Figure number of images per country
Install marss2l package
pip install marss2l
%%time
import matplotlib
from marss2l.utils import setup_stream_logger
import logging
from huggingface_hub import hf_file_system
from marss2l.huggingface import REPO_ID
logger = logging.getLogger(__name__)
setup_stream_logger(logger)
matplotlib.rcParams['mathtext.fontset'] = 'stix'
matplotlib.rcParams['font.family'] = 'STIXGeneral'
C0 = "#648FFF"
C1 = "#785EF0"
C2 = "#DC267F"
C3 = "#FE6100"
C4 = "#FFB000" # #FFB000
import os
fs = hf_file_system.HfFileSystem()
os.makedirs("figures", exist_ok=True)
%%time
from marss2l import loaders
LOCS_TRAINING_IN_DISTRIBUTION = loaders.LOCS_TRAINING_ABLATION
csv_path = f"datasets/{REPO_ID}/validated_images_all.csv"
dataframe_data_traintest = loaders.read_csv(csv_path,
add_columns_for_analysis=True,
fs=fs, split="all", add_case_study=True,
add_loc_type=True)
dataframe_data_traintest["country"] = dataframe_data_traintest.apply(lambda row: "Offshore" if row.offshore else row.country,
axis=1)
# Count number of images with more than one plume
# from shapely import wkt
# import geopandas as gpd
# geometries = dataframe_data_traintest.plume.apply(wkt.loads)
# dataframe_data_traintest["nplumes"] = geometries.apply(lambda geom: 0 if geom.is_empty else len(geom.geoms))
# dataframe_data_traintest.nplumes.value_counts()
dataframe_data_traintest = dataframe_data_traintest[~dataframe_data_traintest.location_name.isin(loaders.LOCATIONS_CONTROL_RELEASES)].copy()
rename_splits = {"test_2023": "test", "train_2023": "train", "val_2023": "val"}
dataframe_data_traintest["split_name"] = dataframe_data_traintest["split_name"].apply(lambda x: rename_splits[x] if x in rename_splits else x)
nplumes_total = dataframe_data_traintest.loc[dataframe_data_traintest.split_name.isin(["train","val","test"]),"isplume"].sum()
nimages_total = dataframe_data_traintest.loc[dataframe_data_traintest.split_name.isin(["train","val","test"])].shape[0]
nlocs_total = dataframe_data_traintest.location_name.nunique()
print(f"In total {nimages_total} images containing {nplumes_total} emissions over {nlocs_total} distinct emitters")
images_by_yearmonth = dataframe_data_traintest.groupby(["country","year_month"])["isplume"].agg(["count","sum"]).rename({"count": "# images", "sum": "# plumes"}, axis=1)
images_by_yearmonth["# locs"] = dataframe_data_traintest.groupby(["country","year_month"])["location_name"].nunique()
images_by_yearmonth = images_by_yearmonth.reset_index()
images_by_yearmonth
images_by_yearmonth_casestudy = dataframe_data_traintest.groupby(["case_study","year_month"])["isplume"].agg(["count","sum"]).rename({"count": "# images", "sum": "# plumes"}, axis=1)
images_by_yearmonth_casestudy["# locs"] = dataframe_data_traintest.groupby(["case_study","year_month"])["location_name"].nunique()
images_by_yearmonth_casestudy = images_by_yearmonth_casestudy.reset_index()
images_by_yearmonth_casestudy
Time series total number of images for Figure 1b
from marss2l.plot import images_by_month
from importlib import reload
import matplotlib.pyplot as plt
from datetime import datetime
reload(images_by_month)
fig, ax = plt.subplots(1, 1, figsize=(11, 2.75))
images_by_month.plot_images_by_month(dataframe_data_traintest, ax, datetime(2018, 1, 1),
step_in_months=3,
datetime_end=datetime(2025, 2, 1))
plt.tight_layout()
plt.savefig("figures/fig1_ts_images_detections.pdf")
import pandas as pd
pd.set_option('display.max_rows', 300)
Number of images in case studies
All data
case_studies_plot = loaders.ORDER_CASE_STUDIES[:6]
ncountries = len(case_studies_plot)
fig, ax = plt.subplots(ncountries,1, figsize=(11, ncountries*2.75), sharex=True, tight_layout=True)
for i, country in enumerate(case_studies_plot):
df_plot = dataframe_data_traintest[dataframe_data_traintest.case_study == country]
images_by_month.plot_images_by_month(df_plot, ax=ax[i],
loc_legend="upper left",
plot_legend= i == 0,
datetime_start=datetime(2018,1,1),
datetime_end=datetime(2025,1,1))
ax[i].set_title(country)
plt.savefig("figures/fig1_ts_images_detections_by_country_1.pdf")
case_studies_plot = loaders.ORDER_CASE_STUDIES[6:]
ncountries = len(case_studies_plot)
fig, ax = plt.subplots(ncountries,1, figsize=(11, ncountries*2.75), sharex=True, tight_layout=True)
for i, country in enumerate(case_studies_plot):
df_plot = dataframe_data_traintest[dataframe_data_traintest.case_study == country]
images_by_month.plot_images_by_month(df_plot, ax=ax[i],
loc_legend="upper left",
plot_legend= i == 0,
datetime_start=datetime(2018,1,1),
datetime_end=datetime(2025,1,1))
ax[i].set_title(country)
plt.savefig("figures/fig1_ts_images_detections_by_country_2.pdf")
Distribution of test data by case study
from matplotlib import gridspec
import seaborn as sns
# Create a figure with a specified size
fig = plt.figure(figsize=(10,2.5*len(loaders.ORDER_CASE_STUDIES)),
layout="constrained")
# Create a GridSpec with 2 rows and 2 columns, specifying the width ratios
gs = gridspec.GridSpec(len(loaders.ORDER_CASE_STUDIES), 2,
width_ratios=[1, 4], figure=fig)
# Create subplots using the GridSpec and share the x-axis between specific subplots
ax1 = fig.add_subplot(gs[0, 0]) # Narrower subplot
ax2 = fig.add_subplot(gs[0, 1]) # Wider subplot
for i, country in enumerate(loaders.ORDER_CASE_STUDIES):
df_plot_hist_fluxrate = dataframe_data_traintest[(dataframe_data_traintest.split_name == "test") & (dataframe_data_traintest.case_study == country)]
if i == 0:
ax1iter = ax1
ax2iter = ax2
else:
ax1iter = fig.add_subplot(gs[i, 0], sharex=ax1) # Narrower subplot, sharing x-axis with ax1
ax2iter = fig.add_subplot(gs[i, 1], sharex=ax2) # Wider subplot, sharing x-axis with ax2
sns.countplot(data=df_plot_hist_fluxrate[df_plot_hist_fluxrate.isplume],
x="interval_ch4_fluxrate_str", ax=ax2iter,color=C1)
sns.countplot(data=df_plot_hist_fluxrate, x="isplume", ax=ax1iter,color=C1)
ax1iter.set_ylabel("# images")
ax2iter.set_ylabel("# plumes")
ax2iter.set_title(country)
# if i != len(loaders.ORDER_CASE_STUDIES) -1:
# ax2iter.xaxis.set_visible(False)
# ax1iter.xaxis.set_visible(False)
# ax2iter.grid(axis="x")
ax2iter.set_xticklabels(ax2iter.get_xticklabels(), rotation=30)
ax2iter.set_xlabel("")
ax1iter.set_xlabel("")
plt.xticks(rotation=30)
ax1.xaxis.set_visible(False)
# ax2.xaxis.set_visible(False)
ax2iter.set_xlabel("Flux rate (t/h)")
ax1iter.set_xlabel("")
ax1iter.set_xticks([False, True],["no plume","plume"])
