Home

Awesome

SACPY -- A Python Package for Statistical Analysis of Climate

Sacpy, an effecient Statistical Analysis tool for Climate and Meteorology data.

Author : Zilu Meng

e-mail : zilumeng@uw.edu

github : https://github.com/ZiluM/sacpy

pypi : https://pypi.org/project/sacpy/

Document: https://zilum.github.io/sacpy/

version : 0.0.20

Why choose Sacpy?

Fast!

For example, Sacpy is more than 60 times faster than the traditional regression analysis with Python (see speed test). The following is the time spent performing the same task. Sacpy is fastest.

Turn to climate data customization!

Compatible with commonly used meteorological calculation libraries such as numpy and xarray.

Concise code

You can finish drawing a following figure with just seven lines of code. see examples of concise.

You can use SVD/MCA to get the image below easily.

Install and update

You can use pip to install.

pip install sacpy

Or you can visit https://gitee.com/zilum/sacpy/tree/main/dist to download .whl file, then

pip install .whl_file

update:

pip install --upgrade sacpy

or you can download .whl file and then install use pip install .whl_file.

Speed

As a comparison, we use the corr function in the xarray library, corrcoef function in numpy library, cdist in scipy, apply_func in xarray and for-loop. The time required to calculate the correlation coefficient between SSTA and nino3.4 for 50 times is shown in the figure below.

It can be seen that we are four times faster than scipy cdist, five times faster than xarray.corr, 60 times faster than forloop, 110 times faster than xr.apply_func and 200 times faster than numpy.corrcoef.

Moreover, xarray and numpy can not return the p value. We can simply check the pvalue attribute of sacpy to get the p value.

All in all, if we want to get p-value and correlation or slope, we only to choose Sacpy is 60 times faster than before.

Example

example1

Calculate the correlation between SST and nino3.4 index

import numpy as np
import scapy as scp
import matplotlib.pyplot as plt
import sacpy.Map # need cartopy or you can just not import
import cartopy.crs as ccrs

# load sst
sst = scp.load_sst()['sst']
# get ssta (method=1, Remove linear trend;method=0, Minus multi-year average)
ssta = scp.get_anom(sst,method=1)
# calculate Nino3.4
Nino34 = ssta.loc[:,-5:5,190:240].mean(axis=(1,2))
# regression
linreg = scp.LinReg(Nino34,ssta)
# plot
fig = plt.figure(figsize=[7, 3])
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
lon ,lat = ssta.lon , ssta.lat
# shading
m = ax.scontourf(lon,lat,linreg.corr)
# significant plot
n = ax.sig_plot(lon,lat,linreg.p_value,color="k",marker="..")
# initialize map
ax.init_map(stepx=50, ysmall=2.5)
# colorbar
plt.colorbar(m)
# save
plt.savefig("../pic/nino34.png",dpi=200)

Result(For a detailed drawing process, see example):

example2

multiple linear regression on Nino3.4 IOD Index and ssta pattern

import numpy as np
import scapy as scp
import matplotlib.pyplot as plt


# load sst
sst = scp.load_sst()['sst']
# get ssta (method=1, Remove linear trend;method=0, Minus multi-year average)
ssta = scp.get_anom(sst,method=1)
# calculate Nino3.4
Nino34 = ssta.loc[:,-5:5,190:240].mean(axis=(1,2))
# calculate IODIdex
IODW = ssta.loc[:,-10:10,50:70].mean(axis=(1,2))
IODE = ssta.loc[:,-10:0,90:110].mean(axis=(1,2))
IODI = +IODW - IODE
# get x
X = np.vstack([np.array(Nino34),np.array(IODI)]).T
# multiple linear regression
MLR = scp.MultLinReg(X,ssta)

# plot IOD's effect
import sacpy.Map
import cartopy.crs as ccrs


fig = plt.figure(figsize=[7, 3])
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
lon ,lat = ssta.lon , ssta.lat
m = ax.scontourf(lon,lat,MLR.slope[1])
# significant plot
n = ax.sig_plot(lon,lat,MLR.pv_i[1],color="k",marker="..")
# initialize map
ax.init_map(stepx=50, ysmall=2.5)
plt.colorbar(m)
plt.savefig("../pic/MLR.png",dpi=200)

Result(For a detailed drawing process, see example):

example3

What effect will ENSO have on the sea surface temperature in the next summer?

import numpy as np
import sacpy as scp
import matplotlib.pyplot as plt
import xarray as xr

# load sst
sst = scp.load_sst()['sst']
ssta = scp.get_anom(sst)

# calculate Nino3.4
Nino34 = ssta.loc[:,-5:5,190:240].mean(axis=(1,2))

# get DJF mean Nino3.4
DJF_nino34 = scp.XrTools.spec_moth_yrmean(Nino34,[12,1,2])

# get JJA mean ssta
JJA_ssta = scp.XrTools.spec_moth_yrmean(ssta, [6,7,8])

# regression
reg = scp.LinReg(DJF_nino34[:-1], JJA_ssta)
# plot
import cartopy.crs as ccrs
import sacpy.Map


fig = plt.figure(figsize=[7, 3])
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
lon ,lat = np.array(ssta.lon) , np.array(ssta.lat)
m = ax.scontourf(lon,lat,reg.slope)
n = ax.sig_plot(lon,lat,reg.p_value,color="k",marker="///")
ax.init_map(stepx=50, ysmall=2.5)
plt.colorbar(m)
plt.savefig("../pic/ENSO_Next_year_JJA.png",dpi=300)

Same as Indian Ocean Capacitor Effect on Indo–Western Pacific Climate during the Summer following El Niño (Xie et al.), the El Nino will lead to Indian ocean warming in next year JJA.

example4

EOF analysis

import sacpy as scp
import numpy as np
import matplotlib.pyplot as plt
# get data
sst = scp.load_sst()["sst"].loc[:, -20:30, 150:275]
ssta = scp.get_anom(sst)
# EOF
eof = scp.EOF(np.array(ssta))
eof.solve()
# get spartial pattern and pc
pc = eof.get_pc(npt=2)
pt = eof.get_pt(npt=2)
# plot
import cartopy.crs as ccrs
import sacpy.Map
lon , lat = np.array(ssta.lon) , np.array(ssta.lat)
fig = plt.figure(figsize=[15,10])
ax = fig.add_subplot(221,projection=ccrs.PlateCarree(central_longitude=180))
m1 = ax.scontourf(lon,lat,pt[0,:,:],cmap='RdBu_r',levels=np.linspace(-0.75,0.75,15),extend="both")
ax.scontour(m1,colors="black")
ax.init_map(ysmall=2.5)
# plt.colorbar(m1)
ax2 = fig.add_subplot(222)
ax2.plot(sst.time,pc[0])
ax3 = fig.add_subplot(223,projection=ccrs.PlateCarree(central_longitude=180))
m2 = ax3.scontourf(lon,lat,pt[1,:,:],cmap='RdBu_r',levels=np.linspace(-0.75,0.75,15),extend="both")
ax3.scontour(m2,colors="black")
ax3.init_map(ysmall=2.5)
ax4 = fig.add_subplot(224)
ax4.plot(sst.time,pc[1])
cb_ax = fig.add_axes([0.1,0.06,0.4,0.02])
fig.colorbar(m1,cax=cb_ax,orientation="horizontal")
plt.savefig("../pic/eof_ana.png",dpi=300)

example5

Mean value (Composite Analysis) t-test for super El Nino (DJF Nino3.4 > 1)


import sacpy as scp
import numpy as np
import matplotlib.pyplot as plt

sst = scp.load_sst()["sst"]
ssta = scp.get_anom(sst, method=0)
# get Dec Jan Feb SSTA
ssta_djf = scp.XrTools.spec_moth_yrmean(ssta,[12,1,2])
Nino34 = ssta_djf.loc[:, -5:5, 190:240].mean(axis=(1, 2))
# select year of Super El Nino
select = Nino34 >= 1
ssta_sl = ssta_djf[select]
mean, pv = scp.one_mean_test(ssta_sl)
# plot
import sacpy.Map
import cartopy.crs as ccrs
fig = plt.figure(figsize=[7, 3])
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
lon ,lat = np.array(ssta.lon) , np.array(ssta.lat)
m = ax.scontourf(lon,lat,mean)
n = ax.sig_plot(lon,lat,pv,color="k",marker="..")
ax.init_map(stepx=50, ysmall=2.5)
plt.colorbar(m)
plt.savefig("../pic/one_test.png")

Result:

example6

SVD(MCA) analysis.

import sacpy as scp
import xarray as xr
import matplotlib.pyplot as plt
import numpy as np
from xmca import array
import sacpy.Map
import cartopy.crs as ccrs

# load data
sst = scp.load_sst()['sst'].loc["1991":"2021", -20:30, 150:275]
ssta = scp.get_anom(sst)
u = scp.load_10mwind()['u']
v = scp.load_10mwind()['v']

uua = scp.get_anom(u)
vua = scp.get_anom(v)
uv = np.concatenate([np.array(uua)[...,np.newaxis],np.array(vua)[...,np.newaxis]],axis=-1)
# calculation
svd = scp.SVD(ssta,uv,complex=False)
svd.solve()
ptl, ptr = svd.get_pt(3)
pcl,pcr = svd.get_pc(3)
upt ,vpt = ptr[...,0] , ptr[...,1]
sst_pt = ptl
# plot progress, see example/SVD.ipynb

result:

examples of concise

If you want to plot example1's figure , you need write:

from cartopy.mpl.ticker import LongitudeFormatter, LatitudeFormatter
from matplotlib.ticker import MultipleLocator
import cartopy.crs as ccrs
plt.rc('font', family='Times New Roman', size=12)
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
m = ax.contourf(ssta.lon,ssta.lat,linreg.corr,
                cmap="RdBu_r",
                levels=np.linspace(-1, 1, 15),
                extend="both",
                transform=ccrs.PlateCarree())
n = plt.contourf(ssta.lon,ssta.lat,linreg.p_value,
                 levels=[0, 0.05, 1],
                 zorder=1,
                 hatches=['..', None],
                 colors="None",
                 transform=ccrs.PlateCarree())
xtk = np.arange(-180,181,60)
ax.set_xticks(xtk)
# ax.set_xticks(xtk,crs=ccrs.PlateCarree())
ax.set_yticks(np.arange(-50,51,20),crs=ccrs.PlateCarree())
ax.yaxis.set_major_formatter(LatitudeFormatter())
ax.xaxis.set_major_formatter(LongitudeFormatter(zero_direction_label=True))
ax.xaxis.set_minor_locator(MultipleLocator(10))
ax.yaxis.set_minor_locator(MultipleLocator(5))
ax.coastlines()
ax.set_aspect("auto")
plt.colorbar(m)

So troublesome!!!

But if you import sacpy.Map, you can easily write:

import sacpy.Map
import cartopy.crs as ccrs
fig = plt.figure(figsize=[7, 3])
ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
lon ,lat = ssta.lon , ssta.lat
m = ax.scontourf(lon,lat,rvalue)
n = ax.sig_plot(lon,lat,p,color="k",marker="..")
ax.init_map(stepx=50, ysmall=2.5)
plt.colorbar(m)

How wonderful, how concise !

Acknowledgements

Thank Prof. Feng Zhu (NUIST,https://fzhu2e.github.io/) for his guidance of this project.

Thank for Prof. Tim Li (University of Hawaii at Mānoa, http://iprc.soest.hawaii.edu/people/li.php) ,Prof. Lin Chen (NUIST, https://faculty.nuist.edu.cn/chenlin12/zh_CN/index.htm) and Dr. Ming Sun (NUIST) 's help.

Sepcial thanks: Lifei Lin (Sun Yat-sen University) 's repr_html.py to visualize class in jupyter!