LHC-Bulk-Ensemble-Plots.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "prescription-avenue",
   "metadata": {},
   "source": [
    "# Ensemble analysis\n",
    "\n",
    "Here I plot the \"bulk\" model behavior, including averages, envelopes, and so on."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "signed-slovenia",
   "metadata": {},
   "source": [
    "## Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "faced-hanging",
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "import datetime\n",
    "import os\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt \n",
    "import matplotlib as mpl\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina' \n",
    "\n",
    "# plotting parameters\n",
    "node_times = [2020, 2030, 2060, 2100, 2150, 2200]\n",
    "x_label = 'Year'\n",
    "\n",
    "# plotting lists\n",
    "color_list = ['#000000', '#E69F00', '#56B4E9', '#009E73', '#F0E442', '#0072B2', '#D55E00', '#CC79A7'] * 4\n",
    "marker_list = ['o', 's', 'P', '*', '<', 'v', '>', 'p'] * 4\n",
    "linestyle_list = ['solid'] * 8 + ['dashdot'] * 8 + ['dashed'] * 8 + ['dotted'] * 8\n",
    "\n",
    "# dictionaries of params\n",
    "cdds_params={'axes.linewidth': 3,\n",
    " 'axes.axisbelow': False,\n",
    " 'axes.edgecolor': 'black',\n",
    " 'axes.facecolor': 'None',\n",
    " 'axes.grid': False,\n",
    " 'axes.labelcolor': 'black',\n",
    " 'axes.spines.right': False,\n",
    " 'axes.spines.top': False,\n",
    " 'axes.titlesize': 20,\n",
    " 'axes.labelsize': 20,\n",
    " 'axes.titlelocation': 'left',\n",
    " 'figure.facecolor': 'white',\n",
    " 'figure.figsize': (18, 10),\n",
    " 'lines.solid_capstyle': 'round',\n",
    " 'lines.linewidth': 2.5,\n",
    " 'patch.edgecolor': 'w',\n",
    " 'patch.force_edgecolor': True,\n",
    " 'text.color': 'black',\n",
    " 'legend.frameon': False,\n",
    " 'xtick.bottom': True,\n",
    " 'xtick.major.width': 3,\n",
    " 'xtick.major.size': 6,\n",
    " 'xtick.color': 'black',\n",
    " 'xtick.direction': 'out',\n",
    " 'xtick.top': False,\n",
    " 'ytick.color': 'black',\n",
    " 'ytick.direction': 'out',\n",
    " 'ytick.left': True,\n",
    " 'ytick.right': False,\n",
    " 'ytick.color' : 'black',\n",
    " 'ytick.major.width': 3, \n",
    " 'ytick.major.size': 6,\n",
    " 'axes.prop_cycle': plt.cycler(linestyle=linestyle_list),\n",
    " 'font.size': 16,\n",
    " 'font.family': 'serif'}\n",
    "\n",
    "plt.rcParams.update(cdds_params)\n",
    "\n",
    "# make base filename\n",
    "today = datetime.datetime.now()\n",
    "year = str(today.year)\n",
    "day = str(today.day)\n",
    "month = str(today.month)\n",
    "\n",
    "basefile = ''.join([month, '-', day, '-', year, '-'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "numerous-diagnosis",
   "metadata": {},
   "source": [
    "## Save figs?!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "stock-holmes",
   "metadata": {},
   "outputs": [],
   "source": [
    "save_figs = True"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "alleged-manual",
   "metadata": {},
   "source": [
    "## Import parameter values and data\n",
    "\n",
    "This might look sloppy, but it includes a cleaning procedure for the data. Some runs of the model will have unusually high utilities and as a result weird output. We don't think this is because of the model being broken, but rather our computational resources are a little limited, so we can only run the optimization for so long and still get 1000 time series. So we clean out unusually high utilities and go from there."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "monthly-trader",
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_nans(ssp_paths_and_util, exp_m, temp_paths, co2_paths, dam_paths, params):\n",
    "    c2020, c2030, c2060, c2100, c2150, c2200, util = ssp_paths_and_util.T\n",
    "    paths = ssp_paths_and_util[:, :-1]\n",
    "    unphys_util_vals = util > 15\n",
    "    c2020 = c2020[~unphys_util_vals]\n",
    "    c2030 = c2030[~unphys_util_vals]\n",
    "    c2060 = c2060[~unphys_util_vals]\n",
    "    c2100 = c2100[~unphys_util_vals]\n",
    "    c2150 = c2150[~unphys_util_vals]\n",
    "    c2200 = c2200[~unphys_util_vals]\n",
    "    util = util[~unphys_util_vals]\n",
    "    paths = paths[~unphys_util_vals]\n",
    "    params = params[~unphys_util_vals]\n",
    "    exp_m = exp_m[~unphys_util_vals]\n",
    "    temp = temp_paths[~unphys_util_vals]\n",
    "    co2 = co2_paths[~unphys_util_vals]\n",
    "    dam = dam_paths[~unphys_util_vals]\n",
    "    return paths[~np.isnan(util)], c2020[~np.isnan(util)], c2030[~np.isnan(util)], c2060[~np.isnan(util)], c2100[~np.isnan(util)], c2150[~np.isnan(util)], c2200[~np.isnan(util)], util[~np.isnan(util)], params[~np.isnan(util)], exp_m[~np.isnan(util)], temp[~np.isnan(util)], co2[~np.isnan(util)], dam[~np.isnan(util)]\n",
    "\n",
    "path = os.getcwd()\n",
    "\n",
    "# sampled parameter values\n",
    "params = np.genfromtxt(''.join([path, '/data/LHC_samples_N1000_DIMS5.csv']), delimiter=',')\n",
    "\n",
    "# pull data from diff files \n",
    "# RENAME AFTER NEW RUNS -- I ADDED AN N TO THE FILE NAME TO SIGNAL HOW MANY DRAWS I TOOK :)\n",
    "ssp5_paths_and_util = np.genfromtxt(''.join([path, '/data/lhc_sampling_paths_ssp5_N1000.csv']), delimiter=',')\n",
    "ssp5_exp_m = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_m_ssp5_N1000.csv']), delimiter=',')\n",
    "ssp5_temp = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_T_ssp5_N1000.csv']), delimiter=',')\n",
    "ssp5_co2 = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_conc_ssp5_N1000.csv']), delimiter=',')\n",
    "ssp5_dam = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_dam_ssp5_N1000.csv']), delimiter=',')\n",
    "ssp5_cpaths, ssp5_c2020, ssp5_c2030, ssp5_c2060, ssp5_c2100, ssp5_c2150, ssp5_c2200, ssp5_util, ssp5_params, ssp5_exp_m, ssp5_exp_T, ssp5_exp_conc, ssp5_exp_dam = remove_nans(ssp5_paths_and_util, ssp5_exp_m, ssp5_temp, ssp5_co2, ssp5_dam, params)\n",
    "ssp5_ra, ssp5_eis, ssp5_ex, ssp5_end, ssp5_prtp = ssp5_params.T\n",
    "\n",
    "ssp4_paths_and_util = np.genfromtxt(''.join([path, '/data/lhc_sampling_paths_ssp4_N1000.csv']), delimiter=',')\n",
    "ssp4_exp_m = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_m_ssp4_N1000.csv']), delimiter=',')\n",
    "ssp4_temp = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_T_ssp4_N1000.csv']), delimiter=',')\n",
    "ssp4_co2 = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_conc_ssp4_N1000.csv']), delimiter=',')\n",
    "ssp4_dam = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_dam_ssp4_N1000.csv']), delimiter=',')\n",
    "ssp4_cpaths, ssp4_c2020, ssp4_c2030, ssp4_c2060, ssp4_c2100, ssp4_c2150, ssp4_c2200, ssp4_util, ssp4_params, ssp4_exp_m, ssp4_exp_T, ssp4_exp_conc, ssp4_exp_dam = remove_nans(ssp4_paths_and_util, ssp4_exp_m, ssp4_temp, ssp4_co2, ssp4_dam, params)\n",
    "ssp4_ra, ssp4_eis, ssp4_ex, ssp4_end, ssp4_prtp = ssp4_params.T\n",
    "\n",
    "ssp3_paths_and_util = np.genfromtxt(''.join([path, '/data/lhc_sampling_paths_ssp3_N1000.csv']), delimiter=',')\n",
    "ssp3_exp_m = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_m_ssp3_N1000.csv']), delimiter=',')\n",
    "ssp3_temp = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_T_ssp3_N1000.csv']), delimiter=',')\n",
    "ssp3_co2 = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_conc_ssp3_N1000.csv']), delimiter=',')\n",
    "ssp3_dam = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_dam_ssp3_N1000.csv']), delimiter=',')\n",
    "ssp3_cpaths, ssp3_c2020, ssp3_c2030, ssp3_c2060, ssp3_c2100, ssp3_c2150, ssp3_c2200, ssp3_util, ssp3_params, ssp3_exp_m, ssp3_exp_T, ssp3_exp_conc, ssp3_exp_dam = remove_nans(ssp3_paths_and_util, ssp3_exp_m, ssp3_temp, ssp3_co2, ssp3_dam, params)\n",
    "ssp3_ra, ssp3_eis, ssp3_ex, ssp3_end, ssp3_prtp = ssp3_params.T\n",
    "\n",
    "ssp2_paths_and_util = np.genfromtxt(''.join([path, '/data/lhc_sampling_paths_ssp2_N1000.csv']), delimiter=',')\n",
    "ssp2_exp_m = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_m_ssp2_N1000.csv']), delimiter=',')\n",
    "ssp2_temp = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_T_ssp2_N1000.csv']), delimiter=',')\n",
    "ssp2_co2 = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_conc_ssp2_N1000.csv']), delimiter=',')\n",
    "ssp2_dam = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_dam_ssp2_N1000.csv']), delimiter=',')\n",
    "ssp2_cpaths, ssp2_c2020, ssp2_c2030, ssp2_c2060, ssp2_c2100, ssp2_c2150, ssp2_c2200, ssp2_util, ssp2_params, ssp2_exp_m, ssp2_exp_T, ssp2_exp_conc, ssp2_exp_dam = remove_nans(ssp2_paths_and_util, ssp2_exp_m, ssp2_temp, ssp2_co2, ssp2_dam, params)\n",
    "ssp2_ra, ssp2_eis, ssp2_ex, ssp2_end, ssp2_prtp = ssp2_params.T\n",
    "\n",
    "ssp1_paths_and_util = np.genfromtxt(''.join([path, '/data/lhc_sampling_paths_ssp1_N1000.csv']), delimiter=',')\n",
    "ssp1_exp_m = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_m_ssp1_N1000.csv']), delimiter=',')\n",
    "ssp1_temp = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_T_ssp1_N1000.csv']), delimiter=',')\n",
    "ssp1_co2 = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_conc_ssp1_N1000.csv']), delimiter=',')\n",
    "ssp1_dam = np.genfromtxt(''.join([path, '/data/lhc_sampling_exp_dam_ssp1_N1000.csv']), delimiter=',')\n",
    "ssp1_cpaths, ssp1_c2020, ssp1_c2030, ssp1_c2060, ssp1_c2100, ssp1_c2150, ssp1_c2200, ssp1_util, ssp1_params, ssp1_exp_m, ssp1_exp_T, ssp1_exp_conc, ssp1_exp_dam = remove_nans(ssp1_paths_and_util, ssp1_exp_m, ssp1_temp, ssp1_co2, ssp1_dam, params)\n",
    "ssp1_ra, ssp1_eis, ssp1_ex, ssp1_end, ssp1_prtp = ssp1_params.T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "suited-merchandise",
   "metadata": {},
   "source": [
    "### Check plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "lonely-tuesday",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = plt.subplots(5)\n",
    "\n",
    "for i in range(len(ssp5_util)):\n",
    "    ax[0].plot(node_times, ssp5_cpaths[i])\n",
    "    \n",
    "for i in range(len(ssp4_util)):\n",
    "    ax[1].plot(node_times, ssp4_cpaths[i])\n",
    "    \n",
    "for i in range(len(ssp3_util)):\n",
    "    ax[2].plot(node_times, ssp3_cpaths[i])\n",
    "    \n",
    "for i in range(len(ssp2_util)):\n",
    "    ax[3].plot(node_times, ssp2_cpaths[i])\n",
    "    \n",
    "for i in range(len(ssp1_util)):          \n",
    "    ax[4].plot(node_times, ssp1_cpaths[i])\n",
    "    \n",
    "fig.tight_layout()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "f65ec1f6",
   "metadata": {},
   "source": [
    "# Quartile ranges\n",
    "\n",
    "Basic idea is to go row by row (cost, temperature, conc, and damages) and plot things up! "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1ffe32fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_std_in_time(paths):\n",
    "    c2020, c2030, c2060, c2100, c2150, c2200 = paths.T\n",
    "    return np.array([np.std(c2020), np.std(c2030), np.std(c2060), np.std(c2100), np.std(c2150), np.std(c2200)])\n",
    "\n",
    "def make_row(ax, row_inds, ssp5_path, ssp4_path, ssp3_path, ssp2_path, ssp1_path, bw, lw, cinds, return_lines, \n",
    "             dam=False, temp=False, conc=False):\n",
    "    ssp5_cind, ssp4_cind, ssp3_cind, ssp2_cind, ssp1_cind = cinds\n",
    "    ssp5_ind, ssp4_ind, ssp3_ind, ssp2_ind, ssp1_ind, var_ind = row_inds\n",
    "    if conc:\n",
    "        LB = 300\n",
    "        UB = 700\n",
    "    elif temp:\n",
    "        LB = 0.8\n",
    "        UB = 4.1\n",
    "    elif dam:\n",
    "        LB = 0\n",
    "        UB = 10.5\n",
    "    else:\n",
    "        UB = 400\n",
    "        LB = 0\n",
    "    \n",
    "    if dam:\n",
    "        ssp5_2020, ssp5_2030, ssp5_2060, ssp5_2100, ssp5_2150, ssp5_2200 = ssp5_path.T * 100\n",
    "        ssp4_2020, ssp4_2030, ssp4_2060, ssp4_2100, ssp4_2150, ssp4_2200 = ssp4_path.T * 100\n",
    "        ssp3_2020, ssp3_2030, ssp3_2060, ssp3_2100, ssp3_2150, ssp3_2200 = ssp3_path.T * 100\n",
    "        ssp2_2020, ssp2_2030, ssp2_2060, ssp2_2100, ssp2_2150, ssp2_2200 = ssp2_path.T * 100\n",
    "        ssp1_2020, ssp1_2030, ssp1_2060, ssp1_2100, ssp1_2150, ssp1_2200 = ssp1_path.T * 100\n",
    "        ssp5_max = np.max((ssp5_2020, ssp5_2030, ssp5_2060, ssp5_2100, ssp5_2150, ssp5_2200))\n",
    "        ssp4_max = np.max((ssp4_2020, ssp4_2030, ssp4_2060, ssp4_2100, ssp4_2150, ssp4_2200))\n",
    "        ssp3_max = np.max((ssp3_2020, ssp3_2030, ssp3_2060, ssp3_2100, ssp3_2150, ssp3_2200))\n",
    "        ssp2_max = np.max((ssp2_2020, ssp2_2030, ssp2_2060, ssp2_2100, ssp2_2150, ssp2_2200))\n",
    "        ssp1_max = np.max((ssp1_2020, ssp1_2030, ssp1_2060, ssp1_2100, ssp1_2150, ssp1_2200))\n",
    "        #UB = np.max([ssp5_max, ssp4_max, ssp3_max, ssp2_max, ssp1_max]) + 2.5\n",
    "\n",
    "        ssp5_std = compute_std_in_time(ssp5_path) * 100\n",
    "        ssp4_std = compute_std_in_time(ssp4_path) * 100\n",
    "        ssp3_std = compute_std_in_time(ssp3_path) * 100\n",
    "        ssp2_std = compute_std_in_time(ssp2_path) * 100\n",
    "        ssp1_std = compute_std_in_time(ssp1_path) * 100\n",
    "\n",
    "        ssp5_mean = np.median(ssp5_path, axis=0) * 100\n",
    "        ssp4_mean = np.median(ssp4_path, axis=0) * 100\n",
    "        ssp3_mean = np.median(ssp3_path, axis=0) * 100\n",
    "        ssp2_mean = np.median(ssp2_path, axis=0) * 100\n",
    "        ssp1_mean = np.median(ssp1_path, axis=0) * 100\n",
    "        \n",
    "        ssp5_64 = np.percentile(ssp5_path, 64, axis=0) * 100\n",
    "        ssp4_64 = np.percentile(ssp4_path, 64, axis=0) * 100\n",
    "        ssp3_64 = np.percentile(ssp3_path, 64, axis=0) * 100\n",
    "        ssp2_64 = np.percentile(ssp2_path, 64, axis=0) * 100\n",
    "        ssp1_64 = np.percentile(ssp1_path, 64, axis=0) * 100\n",
    "        \n",
    "        ssp5_36 = np.percentile(ssp5_path, 36, axis=0) * 100\n",
    "        ssp4_36 = np.percentile(ssp4_path, 36, axis=0) * 100\n",
    "        ssp3_36 = np.percentile(ssp3_path, 36, axis=0) * 100\n",
    "        ssp2_36 = np.percentile(ssp2_path, 36, axis=0) * 100\n",
    "        ssp1_36 = np.percentile(ssp1_path, 36, axis=0) * 100\n",
    "        \n",
    "        ssp5_5 = np.percentile(ssp5_path, 1, axis=0) * 100\n",
    "        ssp4_5 = np.percentile(ssp4_path, 1, axis=0) * 100\n",
    "        ssp3_5 = np.percentile(ssp3_path, 1, axis=0) * 100\n",
    "        ssp2_5 = np.percentile(ssp2_path, 1, axis=0) * 100\n",
    "        ssp1_5 = np.percentile(ssp1_path, 1, axis=0) * 100\n",
    "        \n",
    "        ssp5_95 = np.percentile(ssp5_path, 99, axis=0) * 100\n",
    "        ssp4_95 = np.percentile(ssp4_path, 99, axis=0) * 100\n",
    "        ssp3_95 = np.percentile(ssp3_path, 99, axis=0) * 100\n",
    "        ssp2_95 = np.percentile(ssp2_path, 99, axis=0) * 100\n",
    "        ssp1_95 = np.percentile(ssp1_path, 99, axis=0) * 100\n",
    "    \n",
    "    else:\n",
    "        ssp5_2020, ssp5_2030, ssp5_2060, ssp5_2100, ssp5_2150, ssp5_2200 = ssp5_path.T\n",
    "        ssp4_2020, ssp4_2030, ssp4_2060, ssp4_2100, ssp4_2150, ssp4_2200 = ssp4_path.T\n",
    "        ssp3_2020, ssp3_2030, ssp3_2060, ssp3_2100, ssp3_2150, ssp3_2200 = ssp3_path.T\n",
    "        ssp2_2020, ssp2_2030, ssp2_2060, ssp2_2100, ssp2_2150, ssp2_2200 = ssp2_path.T\n",
    "        ssp1_2020, ssp1_2030, ssp1_2060, ssp1_2100, ssp1_2150, ssp1_2200 = ssp1_path.T\n",
    "        ssp5_max = np.max((ssp5_2020, ssp5_2030, ssp5_2060, ssp5_2100, ssp5_2150, ssp5_2200))\n",
    "        ssp4_max = np.max((ssp4_2020, ssp4_2030, ssp4_2060, ssp4_2100, ssp4_2150, ssp4_2200))\n",
    "        ssp3_max = np.max((ssp3_2020, ssp3_2030, ssp3_2060, ssp3_2100, ssp3_2150, ssp3_2200))\n",
    "        ssp2_max = np.max((ssp2_2020, ssp2_2030, ssp2_2060, ssp2_2100, ssp2_2150, ssp2_2200))\n",
    "        ssp1_max = np.max((ssp1_2020, ssp1_2030, ssp1_2060, ssp1_2100, ssp1_2150, ssp1_2200))\n",
    "        #UB = np.max([ssp5_max, ssp4_max, ssp3_max, ssp2_max, ssp1_max])\n",
    "\n",
    "        ssp5_std = compute_std_in_time(ssp5_path)\n",
    "        ssp4_std = compute_std_in_time(ssp4_path)\n",
    "        ssp3_std = compute_std_in_time(ssp3_path)\n",
    "        ssp2_std = compute_std_in_time(ssp2_path)\n",
    "        ssp1_std = compute_std_in_time(ssp1_path)\n",
    "\n",
    "        ssp5_mean = np.median(ssp5_path, axis=0)\n",
    "        ssp4_mean = np.median(ssp4_path, axis=0)\n",
    "        ssp3_mean = np.median(ssp3_path, axis=0)\n",
    "        ssp2_mean = np.median(ssp2_path, axis=0)\n",
    "        ssp1_mean = np.median(ssp1_path, axis=0)\n",
    "        \n",
    "        ssp5_64 = np.percentile(ssp5_path, 64, axis=0)\n",
    "        ssp4_64 = np.percentile(ssp4_path, 64, axis=0)\n",
    "        ssp3_64 = np.percentile(ssp3_path, 64, axis=0)\n",
    "        ssp2_64 = np.percentile(ssp2_path, 64, axis=0)\n",
    "        ssp1_64 = np.percentile(ssp1_path, 64, axis=0)\n",
    "        \n",
    "        ssp5_36 = np.percentile(ssp5_path, 36, axis=0)\n",
    "        ssp4_36 = np.percentile(ssp4_path, 36, axis=0)\n",
    "        ssp3_36 = np.percentile(ssp3_path, 36, axis=0)\n",
    "        ssp2_36 = np.percentile(ssp2_path, 36, axis=0)\n",
    "        ssp1_36 = np.percentile(ssp1_path, 36, axis=0)\n",
    "        \n",
    "        ssp5_5 = np.percentile(ssp5_path, 1, axis=0)\n",
    "        ssp4_5 = np.percentile(ssp4_path, 1, axis=0)\n",
    "        ssp3_5 = np.percentile(ssp3_path, 1, axis=0)\n",
    "        ssp2_5 = np.percentile(ssp2_path, 1, axis=0)\n",
    "        ssp1_5 = np.percentile(ssp1_path, 1, axis=0)\n",
    "        \n",
    "        ssp5_95 = np.percentile(ssp5_path, 99, axis=0)\n",
    "        ssp4_95 = np.percentile(ssp4_path, 99, axis=0)\n",
    "        ssp3_95 = np.percentile(ssp3_path, 99, axis=0)\n",
    "        ssp2_95 = np.percentile(ssp2_path, 99, axis=0)\n",
    "        ssp1_95 = np.percentile(ssp1_path, 99, axis=0)\n",
    "        \n",
    "    alpha_36_64 = 0.45\n",
    "    alpha_5_95 = 0.1\n",
    "    \n",
    "    ssp5, = ax[ssp5_ind].plot(node_times, ssp5_mean, color=color_list[ssp5_cind], lw=4, label=\"SSP5\")\n",
    "    ax[ssp5_ind].fill_between(node_times, ssp5_36, ssp5_64, color=color_list[ssp5_cind], alpha=alpha_36_64)\n",
    "    ax[ssp5_ind].fill_between(node_times, ssp5_5, ssp5_95, color=color_list[ssp5_cind], alpha=alpha_5_95)\n",
    "    ax[ssp5_ind].set_xticks([2020, 2100, 2200])\n",
    "    ax[ssp5_ind].set_ylim((LB, UB))\n",
    "\n",
    "    ssp4, = ax[ssp4_ind].plot(node_times, ssp4_mean, color=color_list[ssp4_cind], lw=4, label=\"SSP4\")\n",
    "    ax[ssp4_ind].fill_between(node_times, ssp4_36, ssp4_64, color=color_list[ssp4_cind], alpha=alpha_36_64)\n",
    "    ax[ssp4_ind].fill_between(node_times, ssp4_5, ssp4_95, color=color_list[ssp4_cind], alpha=alpha_5_95)\n",
    "    ax[ssp4_ind].set_xticks([2020, 2100, 2200])\n",
    "    ax[ssp4_ind].set_ylim((LB, UB))\n",
    "\n",
    "    ssp3, = ax[ssp3_ind].plot(node_times, ssp3_mean, color=color_list[ssp3_cind], lw=4, label=\"SSP3\")\n",
    "    ax[ssp3_ind].fill_between(node_times, ssp3_36, ssp3_64, color=color_list[ssp3_cind], alpha=alpha_36_64)\n",
    "    ax[ssp3_ind].fill_between(node_times, ssp3_5, ssp3_95, color=color_list[ssp3_cind], alpha=alpha_5_95)\n",
    "    ax[ssp3_ind].set_xticks([2020, 2100, 2200])\n",
    "    ax[ssp3_ind].set_ylim((LB, UB))\n",
    "\n",
    "    ssp2, = ax[ssp2_ind].plot(node_times, ssp2_mean, color=color_list[ssp2_cind], lw=4, label=\"SSP2\")\n",
    "    ax[ssp2_ind].fill_between(node_times, ssp2_36, ssp2_64, color=color_list[ssp2_cind], alpha=alpha_36_64)\n",
    "    ax[ssp2_ind].fill_between(node_times, ssp2_5, ssp2_95, color=color_list[ssp2_cind], alpha=alpha_5_95)\n",
    "    ax[ssp2_ind].set_xticks([2020, 2100, 2200])\n",
    "    ax[ssp2_ind].set_ylim((LB, UB))\n",
    "\n",
    "    ssp1, = ax[ssp1_ind].plot(node_times, ssp1_mean, color=color_list[ssp1_cind], lw=4, label=\"SSP1\")\n",
    "    ax[ssp1_ind].fill_between(node_times, ssp1_36, ssp1_64, color=color_list[ssp1_cind], alpha=alpha_36_64)\n",
    "    ax[ssp1_ind].fill_between(node_times, ssp1_5, ssp1_95, color=color_list[ssp1_cind], alpha=alpha_5_95)\n",
    "    ax[ssp1_ind].set_xticks([2020, 2100, 2200])\n",
    "    ax[ssp1_ind].set_ylim((LB, UB))\n",
    "    \n",
    "    ax[var_ind].plot(node_times, ssp5_std, color=color_list[ssp5_cind])\n",
    "    ax[var_ind].plot(node_times, ssp4_std, color=color_list[ssp4_cind])\n",
    "    ax[var_ind].plot(node_times, ssp3_std, color=color_list[ssp3_cind])\n",
    "    ax[var_ind].plot(node_times, ssp2_std, color=color_list[ssp2_cind])\n",
    "    ax[var_ind].plot(node_times, ssp1_std, color=color_list[ssp1_cind])\n",
    "    if temp:\n",
    "        ax[var_ind].set_ylim((0, 0.6))\n",
    "    elif conc:\n",
    "        ax[var_ind].set_ylim((0, 80))\n",
    "    elif dam:\n",
    "        ax[var_ind].set_ylim((0, 2))\n",
    "    else:\n",
    "        ax[var_ind].set_ylim((0, 80))\n",
    "    if return_lines:\n",
    "        return ssp5, ssp4, ssp3, ssp2, ssp1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e29ba66a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.transforms as mtransforms\n",
    "\n",
    "bw = 4.5\n",
    "lw = 1.5\n",
    "\n",
    "ssp5_cind = 0\n",
    "ssp4_cind = 6\n",
    "ssp3_cind = 7\n",
    "ssp2_cind = 2\n",
    "ssp1_cind = 3\n",
    "cinds = [ssp5_cind, ssp4_cind, ssp3_cind, ssp2_cind, ssp1_cind]\n",
    "\n",
    "cost_row = ['A', 'B', 'C', 'D', 'E', 'F']\n",
    "t_row = ['G', 'H', 'I', 'J', 'K', 'L']\n",
    "conc_row = ['M', 'N', 'O', 'P', 'Q', 'R']\n",
    "dam_row = ['S', 'T', 'U', 'V', 'W', 'X']\n",
    "\n",
    "fig, ax = plt.subplot_mosaic([cost_row, t_row, conc_row, dam_row], sharex=True,\n",
    "                             gridspec_kw={'height_ratios': [1, 1, 1, 1], 'width_ratios': [1, 1, 1, 1, 1, 1]},\n",
    "                            figsize=(25,19))\n",
    "\n",
    "ssp5, ssp4, ssp3, ssp2, ssp1 = make_row(ax, cost_row, ssp5_cpaths, ssp4_cpaths, ssp3_cpaths, ssp2_cpaths, ssp1_cpaths, \n",
    "         bw, lw, cinds, return_lines=True)\n",
    "\n",
    "make_row(ax, t_row, ssp5_exp_T, ssp4_exp_T, ssp3_exp_T, ssp2_exp_T, ssp1_exp_T, \n",
    "         bw, lw, cinds, return_lines=True, dam=False, temp=True)\n",
    "\n",
    "make_row(ax, conc_row, ssp5_exp_conc, ssp4_exp_conc, ssp3_exp_conc, ssp2_exp_conc, ssp1_exp_conc, \n",
    "         bw, lw, cinds, return_lines=True, dam=False, temp=False, conc=True)\n",
    "\n",
    "make_row(ax, dam_row, ssp5_exp_dam, ssp4_exp_dam, ssp3_exp_dam, ssp2_exp_dam, ssp1_exp_dam, \n",
    "         bw, lw, cinds, return_lines=True, dam=True)\n",
    "\n",
    "ax['A'].set_ylabel(\"Cost (2020 $USD)\")\n",
    "ax['G'].set_ylabel(\"Temperature ($^{\\circ}$C)\")\n",
    "ax['M'].set_ylabel(\"CO$_2$ concentration (ppm)\")\n",
    "ax['S'].set_ylabel(\"Damage (% GDP)\")\n",
    "\n",
    "# turn off y labels\n",
    "turn_off_ylabels = ['B', 'C', 'D', 'E', 'H', 'I', 'J', 'K', 'N', 'O', 'P', 'Q', 'T', 'U', 'V', 'W']\n",
    "for i in turn_off_ylabels:\n",
    "    ax[i].set_yticklabels([])\n",
    "    \n",
    "# make x axis label\n",
    "for i in dam_row:\n",
    "    ax[i].set_xlabel(\"Year\")\n",
    "    ax[i].set_xticklabels(['2020', '2100', '2200'])\n",
    "    \n",
    "    \n",
    "for i in dam_row[:-1]:\n",
    "    ax[i].set_yticks([0, 5, 10])\n",
    "    \n",
    "for i in t_row[:-1]:\n",
    "    ax[i].set_yticks([1,2,3,4])\n",
    "    \n",
    "for i in conc_row[:-1]:\n",
    "    ax[i].set_yticks([300, 400, 500, 600, 700])\n",
    "    \n",
    "for i in cost_row[:-1]:\n",
    "    ax[i].set_yticks([0, 100, 200, 300, 400])\n",
    "\n",
    "# label panels\n",
    "for label, ax in ax.items():\n",
    "    # label physical distance in and down:\n",
    "    trans = mtransforms.ScaledTranslation(2.2, 0.0, fig.dpi_scale_trans)\n",
    "    ax.text(0.0, 1.0, label, transform=ax.transAxes + trans, fontsize=22, fontweight='bold',\n",
    "            verticalalignment='top', bbox=dict(facecolor='none', edgecolor='none', pad=1))\n",
    "    \n",
    "fig.legend([ssp5, ssp4, ssp3, ssp2, ssp1], [\"SSP5$-$8.5\", \"SSP4$-$6.0\", \"SSP3$-$7.0\", \"SSP2$-$4.5\", \"SSP1$-$1.9\"], bbox_to_anchor=(0.5, 0.03),\n",
    "      loc='lower center', ncol=5, fancybox=True, shadow=True, fontsize=22)\n",
    "\n",
    "sns.despine(offset=5, trim=True)\n",
    "\n",
    "if save_figs:\n",
    "    fig.savefig(''.join([basefile, \"quartile-ensemble.png\"]), dpi=400, bbox_inches=\"tight\")    "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}