datasci/nuclear_explosions
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update article and rename to index.qmd

Browse source
This commit is contained in:
Marty Oehme 2024-07-03 16:29:23 +02:00
parent fed35fcfd2
commit 7d1d929b0e
Signed by: Marty
GPG key ID: EDBF2ED917B2EF6A
1 changed files with 145 additions and 76 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

221
nuclear_explosions.qmd → index.qmd
View file

@ -1,7 +1,5 @@
---
title: Nuclear Explosions
author: Marty Oehme
output-dir: out
references:
- type: techreport
id: Bergkvist2000
@ -18,16 +16,6 @@ references:
title: "Nuclear Explosions 1945 - 1998"
page: 1-42
issn: 1104-9154
format:
html:
toc: true
code-fold: true
typst:
toc: true
echo: false
docx:
toc: true
echo: false
---
```{python}
@ -43,7 +31,21 @@ from matplotlib import pyplot as plt
sns.set_theme(style="darkgrid")
sns.set_context("notebook")
cp=sns.color_palette()
country_colors = {
"US": cp[0],
"USSR": cp[3],
"France": cp[6],
"UK": cp[5],
"China": cp[4],
"India": cp[1],
"Pakistan": cp[2],
}
```
```{python}
# | label: data-prep
# | echo: false
schema_overrides = (
{
col: pl.Categorical
@ -53,13 +55,31 @@ schema_overrides = (
| {col: pl.String for col in ["year", "name"]}
)
cty_alias = {
"PAKIST": "Pakistan",
"FRANCE": "France",
"CHINA": "China",
"INDIA": "India",
"USA": "US",
}
def cty_replace(name: str) -> str:
if name in cty_alias:
return cty_alias[name]
return name
df = (
pl.read_csv(
"data/nuclear_explosions.csv",
schema_overrides=schema_overrides,
null_values=["NA"],
)
.with_columns(date=pl.col("year").str.strptime(pl.Date, "%Y"))
.with_columns(
date=pl.col("year").str.strptime(pl.Date, "%Y"),
country=pl.col("country").map_elements(cty_replace, return_dtype=pl.String),
)
.with_columns(year=pl.col("date").dt.year().cast(pl.Int32))
)
```
@ -69,7 +89,7 @@ df = (
The following is a re-creation and expansion of some of the graphs found in the
@Bergkvist2000 produced report on nuclear explosions between 1945 and 1998. It
is primarily a reproduction of key plots from the original report.
Additionally, it serves as a exercise in plotting with the python library
Additionally, it serves as an exercise in plotting with the python library
seaborn and the underlying matplotlib. Lastly, it approaches some less well
tread territory for data science in the python universe as it uses the python
library polars-rs for data loading and transformation. All the code used to
@ -77,9 +97,9 @@ transform the data and create the plots is available directly within the full
text document, and separately as well. PDF and Docx formats are available with
the plotting results only.
Their original purpose was the collection of a long list of all the nuclear
explosions occurring between those years, as well as analysing the responsible
nations, tracking the types and purposes of the explosions, as well as
The authors' original purpose was the collection of a long list of all the
nuclear explosions occurring between those years, as well as analysing the
responsible nations, tracking the types and purposes of the explosions and
connecting the rise and fall of nuclear explosion numbers to historical events
throughout.
@ -90,13 +110,13 @@ throughout.
## Nuclear devices
There are two main kinds of nuclear device: those based entirely, on fission,
or the splitting of heavy atomic nucleii (previously known as atomic devices)
or the splitting of heavy atomic nuclei (previously known as atomic devices)
and those in which the main energy is obtained by means of fusion, or of -light
atomic nucleii (hydrogen or thermonuclear devices). A fusion explosion must
atomic nuclei (hydrogen or thermonuclear devices). A fusion explosion must
however be initiated with the help of a fission device. The strength of a
fusion explosion can be practically unlimited. The explosive power of a
nuclear explosion is expressed in ktlotons, (kt) or megatons (Mt), which
correspond to 1000 and i million'tonnes, of conventional explosive (TNT),
nuclear explosion is expressed in kilotons, (kt) or megatons (Mt), which
correspond to 1000 and 1 million tonnes, of conventional explosive (TNT),
respectively.
[@Bergkvist2000, 6]
@ -108,8 +128,9 @@ each country had explode, seen in @tbl-yields.
```{python}
# | label: tbl-yields
# | tbl-cap: "Total number and yields of explosions"
# | output: asis
from great_tables import GT, md
from great_tables import GT
df_yields = (
df.select(["country", "id_no", "yield_lower", "yield_upper"])
@ -119,11 +140,17 @@ df_yields = (
pl.col("id_no").len().alias("count"),
pl.col("yield_avg").sum(),
)
# .with_columns(country=pl.col("country").cast(pl.String).str.to_titlecase())
.with_columns(yield_per_ex=pl.col("yield_avg") / pl.col("count"))
.sort("count", descending=True)
)
(
us_row = df_yields.filter(pl.col("country") == "US")
yields_above_us = df_yields.filter(
pl.col("yield_per_ex") > us_row["yield_per_ex"]
).sort("yield_per_ex", descending=True)
assert len(yields_above_us) == 3, "Yield per explosion desc needs updating!"
tab=(
GT(df_yields)
.tab_source_note(
source_note="Source: Author's elaboration based on Bergkvist and Ferm (2000)."
@ -131,22 +158,33 @@ df_yields = (
.tab_spanner(label="Totals", columns=["count", "yield_avg"])
.tab_stub(rowname_col="country")
.tab_stubhead(label="Country")
.cols_label(
count="Count",
yield_avg="Yield in kt",
)
.cols_label(count="Count", yield_avg="Yield in kt", yield_per_ex="Yield average")
.fmt_integer(columns="count")
.fmt_number(columns="yield_avg", decimals=1)
.fmt_number(columns="yield_per_ex", decimals=1)
)
del df_yields
tab
```
It is interesting to note that while the US undoubtedly had the highest raw
number of explosions, it did not, in fact, output the highest estimated
detonation yields.
In fact, `{python} len(yields_above_us)` countries have a higher average
explosion yield per detonation than the US:
`{python} yields_above_us[0]["country"].item()` leads with an average of
`{python} f"{yields_above_us[0]['yield_per_ex'].item():.2f}"` kt,
before
`{python} yields_above_us[1]["country"].item()` with an average of
`{python} f"{yields_above_us[1]['yield_per_ex'].item():.2f}"` kt.
## Numbers over time
When investigating the nuclear explosions in the world, let us first start by
looking at how many explosions occurred each year in total. This hides the
specific details of who was responsible and which types were involved but
instead paints a much stronger picture of the overall dimension of nuclear
testing, as can be seen in @fig-total.
In the examination of global nuclear detonations, our initial focus shall be
quantifying the annual incidence of the events in aggregate. While it obscures
the specific details of the responsible nations and which diversity of types
tested, it instead paints a much stronger picture of the overall abstracted
dimension of nuclear testing throughout history, as depicted in @fig-total.
```{python}
# | label: fig-total
@ -171,17 +209,29 @@ with sns.axes_style(
del per_year
```
As we can see, the numbers of explosions rise increasingly towards 1957 and
sharply until 1958, before dropping off for a year in 1959. The reasons for
this drop are not entirely clear, but it is very likely that the data are
simply missing for these years.
<!-- FIXME: The reasons for this are a non-proliferation pact, in article -->
As we can see, the number of explosions rises increasingly towards 1957 and
sharply until 1958, before dropping off for a year in 1959. The reason for this
drop should primarily be found in the start of the 'Treaty of Test Ban' which
put limits and restraints on the testing of above-ground nuclear armaments, as
discussed in the original article. Above all the contract signals the
prohibition of radioactive debris to fall beyond a nation's respective
territorial bounds.
However, this contract should perhaps not be viewed as the only reason: With
political and cultural shifts throughout the late 1950s and early 1960s
increasingly focusing on the fallout and horror of nuclear warfare a burgeoning
public opposition to nuclear testing and instead a push towards disarmament was
taking hold. The increased focus on the space race between the US and USSR may
have detracted from the available funds, human resources and agenda attention
for nuclear testing. Lastly, with nuclear testing policies strongly shaped by
the political dynamics of the Cold War, a period of improved diplomatic
relations such as the late 1950s prior to the Cuban missile crisis may directly
affect the output of nuclear testing facilities between various powers.
<!-- TODO: Extract exact numbers from data on-the-fly -->
There is another, very steep, rise in 1962 with over 175 recorded explosions,
before an even sharper drop-off the following year down to just 50 explosions.
Afterwards the changes appear less sharp and the changes remain between 77 and
Afterward the changes appear less sharp and the changes remain between 77 and
24 explosions per year, with a slight downward tendency.
While these numbers show the overall proliferation of nuclear power, let us now
@ -191,6 +241,7 @@ of explosions over time by country can be seen in @fig-percountry.
```{python}
# | label: fig-percountry
# | fig-cap: "Nuclear explosions by country, 1945-98"
keys = df.select("date").unique().join(df.select("country").unique(), how="cross")
per_country = keys.join(
df.group_by(["date", "country"], maintain_order=True).len(),
@ -199,7 +250,7 @@ per_country = keys.join(
coalesce=True,
).with_columns(pl.col("len").fill_null(0))
g = sns.lineplot(data=per_country, x="date", y="len", hue="country")
g = sns.lineplot(data=per_country, x="date", y="len", hue="country", palette=country_colors)
g.set_xlabel("Year")
g.set_ylabel("Count")
plt.setp(
@ -211,14 +262,14 @@ del per_country
Once again we can see the visibly steep ramp-up to 1962, though it becomes
clear that this was driven both by the USSR and the US. Of course the graph
also makes visible the sheer unmatched number of explosions emenating from both
also makes visible the sheer unmatched number of explosions emanating from both
of the countries, with only France catching up to the US numbers and China
ultimately overtaking them in the 1990s.
However, here it also becomes more clear how the UK was responsible for some
early explosions in the late 1950s and early 1960s already, as well as the rise
in France's nuclear testing from the early 1960s onwards to around 1980, before
slowly decreasing in intensity afterwards.
slowly decreasing in intensity afterward.
Let us turn to a cross-cut through the explosions in @fig-groundlevel, focusing
on the number of explosions that have occurred underground and above-ground
@ -273,6 +324,7 @@ with sns.axes_style("darkgrid", {"xtick.bottom": True, "ytick.left": True}):
hue="country",
multiple="stack",
binwidth=365,
palette=country_colors,
)
g.xaxis.set_major_locator(mdates.YearLocator(base=5))
@ -293,25 +345,19 @@ shift from above-ground to underground tests, starting with the year 1962.
## Locations
Finally, let's view a map of the world with the explosions marked.
Finally, let's view a map of the world with the explosions marked, separated by country.
::: {.content-visible when-format="html"}
Hovering over individual explosions will show their year
while a click will open more information in a panel.
:::
The map can be seen in @fig-worldmap.
```{python}
# | label: fig-worldmap
# | fig-cap: "World map of nuclear explosions, 1945-98"
# | label: worldmap-setup
# | output: false
import folium
import geopandas as gpd
from shapely.geometry import Point
def set_style() -> pl.Expr:
return (
pl.when(pl.col("country") == "USSR")
.then(pl.lit({"color": "red"}, allow_object=True))
.otherwise(pl.lit({"color": "blue"}, allow_object=True))
)
geom = [Point(xy) for xy in zip(df["longitude"], df["latitude"])]
# df_pd = df.with_columns(style=set_style()).to_pandas().set_index("date")
df_pd = df.with_columns().to_pandas().set_index("date")
gdf = gpd.GeoDataFrame(
df_pd,
@ -320,25 +366,18 @@ gdf = gpd.GeoDataFrame(
)
del df_pd
country_colors = {
"USA": "darkblue",
"USSR": "darkred",
"FRANCE": "pink",
"UK": "black",
"CHINA": "purple",
"INDIA": "orange",
"PAKIST": "green",
}
def rgb_to_hex(rgb: tuple[float,float,float]) -> str:
return "#" + "".join([format(int(c*255), '02x') for c in rgb])
m = folium.Map(tiles="cartodb positron")
for country in country_colors.keys():
fg = folium.FeatureGroup(name=country, show=True).add_to(m)
folium.GeoJson(
gdf[gdf["country"].str.contains(country)],
gdf[gdf["country"] == country],
name="Nuclear Explosions",
marker=folium.Circle(radius=3, fill_opacity=0.4),
style_function=lambda x: {
"color": country_colors[x["properties"]["country"]],
"color": rgb_to_hex(country_colors[x["properties"]["country"]]),
"radius": (
x["properties"]["magnitude_body"]
if x["properties"]["magnitude_body"] > 0
@ -368,15 +407,45 @@ for country in country_colors.keys():
),
).add_to(fg)
folium.LayerControl().add_to(m)
```
::: {#fig-worldmap}
:::: {.content-visible when-format="html"}
```{python}
# | label: worldmap-html
m
```
That is all for now.
There are undoubtedly more explorations to undertake,
but this is it for the time being.
::::
<!-- Ideas TODO:
- do not just use 'count' of explosions but yields
- compare number to yields for ctrys
- count up total number per country in table
-->
:::: {.content-visible unless-format="html" width=80%}
```{python}
# | label: worldmap-non-html
# ENSURE SELENIUM IS INSTALLED
m.png_enabled = True
m
```
::::
World map of nuclear explosions, 1945-98
:::
While there are undoubtedly more aspects of the data that provide interesting
patterns for analysis, this shall be the extent of review for the time being
for this reproduction.
We can see how the combination of python polars and seaborn makes the process
relatively approachable, understandable and, combined with the rendering output
by quarto, fully reproducible.
Additionally, we can see how additional projects can be included to produce
interactive graphs and maps with tools such as folium and geopandas.
## References
::: {#refs}
:::