Esempio n. 1
0
File: mad.py Progetto: acviana/agpy
def MAD(a, c=0.6745, axis=None):
    """
    Median Absolute Deviation along given axis of an array:

    median(abs(a - median(a))) / c

    c = 0.6745 is the constant to convert from MAD to std; it is used by
    default

    """

    a = ma.masked_where(a!=a, a)
    if a.ndim == 1:
        d = ma.median(a)
        m = ma.median(ma.fabs(a - d) / c)
    else:
        d = ma.median(a, axis=axis)
        # I don't want the array to change so I have to copy it?
        if axis > 0:
            aswp = ma.swapaxes(a,0,axis)
        else:
            aswp = a
        m = ma.median(ma.fabs(aswp - d) / c, axis=0)

    return m
Esempio n. 2
0
File: mad.py Progetto: wiai/agpy
def MAD(a, c=0.6745, axis=None):
    """
    Median Absolute Deviation along given axis of an array:

    median(abs(a - median(a))) / c

    c = 0.6745 is the constant to convert from MAD to std; it is used by
    default

    """

    a = ma.masked_where(a != a, a)
    if a.ndim == 1:
        d = ma.median(a)
        m = ma.median(ma.fabs(a - d) / c)
    else:
        d = ma.median(a, axis=axis)
        # I don't want the array to change so I have to copy it?
        if axis > 0:
            aswp = ma.swapaxes(a, 0, axis)
        else:
            aswp = a
        m = ma.median(ma.fabs(aswp - d) / c, axis=0)

    return m
def MAD(a, c=0.6745, axis=None):


    a = ma.masked_where(a!=a, a)
    if a.ndim == 1:
        d = ma.median(a)
        m = ma.median(ma.fabs(a - d) / c)
    else:
        d = ma.median(a, axis=axis)
        # I don't want the array to change so I have to copy it?
        if axis > 0:
            aswp = ma.swapaxes(a,0,axis)
        else:
            aswp = a
        m = ma.median(ma.fabs(aswp - d) / c, axis=0)

    return m
def MAD(a, c=0.6745, axis=None):


    a = ma.masked_where(a!=a, a)
    if a.ndim == 1:
        d = ma.median(a)
        m = ma.median(ma.fabs(a - d) / c)
    else:
        d = ma.median(a, axis=axis)
        # I don't want the array to change so I have to copy it?
        if axis > 0:
            aswp = ma.swapaxes(a,0,axis)
        else:
            aswp = a
        m = ma.median(ma.fabs(aswp - d) / c, axis=0)

    return m
Esempio n. 5
0
def MAD(a, c=0.6745, axis=None):
    """
    Median Absolute Deviation along given axis of an array:
    median(abs(a - median(a))) / c
    c = 0.6745 is the constant to convert from MAD to std
    """
    
    a = ma.masked_where(a!=a, a)
    if a.ndim == 1:
        d = ma.median(a)
        m = ma.median(ma.fabs(a - d) / c)
    else:
        d = ma.median(a, axis=axis)
        if axis > 0:
            aswp = ma.swapaxes(a,0,axis)
        else:
            aswp = a
        m = ma.median(ma.fabs(aswp - d) / c, axis=0)

    return m
def carbchem_revelle(op_swtch,mdi,T_cube,S_cube,TCO2_cube,TALK_cube,Pr=0.0,TB=0.0,Ni=100.0,Tl=1.0e-5):
# This function calculates the inorganic carbon chemistry balance
# according to the method of Peng et al 1987
# The parameters are set in the first few lines

#salinity needs to be converted into psu
#TCO2 and TALK must be in mol/kg
#the ones below here are not needed

# This procedure calculates the inorganic carbon chemistry balance
# according to the method of Peng et al 1987
# The parameters are set in the first few lines
#
#  ops= 0 ;  output is iteration count
#       1 ;            pCO2
#       2 ;            pH
#       3 ;            [H2CO3]
#       4 ;            [HCO3]
#       5 ;            [CO3]
#       6 ;            satn [co3] : calcite
#       7 ;            saturation state: calcite
#       8 ;            satn [CO3] : aragonite
#       9 ;            saturation state: aragonite
#	10;            Ravelle factor (DIC) calculated from Egleston et al. 2010
#	11;            Alkalinity buffer factor calculated from Egleston et al. 2010

    #make sure grids are same size
    #make sure rthey years are the same
    #extarct the data from the cubes
    
# from iris import *
# from iris.analysis import *
# import iris.analysis
# from numpy import *
# from matplotlib.pyplot import *
# from scipy.stats.mstats import *
# import iris.plot as iplt
# import seawater
# import numpy
# import iris.quickplot as quickplot
# import iris.analysis.stats as istats
# temp = iris.load_cube('/home/ph290/tmp/hadgem2es_potential_temperature_historical_regridded.nc').extract(Constraint(depth = 0))
# sal = iris.load_cube('/home/ph290/tmp/hadgem2es_salinity_historical_regridded.nc').extract(Constraint(depth = 0))
# carb = iris.load_cube('/home/ph290/tmp/hadgem2es_dissolved_inorganic_carbon_historical_regridded.nc').extract(Constraint(depth = 0))
# alk = iris.load_cube('/home/ph290/tmp/hadgem2es_total_alkalinity_historical_regridded.nc').extract(Constraint(depth = 0))
# import carbchem
# co2 = carbchem.carbchem(1,temp.data.fill_value,temp,sal,carb,alk)
# T_cube = temp
# S_cube = sal
# TCO2_cube = carb
# TALK_cube = alk  
# mdi = temp.data.fill_value
	
    t_lat = np.size(T_cube.coord('latitude').points)    
    s_lat = np.size(S_cube.coord('latitude').points)
    c_lat = np.size(TCO2_cube.coord('latitude').points)
    a_lat = np.size(TALK_cube.coord('latitude').points)
    lat_test = t_lat == s_lat == c_lat == a_lat

    t_lon = np.size(T_cube.coord('longitude').points) 
    s_lon = np.size(S_cube.coord('longitude').points)
    c_lon = np.size(TCO2_cube.coord('longitude').points)
    a_lon = np.size(TALK_cube.coord('longitude').points)
    lon_test = t_lon == s_lon == c_lon == a_lon

    if lat_test and lon_test:

		output_cube = T_cube.copy()
		T_cube = T_cube-273.15
		T = T_cube.data.copy()
		S = S_cube.data.copy()
		TCO2_cube = TCO2_cube/1026.0
		TCO2 = np.roll(ma.swapaxes(TCO2_cube.data.copy(),0,1),180)
		#NOTE - this is only required here 'cos glodap and WOA are differently ordered - not necessary for other stuff
		TALK_cube = TALK_cube/1026.0
		TALK = np.roll(ma.swapaxes(TALK_cube.data.copy(),0,1),180)
		
		print np.mean(T)
		print np.mean(S)
		print np.mean(TCO2)
		print np.mean(TALK)
		
		msk1=ma.masked_greater_equal(T,mdi-1.0,copy=True)
		msk2=ma.masked_greater_equal(S,mdi-1.0,copy=True)
		msk3=ma.masked_greater_equal(TCO2,mdi-1.0,copy=True)
		msk4=ma.masked_greater_equal(TALK,mdi-1.0,copy=True)

		msk=msk1.mask | msk2.mask | msk3.mask | msk4.mask

		T[msk]=np.nan
		S[msk]=np.nan
		TALK[msk]=np.nan
		TCO2[msk]=np.nan
# 		plt.contourf(T)
# 		plt.show()
# 		plt.contourf(TCO2)
# 		plt.show()

		# T = np.array([13.74232016,25.0])
		# S = np.array([33.74096661,35.0])
		# TCO2 = np.array([0.0019863,2.0e-3])
		# TALK = np.array([0.00226763,2.2e-3])
		# msk = ma.masked_greater_equal(T,mdi-1.0,copy=True)

		#create land-sea mask used by sea_msk.mask
		salmin = 1.0
		S2=np.copy(S)
		S2[np.abs(S) < salmin]=salmin

		tol = Tl
		mxiter = Ni

		op_fld = np.empty(T.shape)
		op_fld.fill(np.NAN)

#    TB = np.ones(T.shape)
#    TB = 4.106e-4*S2/35.0
		TB = np.empty_like(T)
		TB = np.multiply(S2,4.106e-4/35.0, TB)
		# this boron is from Peng

		#convert to Kelvin
		TK=np.copy(T[:])
		TK += +273.15

		alpha_s = np.ones(T.shape)
		alpha_s = np.exp( ( -60.2409 + 9345.17/TK  + 23.3585*np.log(TK/100.0) )  + ( 0.023517 - 0.023656*(TK/100.0) + 0.0047036*np.power((TK/100.0),2.0) )*S )

		K1 = np.ones(T.shape)
		K1 = np.exp( ( -2307.1266/TK + 2.83655  - 1.5529413*np.log(TK) ) - ( 4.0484/TK + 0.20760841 )*np.sqrt(S) + 0.08468345*S - 0.00654208*np.power(S,1.5) + np.log( 1.0 - 0.001005*S ) )

		a = np.array([-25.50,-15.82,-29.48,-25.60,-48.76,-46.0])
		b = np.array([0.1271,0.0219,0.2324,0.5304,0.5304])
		c = np.array([0.0,0.0,0.0026080,0.0036246,0.0,0.0])
		d = np.array([-3.08,1.13,(-2.84e-3)/(1.0e-3),-5.13,-11.76,-11.76])
		e = np.array([0.0877,0.1475,0.0,0.0794,0.3692,0.3692])

		if keyword.iskeyword(Pr):
				instance = 0
				pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
				K1 = K1*pf

		K2 = np.ones(T.shape)
		K2 = np.exp( ( -3351.6106/TK - 9.226508 - 0.2005743*np.log(TK) ) - ( 23.9722/TK + 0.106901773 )*np.power(S,0.5) + 0.1130822*S - 0.00846934*np.power(S,1.5) + np.log( 1.0 - 0.001005*S ) )

		if keyword.iskeyword(Pr):
				instance = 1
				pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
				K2 = K2*pf

		KB = np.ones(T.shape)
		KB = np.exp( ( -8966.90 - 2890.53*np.power(S,0.5) - 77.942*S + 1.728*np.power(S,1.5)- 0.0996*np.power(S,2.0) )/TK + ( 148.0248 + 137.1942*np.power(S,0.5) + 1.62142*S ) - ( 24.4344 + 25.085*np.power(S,0.5) + 0.2474*S )*np.log(TK) + 0.053105*(np.power(S,0.5))*TK )

		if keyword.iskeyword(Pr):
				instance = 2
				pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
				KB = KB*pf

		KW = np.ones(T.shape)
		KW = np.exp( ( -13847.26/TK + 148.96502 - 23.6521*np.log(TK) ) + ( 118.67/TK - 5.977 + 1.0495*np.log(TK) )*np.power(S,0.5) - 0.01615*S )

		if keyword.iskeyword(Pr):
				instance = 3
				pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
				KW = KW*pf

		if ( op_swtch >= 6 or op_swtch <= 9 ):
				ca_conc = np.ones(T.shape)
				ca_conc = 0.01028*S2/35.0

		if ( op_swtch == 6 or op_swtch == 7 ):
				K_SP_C = np.ones(T.shape)
				K_SP_C = np.power(10.0,( ( -171.9065 - 0.077993*TK + 2839.319/TK + 71.595*np.log10(TK) ) + ( -0.77712 + 0.0028426*TK + 178.34/TK )*np.power(S,0.5) - 0.07711*S+ 0.0041249*np.power(S,1.5) ))
				if keyword.iskeyword(Pr):
						instance = 4
						pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
						K_SP_C = K_SP_C*pf


		if ( op_swtch == 8 or op_swtch == 9 ):
				K_SP_A = np.ones(T.shape)
				K_SP_A = np.power(10,( ( -171.945 - 0.077993*TK + 2903.293/TK + 71.595*np.log10(TK) ) + ( -0.068393 + 0.0017276*TK + 88.135/TK )*np.power(S,0.5) - 0.10018*S + 0.0059415*np.power(S,1.5) ))
				if keyword.iskeyword(Pr):
						instance = 5
						pf = pressure_fun(a[instance],b[instance],c[instance],d[instance],e[instance],T)
						K_SP_A = K_SP_A*pf


		# Get first estimate for H+ concentration.

		AC, AW, AB, aH, count = carbiter(T, TCO2, TALK, TB, msk, tol, mxiter, K1, K2, KB, KW)

# 		plt.contourf(aH)
# 		plt.show()
# 		plt.contourf(AC)
# 		plt.show()
# 		plt.contourf(AW)
# 		plt.show()
# 		plt.contourf(aH)
# 		plt.show()

		# now we have aH we can calculate...
		denom = np.zeros(T.shape)
		H2CO3 = np.zeros(T.shape)
		HCO3 = np.zeros(T.shape)
		CO3 = np.zeros(T.shape)
		pH = np.zeros(T.shape)
		pCO2 = np.zeros(T.shape)
		if ( op_swtch == 6 or op_swtch == 7 ):
				sat_CO3_C = np.zeros(T.shape)
		if ( op_swtch == 7 ):
				sat_stat_C = np.zeros(T.shape)
		if ( op_swtch == 8 or op_swtch == 9 ):
				sat_CO3_A = np.zeros(T.shape)
		if ( op_swtch == 9 ):
				sat_stat_A = np.zeros(T.shape)

		denom = np.power(aH,2.0) + K1*aH + K1*K2
		H2CO3 = TCO2*np.power(aH,2.0)/denom
		HCO3 = TCO2*K1*aH/denom
		CO3 = TCO2*K1*K2/denom
# 		plt.contourf(K1)
# 		plt.show()
# 		plt.contourf(aH) -no
# 		plt.show()
# 		plt.contourf(denom) -no
# 		plt.show()

		pH = -np.log10(aH)
		pCO2 = H2CO3/alpha_s

		if ( op_swtch == 6 or op_swtch == 7 ):
				sat_CO3_C = K_SP_C/ca_conc
				if ( op_swtch == 7 ):
						sat_stat_C = CO3/sat_CO3_C

		if ( op_swtch == 8 or op_swtch == 9 ):
				sat_CO3_A = K_SP_A/ca_conc
				if ( op_swtch == 9 ):
						sat_stat_A = CO3/sat_CO3_A
						
		TALKc=+HCO3+2*(CO3)
		var1=HCO3
		DIC_buffer=HCO3
		ALK_buffer=HCO3
		
		var1=HCO3+4*(CO3)+((aH*AB)/(KB+aH))-AW
		
		DIC_buffer=TCO2-((TALKc*TALKc)/var1)
		
		ALK_buffer=((TALKc*TALKc)-TCO2*var1)/TALKc

		output_cube = output_cube*0.0+np.nan
		if ( op_swtch == 0 ):
				op_fld = np.zeros(T.shape)
				op_fld = count
		elif ( op_swtch == 1 ):
				print np.mean(pCO2)
				output_cube.data = pCO2*1.0e6
				output_cube.standard_name = 'surface_partial_pressure_of_carbon_dioxide_in_sea_water'
				output_cube.long_name = 'CO2 concentration'
				output_cube.units = 'uatm'
		elif ( op_swtch == 2 ):
				output_cube.data = pH
				output_cube.standard_name = 'sea_water_ph_reported_on_total_scale'
				output_cube.long_name = 'pH'
				output_cube.units = '1'
		elif ( op_swtch == 3 ):
				output_cube.data = H2CO3
		elif ( op_swtch == 4 ):
				output_cube.data = HCO3
		elif ( op_swtch == 5 ):
				output_cube.data = CO3
		elif ( op_swtch == 6 ):
				output_cube.data = sat_CO3_C
		elif ( op_swtch == 7 ):
				output_cube.data = sat_stat_C
		elif ( op_swtch == 8 ):
				output_cube.data = sat_CO3_A
		elif ( op_swtch == 9 ):
				output_cube.data = sat_stat_A
		elif ( op_swtch == 10 ):
				output_cube.data = TCO2/DIC_buffer
		elif ( op_swtch == 11 ):
				output_cube.data = ALK_buffer*1000.0

		return output_cube