This tutorial will teach you how to plot a loggerhead turtle track on a map. That turtle was raised in captivity in Japan, then tagged and released on 05/04/2005 in the Central Pacific. It transmitted for over 3 years and went all the way to the Southern tip of Baja California!

The track data can be downloaded here.

Then we'll extract SST and chlorophyll concentration at each location along the track, and plot the data.

This tutorial is also available as a Jupyter notebook.

Load packages

import pandas as pd
import numpy as np
import urllib.request
import xarray as xr
import netCDF4 as nc
import time
from matplotlib import pyplot as plt
from matplotlib.colors import LinearSegmentedColormap,BoundaryNorm,Normalize
from mpl_toolkits.basemap import Basemap
from datetime import date,datetime

np.warnings.filterwarnings('ignore')

Let's load the track data:

df=pd.read_csv('25317_05.dat') 
df.head()

first 5 rows of track data
	mean_lon	mean_lat	year	month	day
0	176.619433	32.678728	2005	5	4
1	176.783786	32.755451	2005	5	5
2	177.086095	32.870337	2005	5	6
3	177.523857	32.859396	2005	5	7
4	178.058145	32.674011	2005	5	8

Let's plot the track on a map

import cartopy.crs as ccrs
import cartopy.feature as cfeature
from cartopy.mpl.ticker import LongitudeFormatter, LatitudeFormatter
import matplotlib.pyplot as plt


plt.figure(figsize=(14, 10))

# Label axes of a Plate Carree projection with a central longitude of 180:

ax1 = plt.subplot(211, projection=ccrs.PlateCarree(central_longitude=180))
ax1.set_extent([120,255, 15, 55], ccrs.PlateCarree())
ax1.add_feature(cfeature.LAND, facecolor='0.6')
ax1.coastlines()
ax1.set_xticks(range(120,255,20), crs=ccrs.PlateCarree())
ax1.set_yticks(range(20,50,10), crs=ccrs.PlateCarree())
lon_formatter = LongitudeFormatter(zero_direction_label=True)
lat_formatter = LatitudeFormatter()
ax1.xaxis.set_major_formatter(lon_formatter)
ax1.yaxis.set_major_formatter(lat_formatter)

x,y = list(df.mean_lon),list(df.mean_lat)
ax1=plt.plot(x,y,transform=ccrs.PlateCarree(),color='k')
ax1=plt.plot(x[0],y[0],marker='v',color='r',transform=ccrs.PlateCarree(),markersize=10)
ax1=plt.plot(x[-1],y[-1],marker='^',color='g',transform=ccrs.PlateCarree(),markersize=10)
plt.title('Turtle #25317', fontsize=20)

plt.show()

Now let's extract data along the track

We are going to grab data from ERDDAP, so we need to set up the ERDDAP URLs using their datasets IDs and the name of the variables we are interested in. Note that we are requesting the data as .csv

Chlorophyll-a concentration

We'll use the ESA OC-CCI product (https://climate.esa.int/en/projects/ocean-colour/) which is a merged product combining many ocean color sensors to create a long time series.

Ideally, we would work with daily data since we have one location per day. But chlorophyll data is severely affected by clouds (i.e. lots of missing data), so you might need to use weekly or even monthly data to get sufficient non-missing data.

Run both of them, and plot a time-series of each to compare (as a separate exercise).

ESA_w = "https://oceanwatch.pifsc.noaa.gov/erddap/griddap/esa-cci-chla-8d-v6-0.csv?chlor_a" 
ESA_m = "https://oceanwatch.pifsc.noaa.gov/erddap/griddap/esa-cci-chla-monthly-v6-0.csv?chlor_a"

lon=df.mean_lon 
lat=df.mean_lat

We need to format the dates in a way that ERDDAP understands, i.e. 2010-12-15

dates=[] 
for i in range(len(df.month)): 
   dates.append(date(df.year[i],df.month[i],df.day[i]).strftime('%Y-%m-%d'))

dates[0] '2005-05-04'

For each date and location, we'll extract a value of CHL or SST. To do this, we need to pass those parameters (which dataset, which date, which lon, and which lat) to ERDDAP by building the URL.

This can take a long time to run (about 15 mins), we are making 1200+ requests to a remote server. For the purpose of the exercise, you can just run the below code on the first 100 points of the turtle track.

start_time=time.time()

col_names = ["date","matched_lat","matched_lon","matched_chla"] 
tot=pd.DataFrame(columns = col_names)

for i in range(len(dates)):
#for i in range(100):
   print(i,len(dates))

   #this is where the URL is built:
   url=ESA_m+"[("+str(dates[i])+"):1:("+str(dates[i])+")][("+str(lat[i])+"):1:("+str(lat[i])+")][("+str(lon[i])+"):1:("+str(lon[i])+")]"

   new=pd.read_csv(url,skiprows=1)
   new.columns=col_names
   tot=tot.append(new,ignore_index=True)

end_time=time.time()

print("total time = %g mins" % ((end_time-start_time)/60.)) total time = 14.4734 mins

tot.head()

first 5 rows of track data and matched chlorophyll concentration values
	date	matched_lat	matched_lon	matched_chla
0	2005-05-16T12:00:00Z	32.687500	176.60417	0.147827
1	2005-05-16T12:00:00Z	32.770832	176.77083	0.168947
2	2005-05-16T12:00:00Z	32.854168	177.10417	0.258081
3	2005-05-16T12:00:00Z	32.854168	177.52083	0.171364
4	2005-05-16T12:00:00Z	32.687500	178.06252	0.296886

We now have a value of monthly chlorophyll-a concentration for each location/date combination along the turtle track.

On your own!

Exercise 1: Repeat the steps above with a different dataset. For example, extract sea surface temperature data using the following dataset: https://oceanwatch.pifsc.noaa.gov/erddap/griddap/CRW_sst_v1_0.html
Exercise 2: Go to an ERDDAP of your choice, find a dataset of interest, generate the URL, copy it and edit the script above to run a match up on that dataset. To find other ERDDAP servers, you can use this search engine: http://erddap.com/

Note! some ERDDAPs are slower than others, so this could take a lot longer. If it takes too long, adjust the "for" loop to request data for only the first 100 days of our track.

Plot #2

Let's plot the track, color coded using values of monthly chlorophyll concentration.

Let's create a color scale

jet=["blue", "#007FFF", "cyan","#7FFF7F", "yellow", "#FF7F00", "red", "#7F0000"]

Let's look at the range of log of monthly chlorophyll values:

np.min(np.log(tot.matched_chla)),np.max(np.log(tot.matched_chla))

(-2.8923874400191183, 2.136305926186285)

n, bins, patches=plt.hist(np.log(tot.matched_chla[~np.isnan(tot.matched_chla)]),50) 
plt.show()

The range of log(chl-a) is -2.9 to 2.2 but most of the values are between -2.9 and 0.

We use the log because the range of chlorophyll values can be pretty big, with lots of very low values, and a few very high values.

levs = np.append(np.arange(-2.9,0,0.1),2.2) 
cm = LinearSegmentedColormap.from_list('my_jet', jet, N=len(levs))

BoundaryNorm will force the colorbar to use the breaks in levs.

norm = BoundaryNorm(levs, len(levs))

#Setup the bounding box for the zoom and bounds of the map
bbox=[120 ,255, 15, 55]

plt.figure(figsize=(10,10))

# Label axes of a Plate Carree projection with a central longitude of 180:

ax1 = plt.subplot(211, projection=ccrs.PlateCarree(central_longitude=180))
ax1.set_extent([120,255, 15, 55], ccrs.PlateCarree())
ax1.add_feature(cfeature.LAND, facecolor='0.6')
ax1.coastlines()
ax1.set_xticks(range(120,255,20), crs=ccrs.PlateCarree())
ax1.set_yticks(range(20,50,10), crs=ccrs.PlateCarree())
lon_formatter = LongitudeFormatter(zero_direction_label=True)
lat_formatter = LatitudeFormatter()
ax1.xaxis.set_major_formatter(lon_formatter)
ax1.yaxis.set_major_formatter(lat_formatter)

# build and plot coordinates onto map
x,y = list(df.mean_lon),list(df.mean_lat)
ax1=plt.scatter(x,y,transform=ccrs.PlateCarree(),c=np.log(tot.matched_chla),cmap=cm,norm=norm)
ax1=plt.plot(x[0],y[0],marker='v',color='r',transform=ccrs.PlateCarree(),markersize=10)
ax1=plt.plot(x[-1],y[-1],marker='^',color='g',transform=ccrs.PlateCarree(),markersize=10)


#let's customize the color bar so the label reflect values of chl-a, not log(chl-a)
#we build levs2 to have the labels more spaced out than the values in levs
levs2=np.append(np.arange(-2.9,0,0.5),2.2)
cbar=plt.colorbar(ticks=levs2,shrink=0.75,aspect=10)
cbar.set_label("Chl a (mg/m^3))", size=15, labelpad=20)
#and set the labels to be exp(levs2)
cbar.ax.set_yticklabels(np.around(np.exp(levs2),2),size=10)

plt.title("Turtle #25317",size=15)
plt.show()

On your own!

Exercise 3: plot the track, color coded using values of monthly sea surface temperature. Note: you do not need to use a log scale for SST.