如何在 plotly scattergeo 中自定义颜色条？

Question

I pulled some fire data from the NASA earthdata website (fires in south america) and plotted the data on a world map. I used a colorbar to display the brightness of each fire.

The variance in brightness of the fires does not correspond to the full colorscale range and most of the fires are in the same color (yellow). Here is my code:

import csv

from plotly.graph_objs import Scattergeo, Layout
from plotly import offline

filename = 'data/MODIS_C6_South_America_24h.csv'
with open(filename) as f:
    reader = csv.reader(f)
    header_row = next(reader)
    print(header_row)

    # Get latitudes, longitudes and brightness from this file.

    lats, lons, brights = [], [], []
    for row in reader:
        lat = float(row[0])
        lats.append(lat)
        lon = float(row[1])
        lons.append(lon)
        bright = float(row[2])
        brights.append(bright)

# Map the fires
data = [{
    'type': 'scattergeo',
    'lon': lons,
    'lat': lats,
    'marker': {
        'size': [1/30* bright for bright in brights],
        'color': brights,
        'colorscale': 'Inferno',
        'reversescale': True,
        'colorbar': {'title': 'Brightness'},
    },
}]
my_layout = Layout(title='South America Fires\npast 24 hours')

fig = {'data': data, 'layout': my_layout}
offline.plot(fig, filename='south_america_fires.html')

Can I somehow change the limits of the colorscale so that the markers have a broader color range and are better distinguishable? Or is there better strategy?

Answer 1

The variance in brightness of the fires does not correspond to the full colorscale range

Yes, they do. Just have look at a simpler visualization of your data:

Plot 1: Seaborn distribution plot

Code 1: Seaborn distribution plot

import seaborn as sns
import numpy as np
sns.set(color_codes=True)
sns.distplot(tuple(brights))

Your plot just ends up looking as it does for three reasons:

1. There are many observations around brightness = 330
2. There are very few observations of the brighter fires
3. And most importanntly, the markers are added to the plot in the order they appear in your dataset.

So if you just sort the data to make sure that the brighter fires aren't covered by the less brighter fires, you'll get this:

*Plot 2: Sorted brights using brights.sort()

I think that should take care of this:

[...] so that the markers have a broader color range and are better distinguishable?

So there's really no need to worry about this:

Can I somehow change the limits of the colorscale [...]

You could consider a log recoding of your data as well. I tested it, but it didn't make much of a visual difference. And do note that I removed the 'size': [1/60* bright for bright in brights] part. I think plot 2 looked better than this:

Complete code:

import csv

from plotly.graph_objs import Scattergeo, Layout
from plotly import offline

filename = 'C:\\pySO\\MODIS_C6_South_America_24h.csv'
with open(filename) as f:
    reader = csv.reader(f)
    header_row = next(reader)
    print(header_row)

# Get latitudes, longitudes and brightness from this file.

    lats, lons, brights = [], [], []
    for row in reader:
        lat = float(row[0])
        lats.append(lat)
        lon = float(row[1])
        lons.append(lon)
        bright = float(row[2])
        brights.append(bright)

brights.sort()

# Map the fires
data = [{
    'type': 'scattergeo',
    'lon': lons,
    'lat': lats,
    'marker': {
        #'size': [1/60* bright for bright in brights],
        'color': brights,
        #'color': brights.sort(),
        'colorscale': 'Inferno',
        'reversescale': True,
        'colorbar': {'title': 'Brightness'},
    },
}]
my_layout = Layout(title='South America Fires\npast 24 hours')

fig = {'data': data, 'layout': my_layout}
offline.plot(fig, filename='south_america_fires.html')