent mempory with the farming surplus or deficit from this year, fuzzing up the means a bit to simulate the vagaries of memory. We do this by changing the actual values of these things by randomizing them through a gaussian probability filter with mu of the mean value and sigma of 0.0333. This means that the value they actually remember can be up to +- %10 of the actual mean value (eg. at the 3-sigma level of a gaussian distribution with sigma of 0.0333), although they have a better change of remembering values closer to the actual average. This more closely models how good people are at "educated guesses" of central tendencies (i.e., it's how people "guesstimate" the "average" value). This also ensures some variation from year to year, regardless of the "optimum" solution.
        farmingmemory.append(random.gauss(float(cerealdif), (float(cerealdif) * 0.0333)))
        farmyieldmemory.append(random.gauss(float(cerealstats["mean"]), (float(cerealstats['mean']) * 0.0333)))
         #find out the percentage of agcatch that was farmed this year.
        basepercent = 100 * (numfarmcells / (float(cerealstats2['cells']) - float(cerealstats2["null_cells"]) ) )
        if basepercent > 100:
            agpercent = 100.00
        else:
            agpercent = basepercent
        #reset region to original resolution
        grass.del_temp_region()
        grass.message('We farmed %s fields, using %.2f percent of agcatch...' % (numfarmcells,agpercent))
        #GENERATE GRAZING IMPACTS
        grass.message("Calculating potential grazing yields")
        #generate basic impact values
        tempimpactg = "%stemporary_grazing_impact" % pid
        grass.run_command("r.random.surface", quiet = "True", output = tempimpactg, distance = grazespatial, exponent = grazepatchy, high = maxgrazeimpact)
        #Calculate temporary grazing yield map in kg/ha
        tempgrazereturnha = "%stemporary_hectares_grazing_returns_map" % pid
        tempgrazereturn = "%stemporary_grazing_returns_map" % pid
        grass.run_command('r.recode', quiet = 'True', flags = 'da', input = oldlcov, output = tempgrazereturnha, rules = fodder_rules)
        #convert to kg / cell, and adjust to impacts
        grass.mapcalc("${tempgrazereturn}=(${tempgrazereturnha}/${cellperhectare}) * ${tempimpactg}", quiet = "True", tempgrazereturn = tempgrazereturn, tempgrazereturnha = tempgrazereturnha, cellperhectare = cellperhectare, tempimpactg = tempimpactg)
        grass.message('Figuring out grazing plans for this year....')
        #Do we graze on the stubble of agricultural fields? If so, how much fodder to we think we will get?
        if use_flags['g'] is False:
            #temporarily set region to match the field size again
            grass.use_temp_region()
            grass.run_command('g.region', quiet = 'True',nsres = nsfieldsize, ewres = ewfieldsize)
            #set up a map with the right values of stubble fodder in it, and get them to the right units (fodder per farm field)
            stubfod1 = "%stemporary_stubblefodder_1" % pid
            stubfod2 = "%stemporary_stubblefodder_2" % pid
            stubfod3 = "%stemporary_stubblefodder_3" % pid
            #make map of the basic landcover value for fields (grass stubbles)
            grass.mapcalc("${stubfod1}=if(${tempfarmzone}, ${farmval}, null())", quiet = "True", stubfod1 = stubfod1, farmval = farmval, tempfarmzone = tempfarmzone)
            #turn that map into baseline grazing yields/ha for that landcover value
            grass.run_command('r.recode', quiet = 'True', flags = 'da', input = stubfod1, output = stubfod2, rules = fodder_rules)
            #Match the variability in stubble yields to that in cereal returns, and convert to yields per field
            grass.mapcalc("${stubfod3}=(${stubfod2} / ${fieldsperhectare}) * (${tempcerealreturn}/${maxcereals})", quiet = "True", stubfod3 = stubfod3, stubfod2 = stubfod2, fieldsperhectare = fieldsperhectare, tempcerealreturn = tempcerealreturn, maxcereals = cerealstats['max'])
            stubblestats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = stubfod3)
            #Since we grazed stubble, let's' estimate how much stubbles we got by padding the mean value to a randomly generated percentage that is drawn from a gaussian probability distribution with mu of the mean value and sigma of 0.0333. This means that the absolute max/min pad can only be up to +- %10 of the mean value (eg. at the 3-sigma level of a gaussian distribution with sigma of 0.0333), and that pad values closer to 0% will be more likely than pad values close to +- 10%. This more closely models how good people are at "educated guesses" of central tendencies (i.e., it's how people "guesstimate" the "average" value). This also ensures some variation from year to year, regardless of the "optimum" solution. Once we've done that, do we think there's still some remaining fodder needs to be grazed from wild patches? If so, how much?
            if (float(fodderreq) - ( float(stubblestats['mean']) * numfarmcells )) <= 0:
                remainingfodder = 0
            else:
                remainingfodder = float(fodderreq) - ( random.gauss(float(stubblestats['mean']), (float(stubblestats['mean']) * 0.0333)) * numfarmcells )
            #reset region
            grass.del_temp_region()
        else:
            stubblestats = {"mean": '0', "sum": "0", "cells": "0", "stddev": "0", "min": "0", "first_quartile": "0", "median": "0", "third_quartile": "0", "max": "0"}
            remainingfodder = float(fodderreq)
        #Do we graze on the "fallowed" portions of the agricultural catchment?
        tempgrazecatch = "%stemporary_grazing_catchment_map" % pid
        if use_flags['f'] is True:
            grass.mapcalc("${tempgrazecatch}=if(isnull(${grazecatch}), null(), if(isnull(${agcatch}), ${tempgrazereturn}, null()))", quiet = "True", tempgrazecatch = tempgrazecatch, grazecatch = grazecatch, agcatch = agcatch, tempgrazereturn = tempgrazereturn)
        else:
            grass.mapcalc("${tempgrazecatch}=if(isnull(${grazecatch}), null(), if(isnull(${fields}), ${tempgrazereturn}, null()))", quiet = "True", tempgrazecatch = tempgrazecatch, grazecatch = grazecatch, fields = fields, tempgrazereturn = tempgrazereturn)
        #Now that we know where we are allowed to graze, how much of the grazing catchment does the agent think it needs to meet its remaining fodder requirements? First grab some general stats from the grazing catchment.
        fodderstats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = tempgrazecatch)
        # Use the agent's memory of past grazing yields and deficits to determine what they think they need to do this year.
        if len(grazeyieldmemory) == 0: #if it's the first year, then just use the fuzzed average potential yield from all cells in agcatch, and make the padded amount 1
            fuzzygyieldmemory = random.gauss(float(fodderstats['mean']), (float(fodderstats['mean']) * 0.0333))
            fuzzygdeficitmemory = -1
        else:
            fuzzygyieldmemory = numpy.mean(grazeyieldmemory[slicer:])
            fuzzygdeficitmemory = numpy.mean(grazingmemory[slicer:])
        #Figure out how many grazing patches the agent thinks it needs based on current average patch yield
        if fuzzygyieldmemory == 0:
            print("fuzzygyieldmemory is 0! numfoddercells broke")
        else:
            try:
                numfoddercells = int(round(float(remainingfodder) / fuzzygyieldmemory))
            except:
                numfoddercells = 0
        grass.debug("total graze patches should be %s" % numfoddercells)
        #discover how many cells to add or subtract based on agent memory of surplus or deficit
        if fuzzygyieldmemory == 0:
            print("fuzzygyieldmemory is 0! fodderpad broke.")
        else:
            try:
                fodderpad = int(round(fuzzygdeficitmemory / fuzzygyieldmemory))
            except:
                fodderpad = 0
        grass.debug("fodderpad (sign reversed) %s" % fodderpad)
        #add or remove the number of extra cells calculated from agent memory surplus or deficit
        numfoddercells = numfoddercells - fodderpad
        grass.debug("did numfoddercells change? %s" % numfoddercells)
        #check if we need more cells than are in the maximum grazing catchment, and clip to that catchment if so
        totgrazecells = int(float(fodderstats['cells'])) - int(float(fodderstats['null_cells']))
        grass.debug("did we have to clip numfoddercells to the catchment size? %s" % numfoddercells)
        #make the actual grazing impacts map
        if numfoddercells > totgrazecells:
            celltarget = totgrazecells
            grass.mapcalc("${grazeimpacts}=if(isnull(${tempgrazecatch}), null(), if(${oldlcov} <= ${mingraze}, null(), ${tempimpactg}))", quiet = "True", grazeimpacts = grazeimpacts, tempimpactg = tempimpactg, tempgrazecatch = tempgrazecatch, oldlcov = oldlcov, mingraze = mingraze)
        elif numfoddercells == 0:
            grass.mapcalc("${grazeimpacts}=null()", quiet = "True", grazeimpacts = grazeimpacts)
        else:
            celltarget = numfoddercells
            #now clip the cost surface to the grazable area (including cells with vegetation too low to graze on), and iterate through it to figure out the actual grazing area to be used this year
            tempgrazecost = "%stemporary_grazing_cost_map" % pid
            grass.mapcalc("${tempgrazecost}=if(isnull(${tempgrazecatch}), null(),if(${oldlcov} <= ${mingraze}, null(), ${costsurf}))", quiet = "True", tempgrazecatch = tempgrazecatch, tempgrazecost = tempgrazecost, costsurf = costsurf, oldlcov = oldlcov, mingraze = mingraze)
            catchstat = [float(x) for x in grass.read_command("r.stats", quiet = "True", flags = "1n", input = tempgrazecost, separator="=", nsteps = "10000").splitlines()]
            target = 0
            cutoff = []
            for x in sorted(catchstat):
                target = target +  1
                cutoff.append(x)
                if target >= celltarget:
                    break
            try:
                #make the actual grazing impacts map
                grass.mapcalc("${grazeimpacts}=if(${tempgrazecost} > ${cutoff}, null(), ${tempimpactg})", quiet = "True", grazeimpacts = grazeimpacts, tempimpactg = tempimpactg, tempgrazecost = tempgrazecost, cutoff = cutoff[-1])
            except:
                grass.fatal("Uh oh! Somethng wierd happened when figuring out this year\'s grazing catchment! Check your numbers and try again! Sorry!")
                sys.exit(1)
        #now get some grazing yields stats
        #Check if stubble grazing got us everything we wanted, and act appropriately
        if numfoddercells == 0:
            grazestats = {"mean": '0', "sum": "0", "cells": "0", "stddev": "0", "min": "0", "first_quartile": "0", "median": "0", "third_quartile": "0", "max": "0"}
            totalfodder = float(stubblestats['sum'])
            grazepercent = 0
            fodderdif = totalfodder - float(fodderreq)
            areagrazed = 0
        else:
            #first make a temporary "zone" map for the grazed areas in which to run r.univar
            tempgrazezone = "%stemporary_grazing_zones_map" % pid
            grass.mapcalc("${tempgrazezone}=if(isnull(${grazeimpacts}), null(), ${tempgrazereturn})", quiet = "True", tempgrazezone = tempgrazezone, grazeimpacts = grazeimpacts, tempgrazereturn = tempgrazereturn)
            #now grab the univar stats with this zone file
            grazestats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = tempgrazezone)
            numgrazecells = (float(grazestats['cells']) - float(grazestats["null_cells"]) )
            totalfodder = float(grazestats['sum']) + float(stubblestats['sum'])
            grazepercent = 100 * (numgrazecells / totgrazecells)
            fodderdif =  totalfodder - float(fodderreq)
            areagrazed = numgrazecells * (float(region['nsres']) * float(region['ewres']))
        # update agent mempory with the grazing surplus or deficit from this year, fuzzing up the means a bit to simulate the vagaries of memory. We do this by changing the actual values of these things by randomizing them through a gaussian probability filter with mu of the mean value and sigma of 0.0333. This means that the value they actually remember can be up to +- %10 of the actual mean value (eg. at the 3-sigma level of a gaussian distribution with sigma of 0.0333), although they have a better change of remembering values closer to the actual average. This more closely models how good people are at "educated guesses" of central tendencies (i.e., it's how people "guesstimate" the "average" value). This also ensures some variation from year to year, regardless of the "optimum" solution.
        grazingmemory.append(random.gauss(float(fodderdif), (float(fodderdif) * 0.0333)))
        grazeyieldmemory.append(random.gauss(float(grazestats['mean']), (float(grazestats['mean']) * 0.0333)))
        grass.message('We got %.2f kg of fodder from stubbles, and so we grazed %.2f percent of grazecatch this year.' % (float(stubblestats['sum']),grazepercent))
        #figure out how many animals and people were fed
        animfed = (totalfodder / indfodreq)
        peoplefed = ((float(cerealstats['sum']) / indcerreq) + (animfed / animals))
        grass.message('We fed %i herd animals, and produced enough food to fully support %i people this year (current population: %i)...' %(animfed, peoplefed, numpeople))
        # If the -p flag was checked, update population levels based on returns.
        if use_flags['p'] is True:
            if peoplefed / numpeople < starvthresh:     #Check if they starved this year and just die deaths if so
                numpeople = numpeople - deathdealer(deathrate, numpeople)
                grass.message("Starved a bit this year, no births will occur. New population: %i" % numpeople)
            else: #otherwise, balance births and deaths, and adjust the population accordingly
                numpeople = numpeople + babymaker(birthrate, numpeople) - deathdealer(deathrate, numpeople)
                grass.message("Balancing births and deaths... New population: %i" % numpeople)
            # Update labor and yeild needs
            cereal_pers = numpeople * agratio
            fodder_anim = animals * numpeople * pratio
            cerealreq = indcerreq * cereal_pers
            fodderreq = indfodreq * fodder_anim
            totlabor = numpeople * aglabor
            maxfields = int(round(totlabor / fieldlabor))
        #write the yield stats to the stats file
        grass.message('Writing some farming and grazing stats from this year....')
        f = open(textout3, 'a')
        if os.path.getsize(textout3) == 0:
            f.write("Year,Effective carrying capacity,Population,Percent of Agricultural Catchment Used,Number of Farm Fields,Number of Tenured Fields,Number of Dropped Fields,Number of New Fields,Total Farmed Area (m2),Agent Memory of Per Field Harvest Mean,Per Field Harvest Mean,Per Field Harvest Standard Deviation,Total Cereals Harvested,Total Cereals Required,Cereal Surplus/Deficit,Agent Memory of Cereal Surplus/Deficit,,Herd Animals Fed,Percent of Grazing Catchment Used,Total Grazed Area (m2),Agent Memory of Wild Grazing Patch Mean,Wild Grazing Patch Mean,Wild Grazing Patch Standard Deviation,Total Wild Fodder,Field Stubbles Mean,Field Stubbles Standard Deviation,Total Stubble Fodder,Total Fodder Consumed,Total Amount of Fodder Required,Fodder Surplus/Deficit,,,Minimum Cereals,First Quartile Cereals,Third Quartile Cereals,Maximum Cereals,,Minimum Wild Fodder,First Quartile Wild Fodder,Third Quartile Wild Fodder,Maximum Wild Fodder,,Minimum Stubble Fodder,Stubble Quartile Stubble Fodder,Third Quartile Stubble Fodder,Maximum Stubble Fodder") # the file is empty, so write the headers on the first line.
        f.write('\n%s' % now + ',' + str(peoplefed) + ',' + str(numpeople) + ',' + str(agpercent) + ',' + str(numfarmcells) + ',' + str(tenuredcells) + ',' + str(droppedcells) + ',' + str(newcells) + ',' + str(areafarmed) + ',' + str(fuzzyyieldmemory) + ',' + cerealstats['mean'] + ',' + cerealstats['stddev'] + ',' + cerealstats['sum'] + ',' + str(cerealreq) + ',' + str(cerealdif) + ',' + str(fuzzydeficitmemory) + ',,' + str(animfed) + ',' + str(grazepercent) + ',' + str(areagrazed) + ',' + str(fuzzygyieldmemory) + ',' + grazestats['mean'] + ',' + grazestats['stddev'] + ',' + grazestats['sum'] + ',' + stubblestats['mean'] + ',' + stubblestats['stddev'] + ',' + stubblestats['sum'] + ',' + str(totalfodder) + ',' + str(fodderreq) + ',' + str(fodderdif) + ',,,' + cerealstats['min'] + ',' + cerealstats['first_quartile'] + ',' + cerealstats['first_quartile'] + ',' + cerealstats['max'] + ',,' + grazestats['min'] + ',' + grazestats['first_quartile'] + ',' + grazestats['third_quartile'] + ',' + grazestats['max'] + ',,' + stubblestats['min'] + ',' + stubblestats['first_quartile'] + ',' + stubblestats['third_quartile'] + ',' + stubblestats['max']) # update this year's row with the data from this year's simulation
        #UPDATE LANDCOVER AND SOIL FERTILITY
        grass.message('Updating landcover and soil fertility with new impacts')
        #update fertility
        tempfertil = "%stemporary_fertility_regain_map" % pid
        #use r.surf.gaussian to cacluate fertily regain map
        grass.run_command('r.surf.gauss', quiet = "True", output = tempfertil, mean = fertilrate[0], sigma = fertilrate[1])
        #figure out what happened to fertility (see if stubble-grazing is enabled, and make sure to add some manure where grazing occured, scaled to the degree of graing that happened)
        if use_flags['g'] is False:
            grass.mapcalc("${outfert}=eval(a=if(isnull(${grazeimpacts}) && isnull(${fields}), ${tempfertil}, ${tempfertil} + (${manurerate} * ${tempimpactg})), b=if(isnull(${fields}), ${oldfert}, ${oldfert} - ${fields}), c=if(b <= ${maxfert} - a, b + a, ${maxfert}), if(c < 0, 0, c))", quiet = "True", outfert = outfert, oldfert = oldfert, fields = fields, tempimpactg = tempimpactg, grazeimpacts = grazeimpacts, manurerate = manurerate, maxfert = maxfert, tempfertil = tempfertil)
        else:
            grass.mapcalc("${outfert}=eval(a=if(isnull(${grazeimpacts}), ${tempfertil}, ${tempfertil} + (${manurerate} * ${tempimpactg})), b=if(isnull(${fields}), ${oldfert}, ${oldfert} - ${fields}), if(b <= ${maxfert} - a, b + a, ${maxfert}))", quiet = "True", outfert = outfert, oldfert = oldfert, fields = fields, tempimpactg = tempimpactg, grazeimpacts = grazeimpacts, manurerate = manurerate, maxfert = maxfert, tempfertil = tempfertil)
        grass.run_command('r.colors', quiet = "True", map = outfert, rules = fertcolors.name)
        #update landcover
        # calculating rate of regrowth based on current soil fertility, spil depths, and precipitation. Recoding fertility (0 to 100%), depth (0 to >= 1m), and precip (0 to >= 1000mm) with a power regression curve from 0 to 1, then taking the mean of the two as the regrowth rate
        growthrate = "%stemporary_vegetation_regrowth_map" % pid
        grass.mapcalc('${growthrate}=eval(x=if(${sdepth} <= 1.0, ( -0.000118528 * (exp((100*${sdepth}),2.0))) + (0.0215056 * (100*${sdepth})) + 0.0237987, 1), y=if(${precip} <= 1.0, ( -0.000118528 * (exp((100*${precip}),2.0))) + (0.0215056 * (100*${precip})) + 0.0237987, 1), z=(-0.000118528 * (exp(${outfert},2.0))) + (0.0215056 * ${outfert}) + 0.0237987, a=if(x <= 0 || z <= 0, 0, (x+y+z)/3), if(a < 0, 0, a) )', quiet = "True", growthrate = growthrate,  sdepth = oldsdepth, outfert = outfert, precip = precip)
        #Calculate this year's landcover impacts and regrowth
        grass.mapcalc("${outlcov}=eval(a=if(${oldlcov} - ${grazeimpacts} + ${growthrate} >= 0, ${oldlcov} - ${grazeimpacts} + ${growthrate}, 0) , b=if(isnull(${fields}), a, ${farmval}), if(${oldlcov} < (${maxlcov} - ${growthrate}) && isnull(b), ${oldlcov} + ${growthrate}, if(isnull(b), ${maxlcov}, b) ))", quiet = "True", outlcov = outlcov, oldlcov = oldlcov, maxlcov = maxlcov, growthrate = growthrate, fields = fields, farmval = farmval, grazeimpacts = grazeimpacts)
        #Make a rainfall excess map to send to r.landcape.evol. This is a logarithmic regression (R^2=0.99.) for the data pairs: 0,90;3,85;8,70;13,60;19,45;38,30;50,20. These are the same succession cutoffs that are used in the c-factor coding.
        grass.mapcalc("${outxs}=193.522 - (42.3272 * log(${lcov} + 10.9718))", quiet = "True", outxs = outxs, lcov = outlcov)
        #if rules set exists, create reclassed landcover labels map
        try:
            temp_reclass = "%stemporary_reclassed_landcover" %pid
            grass.run_command('r.reclass', quiet = "True",  input = outlcov,  output = temp_reclass,  rules = lc_rules)
            grass.mapcalc('${out}=${input}', quiet = "True", overwrite = "True", out = outlcov, input = temp_reclass)
        except:
            grass.warning("No landcover labling rules found at path \"%s\"\nOutput landcover map will not have text labels in queries" % lc_rules)
        grass.run_command('r.colors',  quiet = "True",  map = outlcov, rules = lccolors.name)
        #collect and write landcover and fertiltiy temporal matrices
        grass.message('Collecting some landcover and fertility stats from this year....')
        f = open(textout, 'a')
        if os.path.getsize(textout) == 0:
            f.write("Temporal Matrix of Landcover\n\nYear," + ",".join(str(i) for i in range(maxval + 1)) + "\n")
        statdict = grass.parse_command('r.stats', quiet = "True",  flags = 'ani', input = outlcov, separator = '=', nv ='*')
        f.write("%s," % now)
        for key in range(maxval + 1):
            try:
                f.write(statdict[str(key)] + "," )
            except:
                f.write("0,")
        f.write("\n")
        f.close()
        f = open(textout2, 'a')
        if os.path.getsize(textout2) == 0:
            f.write("Temporal Matrix of Soil Fertility\n\nYear," + ",".join(str(i) for i in range(maxfertval + 1)) + "\n")
        statdict = grass.parse_command('r.stats', quiet = "True",  flags = 'ani', input = outfert, separator = '=', nv ='*')
        f.write("%s," % now)
        for key in range(maxfertval + 1):
            try:
                f.write(statdict[str(key)] + "," )
            except:
                f.write("0,")
        f.write("\n")
        f.close()
        #collect and write univariate stats
        #grass.run_command('r.mask', quiet = "True", raster = grazecatch)
        grass.mapcalc("MASK=if(isnull(${grazecatch}), null(), 1)", quiet = "True", overwrite = "True", grazecatch = grazecatch)
        lcovstats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = outlcov)
        grass.run_command('g.remove', quiet = "True", flags = "f", type = "rast", name = "MASK")
        #grass.run_command('r.mask', quiet = "True", raster = agcatch)
        grass.mapcalc("MASK=if(isnull(${agcatch}), null(), 1)", quiet = "True", overwrite = "True", agcatch = agcatch)
        fertstats = grass.parse_command('r.univar', flags = 'ge', percentile = '90', map = outfert)
        grass.run_command('g.remove', quiet = "True", flags = "f", type = "rast", name = "MASK")
        f = open(textout4, 'a')
        if os.path.getsize(textout4) == 0:
            f.write("Landcover and Soil Fertility Stats\nNote that these stats are collected within the grazing catchment (landcover) and agricultural catchment (fertility) ONLY. Rest of the map is ignored.\n\n,,Basic Stats,,,,Extended Stats\nYear,,Mean Landcover,Standard Deviation Landcover,Mean Soil Fertility,Standard Deviation Soil Fertility,,Minimum Landcover,First Quartile Landcover,Median Landcover,Third Quartile Landcover,Maximum Landcover,,Minimum Soil Fertility,First Quartile Soil Fertility,Median Soil Fertility,Third Quartile Soil Fertility,Maximum Soil Fertility")
        f.write('\n%s' % now + ',,' + lcovstats['mean'] + ',' + lcovstats['stddev'] + ',' + fertstats['mean'] + ',' + fertstats['stddev'] + ',,' + lcovstats['max'] + ',' + lcovstats['third_quartile'] + ',' + lcovstats['median'] + ',' + lcovstats['first_quartile'] + ',' + lcovstats['min'] + ',,' + fertstats['min'] + ',' + fertstats['first_quartile'] + ',' + fertstats['median'] + ',' + fertstats['third_quartile'] + ',' + fertstats['max'])
        #creating c-factor map
        grass.message('Creating C-factor map for r.landscape.evol')
        try:
            grass.run_command('r.recode', quiet = True, input = outlcov, output = outcfact, rules = cfact_rules)
        except:
            grass.fatal("NO CFACTOR RECLASS RULES WERE FOUND AT PATH \"%s\"\nPLEASE ENSURE THAT THE CFACTOR RECODE RULES EXIST AND ARE WRITTEN PROPERLY, AND THEN TRY AGAIN" % cfact_rules)
            sys.exit(1)
        grass.run_command('r.colors',  quiet = True, map = outcfact, rules = cfcolors.name)
        #Run r.landscape.evol with this years' cfactor map
        grass.message('Running landscape evolution for this year....')
        #check if this is year one, and use the starting dem if so
        if now == 1:
            inelev = elev
        else:
            inelev = "%s%04d_Elevation" % (prfx, then)
        # try:            
        grass.run_command('r.landscape.evol', quiet = "True", overwrite=True, number = 1, prefx = "%s%04d_" % (prfx, now), c = outcfact, elev = inelev, initbdrk = initbdrk, transp_eq = transp_eq, outdem = "Elevation", outsoil = "Soil_Depth", exp_m = exp_m, exp_n=exp_n, k = k, sdensity = sdensity, manningn = manningn, flowcontrib = outxs, convergence = convergence, r = r, rain = rain, storms = storms, stormlength = stormlength, stormi=stormi, statsout = statsout, flags = ''.join(levol_flags))
        # except:
            # grass.fatal("Something is wrong with the values you sent to r.landscape.evol. Did you forget something? Check the values and try again...\nSimulation terminated with an error at time step %s" % now)
            # sys.exit(1)
        #delete C-factor map, unless asked to save it
        if use_flags['c'] is False:
            grass.run_command("g.remove", quiet = "True", flags = 'f', type = "rast", name = "%s,%s" %  (outcfact,outxs))
        else:
            pass
        #clean up temporary maps
        grass.run_command('g.remove', quiet = "True", flags = 'f', type = "rast", pattern = '%s*' % pid)
        grass.message('Completed year %s of the simulation' % now)
    lccolors.close()
    cfcolors.close()
    fertcolors.close()
    return(grass.message(".........................SIMULATION COMPLETE...........................\nCheck in the current mapset for farming/grazing yields, landcover, fertility, and erosion/depostion stats files from this run."))


#Here is where the code in "main" actually gets executed. This way of programming is neccessary for the way g.parser needs to run.
if __name__ == "__main__":
    options, flags = grass.parser()
    main()
    exit(0)