illustrates the use of the `proxy`, `analogs` and `scalar_plot` classes



In [ ]:

    
%matplotlib inline
import pandas as pd
import numpy as np
try:
    import xarray as xray 
except: 
    import xray
import matplotlib.pyplot as plt

import the development version of paleopy



In [2]:

    
import sys



In [3]:

    
sys.path.insert(0, '../')



In [4]:

    
from paleopy import proxy 
from paleopy import analogs
from paleopy.plotting import scalar_plot

example 1: defines one proxy

defines the folder where the JSON files are (for the datasets) and where to save the proxy JSON files



In [5]:

    
djsons = '../jsons/'
pjsons = '../jsons/proxies'

instantiates a proxy instance



In [6]:

    
p = proxy(sitename='Rarotonga', \
          lon = -159.82, \
          lat = -21.23, \
          djsons = djsons, \
          pjsons = pjsons, \
          pfname = 'Rarotonga.json', \
          dataset = 'ersst', \
          variable ='sst', \
          measurement ='delta O18', \
          dating_convention = 'absolute', \
          calendar = 'gregorian',\
          chronology = 'historic', \
          season = 'DJF', \
          value = 0.6, \
          qualitative = 0, \
          calc_anoms = 1, \
          detrend = 1, \
        method = 'quintiles')

find the analogs, calls the internal functions necessary to get there



In [8]:

    
p.find_analogs()



In [10]:

    
p.proxy_repr(pprint=True)









    



{
sitename:Rarotonga
proxy_type:None
measurement:delta O18
dating_convention:absolute
calendar:gregorian
chronology:historic
coords:(200.18, -21.23)
aspect:None
elevation:None
season:DJF
dataset:ersst
variable:sst
calc_anoms:True
detrend:True
value:0.6
climatology:(1981, 2010)
period:(1979, 2014)
extracted_coords:[200.0, -22.0]
distance_point:87.56368858840081
trend_params:{'intercept': -60.965016377755163, 'slope': 0.030535947029886677}
category:WA
analog_years:[1982, 1997, 1999, 2000, 2001, 2006, 2009]
weights:[0.15645480307020584, 0.12800536294956205, 0.13090508326029401, 0.16334634614636281, 0.15762466128135916, 0.1220029495538857, 0.1416607937383304]
}

those are the extracted analog years



In [11]:

    
p.analog_years









    Out[11]:





array([1982, 1997, 1999, 2000, 2001, 2006, 2009], dtype=int32)

`analogs` is an attribute of the proxy: it's a `pandas` dataframe containing info about the analog years



In [12]:

    
p.analogs









    Out[12]:






  
    
      
      d_anomalies
      cat
      weights
    
    
      time
      
      
      
    
  
  
    
      1982-02-15
      0.638992
      WA
      0.156455
    
    
      1997-02-15
      0.747619
      WA
      0.128005
    
    
      1999-02-15
      0.736547
      WA
      0.130905
    
    
      2000-02-15
      0.612678
      WA
      0.163346
    
    
      2001-02-15
      0.565475
      WA
      0.157625
    
    
      2006-02-15
      0.429462
      WA
      0.122003
    
    
      2009-02-15
      0.504521
      WA
      0.141661

Adding `outfile=True` creates the json file containing the proxy information, with filename `self.pfname`



In [14]:

    
p.proxy_repr(pprint=False, outfile=True)



In [15]:

    
!ls -lt ../jsons/proxies/Rarotonga.json









    



-rw-r--r--  1 nicolasf  staff  771 May  3 12:14 ../jsons/proxies/Rarotonga.json

plot the seasonal time-series of anomalies and indicates the analog years



In [16]:

    
f = p.plot_season_ts()

instantiate the analogs class with the proxy object and the dataset + variable to composite



In [26]:

    
compos = analogs(p, 'ncep', 'hgt_1000').composite()



In [27]:

    
f = scalar_plot(compos, test=0.1).plot(subplots=False)

a smaller domain (NZ)



In [28]:

    
f = scalar_plot(compos, test=0.1, proj='cyl', domain=[165, 180, -50., -30], res='h').plot(subplots=False)

now one plot per year in the analog sample



In [29]:

    
f = scalar_plot(compos, test=0.1, proj='cyl', domain=[165, 180, -50., -30], res='h').plot(subplots=True)

Tmean composite from VCSN



In [30]:

    
compos = analogs(p, 'vcsn', 'TMean').composite()



In [31]:

    
f = scalar_plot(compos, test=0.1, proj='cyl', res='h', vmin=-0.5, vmax=0.5).plot(subplots=False)

saves to disk the xray Dataset containing the composite sample, anomalies and associated metadata



In [32]:

    
compos.dset









    Out[32]:





<xarray.Dataset>
Dimensions:              (agent: 11491, latitudes: 260, longitudes: 243, years: 7)
Coordinates:
  * longitudes           (longitudes) float64 166.4 166.4 166.5 166.5 166.6 ...
  * latitudes            (latitudes) float64 -47.35 -47.3 -47.25 -47.2 ...
  * agent                (agent) int64 3027 3106 3380 4782 5878 7317 7318 ...
  * years                (years) int32 1982 1997 1999 2000 2001 2006 2009
Data variables:
    mask                 (latitudes, longitudes) int64 1 1 1 1 1 1 1 1 1 1 1 ...
    composite_sample     (years, latitudes, longitudes) float64 nan nan nan ...
    composite_anomalies  (latitudes, longitudes) float64 nan nan nan nan nan ...
    weights              (years) float64 0.1565 0.128 0.1309 0.1633 0.1576 ...
    pvalues              (latitudes, longitudes) float64 nan nan nan nan nan ...



In [40]:

    
compos.save_to_file('/Users/nicolasf/Desktop/vcsn_tmean.nc')



In [41]:

    
!ncdump -h /Users/nicolasf/Desktop/vcsn_tmean.nc









    



netcdf vcsn_tmean {
dimensions:
	years = 7 ;
	latitudes = 89 ;
	longitudes = 180 ;
variables:
	double composite_sample(years, latitudes, longitudes) ;
		composite_sample:_FillValue = -999.9 ;
		composite_sample:missing_value = -999.9 ;
	double composite_anomalies(latitudes, longitudes) ;
		composite_anomalies:_FillValue = -999.9 ;
		composite_anomalies:missing_value = -999.9 ;
	double pvalues(latitudes, longitudes) ;
	float latitudes(latitudes) ;
		string latitudes:units = "degrees_north" ;
		string latitudes:long_name = "Latitudes" ;
		string latitudes:axis = "Y" ;
	int years(years) ;
	float longitudes(longitudes) ;
		string longitudes:units = "degrees_east" ;
		string longitudes:long_name = "Longitudes" ;
		string longitudes:axis = "X" ;
}

Now with another dataset (SST from ERSST)



In [42]:

    
compos = analogs(p, 'ersst', 'sst').composite()



In [43]:

    
f = scalar_plot(compos, test=0.1).plot(subplots=False)

now maps: global domain, Equidistant Cylindrical projection, significance level 90%



In [44]:

    
f = scalar_plot(compos, test=0.1, proj='spstere').plot(subplots=False)



In [38]:

    
compos.dset









    Out[38]:





<xarray.Dataset>
Dimensions:              (latitudes: 89, longitudes: 180, years: 7)
Coordinates:
  * longitudes           (longitudes) float32 0.0 2.0 4.0 6.0 8.0 10.0 12.0 ...
  * latitudes            (latitudes) float32 -88.0 -86.0 -84.0 -82.0 -80.0 ...
  * years                (years) int32 1982 1997 1999 2000 2001 2006 2009
Data variables:
    mask                 (latitudes, longitudes) int64 1 1 1 1 1 1 1 1 1 1 1 ...
    composite_sample     (years, latitudes, longitudes) float64 nan nan nan ...
    composite_anomalies  (latitudes, longitudes) float64 nan nan nan nan nan ...
    weights              (years) float64 0.1565 0.128 0.1309 0.1633 0.1576 ...
    pvalues              (latitudes, longitudes) float64 nan nan nan nan nan ...



In [39]:

    
compos.save_to_file('/Users/nicolasf/Desktop/ersst_sst.nc')



In [45]:

    
!ncdump -h /Users/nicolasf/Desktop/ersst_sst.nc









    



netcdf ersst_sst {
dimensions:
	years = 7 ;
	latitudes = 89 ;
	longitudes = 180 ;
variables:
	double composite_sample(years, latitudes, longitudes) ;
		composite_sample:_FillValue = -999.9 ;
		composite_sample:missing_value = -999.9 ;
	double composite_anomalies(latitudes, longitudes) ;
		composite_anomalies:_FillValue = -999.9 ;
		composite_anomalies:missing_value = -999.9 ;
	double pvalues(latitudes, longitudes) ;
	float latitudes(latitudes) ;
		string latitudes:units = "degrees_north" ;
		string latitudes:long_name = "Latitudes" ;
		string latitudes:axis = "Y" ;
	int years(years) ;
	float longitudes(longitudes) ;
		string longitudes:units = "degrees_east" ;
		string longitudes:long_name = "Longitudes" ;
		string longitudes:axis = "X" ;
}



In [ ]:

	d_anomalies	cat	weights
time
1982-02-15	0.638992	WA	0.156455
1997-02-15	0.747619	WA	0.128005
1999-02-15	0.736547	WA	0.130905
2000-02-15	0.612678	WA	0.163346
2001-02-15	0.565475	WA	0.157625
2006-02-15	0.429462	WA	0.122003
2009-02-15	0.504521	WA	0.141661

illustrates the use of the proxy, analogs and scalar_plot classes

import the development version of paleopy

example 1: defines one proxy

defines the folder where the JSON files are (for the datasets) and where to save the proxy JSON files

instantiates a proxy instance

find the analogs, calls the internal functions necessary to get there

now we can print the information about the proxy and the derived information related to the extraction

those are the extracted analog years

analogs is an attribute of the proxy: it's a pandas dataframe containing info about the analog years

Adding outfile=True creates the json file containing the proxy information, with filename self.pfname

plot the seasonal time-series of anomalies and indicates the analog years

instantiate the analogs class with the proxy object and the dataset + variable to composite

a smaller domain (NZ)

now one plot per year in the analog sample

Tmean composite from VCSN

saves to disk the xray Dataset containing the composite sample, anomalies and associated metadata

Now with another dataset (SST from ERSST)

now maps: global domain, Equidistant Cylindrical projection, significance level 90%

illustrates the use of the `proxy`, `analogs` and `scalar_plot` classes

`analogs` is an attribute of the proxy: it's a `pandas` dataframe containing info about the analog years

Adding `outfile=True` creates the json file containing the proxy information, with filename `self.pfname`