Created: Tuesday 31 January 2017

Quantifying the uncertainity of a global fire limitation model using Bayesian inference

Part 3: Data visualisation

^1,*Douglas Kelley, ²Ioannis Bistinas, ^{3, 4}Chantelle Burton, ¹~~Tobias Marthews,~~ ⁵Rhys Whitley

¹Centre for Ecology and Hydrology, Maclean Building, Crowmarsh Gifford, Wallingford, Oxfordshire, United Kingdom
²Vrije Universiteit Amsterdam, Faculty of Earth and Life Sciences, Amsterdam, Netherlands
³Met Office United Kingdom, Exeter, United Kingdom
⁴Geography, University of Exeter, Exeter, United Kingdom
⁵Natural Perils Pricing, Commercial & Consumer Portfolio & Pricing, Suncorp Group, Sydney, Australia

Load libraries



In [1]:

    
# data munging and analytical libraries 
import re
import os
import numpy as np
import pandas as pd

# graphical libraries
import matplotlib.pyplot as plt
import matplotlib.pylab as pylab

# default plot settings for notebook
from IPython.core.pylabtools import figsize
figsize(11, 9)
plt.style.use('ggplot')

%matplotlib inline

dataPath = "../data/globfire.csv"

Import data



In [2]:

    
DATAPATH = os.path.expanduser(dataPath)

fd = pd.read_csv(DATAPATH)

Function definitions



In [3]:

    
def ignition(lightning, pasture_area, pop_density, kp, kd1):
    """
    Definition for the measure of ignition
    """
    return lightning + kp*pasture_area + kd1*pop_density

def supression(crop_area, pop_density, kd2):
    """
    Definition for the measure of fire supression
    """
    return crop_area + kd2*pop_density

def np_sigmoid(x, a, b):
    """
    Sigmoid function to describe limitation using tensor
    """
    return 1.0/(1.0 + np.exp(a*x + b))

Extra model covariates determined

Determine the ignition and supression covariates as per the model documentation



In [4]:

    
fd['ignite'] = ignition(fd["lightning_ignitions"].values, \
               fd["pasture"].values, \
               fd["population_density"].values, \
               500, 10)

fd['suppress'] = supression(fd["cropland"].values, \
                 fd["population_density"].values, \
                 10)

Data viz below

Base Data



In [5]:

    
def pltVsFire(x, xlab):
    plt.plot(x, fd.fire, 'o', alpha = 0.01)
    plt.ylabel("Burnt area fraction")
    plt.xlabel(xlab);



In [6]:

    
pltVsFire(fd.NPP, "NPP weight per area");
plt.xscale('log')



In [7]:

    
pltVsFire(fd.alpha, "Alpha")



In [8]:

    
pltVsFire(fd.lightning_ignitions, "lightning strikes")
plt.xscale('log')



In [9]:

    
pltVsFire(fd.cropland, "Crop cover")

Secondary Data



In [10]:

    
pltVsFire(fd.pasture, "Pasture")



In [11]:

    
pltVsFire(fd.population_density, "Population Density")
plt.xscale('log')



In [ ]: