downscaling-baselines.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add submodule paths\n",
    "import sys\n",
    "sys.path += ['./normalizing_flows', './baselines', './climdex']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow_probability as tfp\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import utils.data as data_util\n",
    "import utils.nn_util as nn\n",
    "import xarray as xr\n",
    "import gcsfs\n",
    "import dscnn\n",
    "import climdex.temperature as tdex\n",
    "import logging\n",
    "from baselines.bcsd import BCSD\n",
    "from normalizing_flows.models import VariationalModel, FlowLVM, JointFlowLVM, adversarial\n",
    "from normalizing_flows.models.variational import nll_loss\n",
    "from normalizing_flows.models.optimization import LinearWarmupSchedule\n",
    "from normalizing_flows.flows import Transform, Flow, Invert\n",
    "from normalizing_flows.flows.image import Upsample\n",
    "from normalizing_flows.flows.glow import GlowFlow, coupling_nn_glow\n",
    "from datasource import EraiRasDataLoader\n",
    "from utils.preprocessing import remove_monthly_means\n",
    "from utils.distributions import normal, bernoulli_gamma\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tqdm import tqdm\n",
    "from regions import southeast_us, pacific_nw\n",
    "\n",
    "gcs = gcsfs.GCSFileSystem(project='thesis-research-255223', token='gcs.secret.json')\n",
    "logging.basicConfig(stream=sys.stdout, level=logging.INFO)\n",
    "#tf.autograph.set_verbosity(1)\n",
    "#tf.config.experimental_run_functions_eagerly(True)\n",
    "#tf.debugging.set_log_device_placement(True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = EraiRasDataLoader(gcs_bucket='erai-rasmussen', gcs_project='thesis-research-255223', auth='gcs.secret.json')\n",
    "# era-interim\n",
    "erai_deg1 = xr.open_zarr(data.erai('daily-1deg'), consolidated=True)\n",
    "# 1-degree regridded rasmussen\n",
    "ras_deg1 = xr.open_zarr(data.rasmussen('daily-1deg'), consolidated=True)\n",
    "# 1/2-degree regridded rasmussen\n",
    "ras_deg12 = xr.open_zarr(data.rasmussen('daily-1-2deg'), consolidated=True)\n",
    "ras_deg14 = xr.open_zarr(data.rasmussen('daily-1-4deg'), consolidated=True)\n",
    "ras_deg18 = xr.open_zarr(data.rasmussen('daily-1-8deg'), consolidated=True)\n",
    "#ras_deg116 = xr.open_zarr(rasmussen('daily-1-16deg'), consolidated=True)\n",
    "erai_deg1_seus = southeast_us(erai_deg1)\n",
    "ras_deg1_seus = southeast_us(ras_deg1)\n",
    "ras_deg12_seus = southeast_us(ras_deg12, scale_factor=2)\n",
    "ras_deg14_seus = southeast_us(ras_deg14, scale_factor=4)\n",
    "ras_deg18_seus = southeast_us(ras_deg18, scale_factor=8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "split_fn = data_util.create_time_series_train_test_generator(n_splits=3)\n",
    "def preprocess_maxt(ds):\n",
    "    return ds[['MAXT']].fillna(0.).clip(min=0.0, max=np.inf)\n",
    "ras_seus_maxt_14 = preprocess_maxt(ras_deg14_seus)\n",
    "ras_seus_maxt_12 = preprocess_maxt(ras_deg12_seus)\n",
    "ras_seus_maxt_1 = preprocess_maxt(ras_deg1_seus)\n",
    "erai_seus_maxt_1 = preprocess_maxt(erai_deg1_seus)\n",
    "folds = list(split_fn(erai_seus_maxt_1, ras_seus_maxt_14))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_vds(data_lo, data_hi, batch_size=100, buffer_size=1000, supervised=True):\n",
    "    if supervised:\n",
    "        data = tf.data.Dataset.zip((data_lo, data_hi)).shuffle(buffer_size)\n",
    "    else:\n",
    "        data = tf.data.Dataset.zip((data_lo.shuffle(buffer_size), data_hi.shuffle(buffer_size)))\n",
    "    return data.batch(batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_1xn(data):\n",
    "    n = len(data)\n",
    "    plt.figure(figsize=(n*9,6))\n",
    "    for i in range(n):\n",
    "        plt.subplot(1,n,i+1)\n",
    "        plt.imshow(data[i].numpy().squeeze(), origin='lower')\n",
    "        plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = tdex.indices('Time')\n",
    "def eval_climdex(true, pred, coords):\n",
    "    true_arr = xr.DataArray(true, coords=coords)\n",
    "    pred_arr = xr.DataArray(pred, coords=coords)\n",
    "    txx_true = indices.monthly_txx(true_arr)\n",
    "    txx_pred = indices.monthly_txx(pred_arr)\n",
    "    txn_true = indices.monthly_txn(true_arr)\n",
    "    txn_pred = indices.monthly_txn(pred_arr)\n",
    "    txx_bias = txx_pred - txx_true\n",
    "    txn_bias = txn_pred - txn_true\n",
    "    return txx_bias, txn_bias"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bias-Correction Spatial Disaggregation (BCSD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:root:initializing tensors for low resolution datasets\n",
      "INFO:root:fitting quantile map\n",
      "INFO:root:computing scaling factors\n",
      "INFO:root:applying bias-correction quantile mapping\n",
      "INFO:root:interpolating low resolution data to target resolution\n"
     ]
    }
   ],
   "source": [
    "bcsd = BCSD(verbose=True)\n",
    "#(lr_train, lr_obs_train, hr_train), (lr_test, lr_obs_test, hr_test) = folds[-1]\n",
    "N = erai_seus_maxt_1.Time.size\n",
    "lr_train = erai_seus_maxt_1.isel(Time=slice(0,N-365)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "lr_test = erai_seus_maxt_1.isel(Time=slice(N-365,N+1)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "lr_obs_train = ras_seus_maxt_1.isel(Time=slice(0,N-365)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "lr_obs_test = ras_seus_maxt_1.isel(Time=slice(N-365,N+1)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "hr_train = ras_seus_maxt_14.isel(Time=slice(0,N-365)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "hr_test = ras_seus_maxt_14.isel(Time=slice(N-365,N+1)).to_array(dim='chan').transpose('Time','lat','lon','chan')\n",
    "bcsd.fit(lr_train, hr_train, lr_obs_train)\n",
    "hr_pred = bcsd.predict(lr_test).values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "plt.subplot(1,2,1)\n",
    "plt.imshow(hr_test[0].squeeze())\n",
    "plt.colorbar()\n",
    "plt.title('hr true')\n",
    "plt.subplot(1,2,2)\n",
    "plt.imshow(hr_pred[0].squeeze())\n",
    "plt.colorbar()\n",
    "plt.title('hr predicted, BCSD')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5.6931621088551765e-06 19.984285152946086\n",
      "BCSD test rmse: 2.3055165569030724 +/- 3.2489679599287693\n"
     ]
    }
   ],
   "source": [
    "_, means = remove_monthly_means(hr_train)\n",
    "hr_test, _ = remove_monthly_means(hr_test, means)\n",
    "hr_pred, _ = remove_monthly_means(xr.DataArray(hr_pred, coords=hr_test.coords, dims=hr_test.dims), means)\n",
    "err = (hr_test - hr_pred)**2\n",
    "print(np.sqrt(np.min(err)).values, np.sqrt(np.max(err)).values)\n",
    "print(f'BCSD test rmse: {float(np.sqrt(np.mean(err)))} +/- {float(np.sqrt(np.std(err)))}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "16.422486732148172"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.max(hr_pred.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = np.random.randn(10,10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.11479822629833375 0.11479822629833374\n"
     ]
    }
   ],
   "source": [
    "print(x.mean(), x.mean(axis=0).mean(axis=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Baseline: Vanilla Bano-Medina inference network for downscaling\n",
    "\n",
    "$f: X \\rightarrow \\Theta$, $\\hat{y} \\sim p(\\Theta)$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fold 1/3\n",
      "1200 training samples, 1200 test samples, 8x16 -> 32x64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "train, epoch 1/50: 100%|██████████| 120/120 [00:18<00:00,  6.40it/s, loss=4.19, nll=4.19]\n",
      "test, epoch 1/50: 100%|██████████| 120/120 [00:07<00:00, 15.56it/s, nll=3.56]\n",
      "train, epoch 2/50: 100%|██████████| 120/120 [00:04<00:00, 27.93it/s, loss=3.31, nll=3.31]\n",
      "test, epoch 2/50: 100%|██████████| 120/120 [00:06<00:00, 17.28it/s, nll=3.42]\n",
      "train, epoch 3/50: 100%|██████████| 120/120 [00:04<00:00, 28.72it/s, loss=3.14, nll=3.14]\n",
      "test, epoch 3/50: 100%|██████████| 120/120 [00:07<00:00, 16.98it/s, nll=3.34]\n",
      "train, epoch 4/50: 100%|██████████| 120/120 [00:04<00:00, 27.85it/s, loss=3.07, nll=3.07]\n",
      "test, epoch 4/50: 100%|██████████| 120/120 [00:07<00:00, 15.81it/s, nll=3.29]\n",
      "train, epoch 5/50: 100%|██████████| 120/120 [00:04<00:00, 25.77it/s, loss=3.04, nll=3.04]\n",
      "test, epoch 5/50: 100%|██████████| 120/120 [00:07<00:00, 16.60it/s, nll=3.25]\n",
      "train, epoch 6/50: 100%|██████████| 120/120 [00:04<00:00, 27.52it/s, loss=3.01, nll=3.01]\n",
      "test, epoch 6/50: 100%|██████████| 120/120 [00:07<00:00, 15.77it/s, nll=3.23]\n",
      "train, epoch 7/50: 100%|██████████| 120/120 [00:04<00:00, 28.37it/s, loss=2.99, nll=2.99]\n",
      "test, epoch 7/50: 100%|██████████| 120/120 [00:07<00:00, 16.82it/s, nll=3.2] \n",
      "train, epoch 8/50: 100%|██████████| 120/120 [00:04<00:00, 26.05it/s, loss=2.98, nll=2.98]\n",
      "test, epoch 8/50: 100%|██████████| 120/120 [00:07<00:00, 15.55it/s, nll=3.19]\n",
      "train, epoch 9/50: 100%|██████████| 120/120 [00:04<00:00, 26.30it/s, loss=2.97, nll=2.97]\n",
      "test, epoch 9/50: 100%|██████████| 120/120 [00:07<00:00, 15.70it/s, nll=3.18]\n",
      "train, epoch 10/50: 100%|██████████| 120/120 [00:04<00:00, 26.81it/s, loss=2.96, nll=2.96]\n",
      "test, epoch 10/50: 100%|██████████| 120/120 [00:07<00:00, 16.40it/s, nll=3.16]\n",
      "train, epoch 11/50: 100%|██████████| 120/120 [00:04<00:00, 25.85it/s, loss=2.95, nll=2.95]\n",
      "test, epoch 11/50: 100%|██████████| 120/120 [00:07<00:00, 16.46it/s, nll=3.16]\n",
      "train, epoch 12/50: 100%|██████████| 120/120 [00:04<00:00, 27.20it/s, loss=2.95, nll=2.95]\n",
      "test, epoch 12/50: 100%|██████████| 120/120 [00:06<00:00, 17.44it/s, nll=3.15]\n",
      "train, epoch 13/50: 100%|██████████| 120/120 [00:04<00:00, 28.60it/s, loss=2.94, nll=2.94]\n",
      "test, epoch 13/50: 100%|██████████| 120/120 [00:07<00:00, 16.74it/s, nll=3.14]\n",
      "train, epoch 14/50: 100%|██████████| 120/120 [00:04<00:00, 26.72it/s, loss=2.93, nll=2.93]\n",
      "test, epoch 14/50: 100%|██████████| 120/120 [00:07<00:00, 16.73it/s, nll=3.13]\n",
      "train, epoch 15/50: 100%|██████████| 120/120 [00:04<00:00, 26.63it/s, loss=2.93, nll=2.93]\n",
      "test, epoch 15/50: 100%|██████████| 120/120 [00:07<00:00, 16.74it/s, nll=3.13]\n",
      "train, epoch 16/50: 100%|██████████| 120/120 [00:04<00:00, 27.41it/s, loss=2.92, nll=2.92]\n",
      "test, epoch 16/50: 100%|██████████| 120/120 [00:07<00:00, 16.00it/s, nll=3.13]\n",
      "train, epoch 17/50: 100%|██████████| 120/120 [00:03<00:00, 30.04it/s, loss=2.91, nll=2.91]\n",
      "test, epoch 17/50: 100%|██████████| 120/120 [00:07<00:00, 17.05it/s, nll=3.12]\n",
      "train, epoch 18/50: 100%|██████████| 120/120 [00:04<00:00, 27.48it/s, loss=2.91, nll=2.91]\n",
      "test, epoch 18/50: 100%|██████████| 120/120 [00:07<00:00, 16.89it/s, nll=3.12]\n",
      "train, epoch 19/50: 100%|██████████| 120/120 [00:04<00:00, 27.26it/s, loss=2.91, nll=2.91]\n",
      "test, epoch 19/50: 100%|██████████| 120/120 [00:06<00:00, 17.25it/s, nll=3.12]\n",
      "train, epoch 20/50: 100%|██████████| 120/120 [00:04<00:00, 27.97it/s, loss=2.9, nll=2.9]  \n",
      "test, epoch 20/50: 100%|██████████| 120/120 [00:07<00:00, 16.68it/s, nll=3.11]\n",
      "train, epoch 21/50: 100%|██████████| 120/120 [00:04<00:00, 27.97it/s, loss=2.89, nll=2.89]\n",
      "test, epoch 21/50: 100%|██████████| 120/120 [00:07<00:00, 16.71it/s, nll=3.11]\n",
      "train, epoch 22/50: 100%|██████████| 120/120 [00:04<00:00, 26.91it/s, loss=2.89, nll=2.89]\n",
      "test, epoch 22/50: 100%|██████████| 120/120 [00:07<00:00, 16.14it/s, nll=3.11]\n",
      "train, epoch 23/50: 100%|██████████| 120/120 [00:04<00:00, 28.13it/s, loss=2.89, nll=2.89]\n",
      "test, epoch 23/50: 100%|██████████| 120/120 [00:07<00:00, 16.81it/s, nll=3.11]\n",
      "train, epoch 24/50: 100%|██████████| 120/120 [00:04<00:00, 25.88it/s, loss=2.88, nll=2.88]\n",
      "test, epoch 24/50: 100%|██████████| 120/120 [00:08<00:00, 14.44it/s, nll=3.1]\n",
      "train, epoch 25/50: 100%|██████████| 120/120 [00:04<00:00, 24.05it/s, loss=2.88, nll=2.88]\n",
      "test, epoch 25/50: 100%|██████████| 120/120 [00:07<00:00, 16.32it/s, nll=3.1]\n",
      "train, epoch 26/50: 100%|██████████| 120/120 [00:04<00:00, 26.92it/s, loss=2.87, nll=2.87]\n",
      "test, epoch 26/50: 100%|██████████| 120/120 [00:07<00:00, 16.80it/s, nll=3.1]\n",
      "train, epoch 27/50: 100%|██████████| 120/120 [00:04<00:00, 27.30it/s, loss=2.87, nll=2.87]\n",
      "test, epoch 27/50: 100%|██████████| 120/120 [00:07<00:00, 16.52it/s, nll=3.1]\n",
      "train, epoch 28/50: 100%|██████████| 120/120 [00:04<00:00, 26.64it/s, loss=2.87, nll=2.87]\n",
      "test, epoch 28/50: 100%|██████████| 120/120 [00:07<00:00, 16.94it/s, nll=3.1]\n",
      "train, epoch 29/50: 100%|██████████| 120/120 [00:04<00:00, 28.75it/s, loss=2.86, nll=2.86]\n",
      "test, epoch 29/50: 100%|██████████| 120/120 [00:07<00:00, 16.12it/s, nll=3.1]\n",
      "train, epoch 30/50: 100%|██████████| 120/120 [00:04<00:00, 26.28it/s, loss=2.86, nll=2.86]\n",
      "test, epoch 30/50: 100%|██████████| 120/120 [00:06<00:00, 17.41it/s, nll=3.1]\n",
      "train, epoch 31/50: 100%|██████████| 120/120 [00:04<00:00, 28.21it/s, loss=2.85, nll=2.85]\n",
      "test, epoch 31/50: 100%|██████████| 120/120 [00:07<00:00, 16.82it/s, nll=3.1] \n",
      "train, epoch 32/50: 100%|██████████| 120/120 [00:04<00:00, 27.11it/s, loss=2.85, nll=2.85]\n",
      "test, epoch 32/50: 100%|██████████| 120/120 [00:07<00:00, 16.62it/s, nll=3.1]\n",
      "train, epoch 33/50: 100%|██████████| 120/120 [00:04<00:00, 25.52it/s, loss=2.85, nll=2.85]\n",
      "test, epoch 33/50: 100%|██████████| 120/120 [00:06<00:00, 17.92it/s, nll=3.1]\n",
      "train, epoch 34/50: 100%|██████████| 120/120 [00:04<00:00, 28.89it/s, loss=2.84, nll=2.84]\n",
      "test, epoch 34/50: 100%|██████████| 120/120 [00:07<00:00, 16.39it/s, nll=3.1]\n",
      "train, epoch 35/50: 100%|██████████| 120/120 [00:04<00:00, 27.71it/s, loss=2.84, nll=2.84]\n",
      "test, epoch 35/50: 100%|██████████| 120/120 [00:07<00:00, 15.87it/s, nll=3.1]\n",
      "train, epoch 36/50: 100%|██████████| 120/120 [00:04<00:00, 27.08it/s, loss=2.84, nll=2.84]\n",
      "test, epoch 36/50: 100%|██████████| 120/120 [00:07<00:00, 16.45it/s, nll=3.1]\n",
      "train, epoch 37/50: 100%|██████████| 120/120 [00:04<00:00, 27.06it/s, loss=2.83, nll=2.83]\n",
      "test, epoch 37/50: 100%|██████████| 120/120 [00:07<00:00, 16.61it/s, nll=3.1]\n",
      "train, epoch 38/50: 100%|██████████| 120/120 [00:04<00:00, 27.57it/s, loss=2.83, nll=2.83]\n",
      "test, epoch 38/50: 100%|██████████| 120/120 [00:06<00:00, 17.38it/s, nll=3.1]\n",
      "train, epoch 39/50: 100%|██████████| 120/120 [00:04<00:00, 28.65it/s, loss=2.82, nll=2.82]\n",
      "test, epoch 39/50: 100%|██████████| 120/120 [00:07<00:00, 16.43it/s, nll=3.1]\n",
      "train, epoch 40/50: 100%|██████████| 120/120 [00:04<00:00, 26.59it/s, loss=2.82, nll=2.82]\n",
      "test, epoch 40/50: 100%|██████████| 120/120 [00:07<00:00, 15.82it/s, nll=3.1]\n",
      "train, epoch 41/50: 100%|██████████| 120/120 [00:04<00:00, 26.94it/s, loss=2.82, nll=2.82]\n",
      "test, epoch 41/50: 100%|██████████| 120/120 [00:07<00:00, 16.65it/s, nll=3.1]\n",
      "train, epoch 42/50: 100%|██████████| 120/120 [00:04<00:00, 29.35it/s, loss=2.81, nll=2.81]\n",
      "test, epoch 42/50: 100%|██████████| 120/120 [00:07<00:00, 16.02it/s, nll=3.1]\n",
      "train, epoch 43/50: 100%|██████████| 120/120 [00:04<00:00, 26.43it/s, loss=2.81, nll=2.81]\n",
      "test, epoch 43/50: 100%|██████████| 120/120 [00:07<00:00, 16.35it/s, nll=3.1]\n",
      "train, epoch 44/50: 100%|██████████| 120/120 [00:04<00:00, 26.69it/s, loss=2.81, nll=2.81]\n",
      "test, epoch 44/50: 100%|██████████| 120/120 [00:07<00:00, 15.85it/s, nll=3.1]\n",
      "train, epoch 45/50: 100%|██████████| 120/120 [00:04<00:00, 26.37it/s, loss=2.8, nll=2.8]  \n",
      "test, epoch 45/50: 100%|██████████| 120/120 [00:07<00:00, 16.48it/s, nll=3.1]\n",
      "train, epoch 46/50: 100%|██████████| 120/120 [00:04<00:00, 27.98it/s, loss=2.8, nll=2.8] \n",
      "test, epoch 46/50: 100%|██████████| 120/120 [00:07<00:00, 16.57it/s, nll=3.1]\n",
      "train, epoch 47/50: 100%|██████████| 120/120 [00:04<00:00, 28.42it/s, loss=2.8, nll=2.8]  \n",
      "test, epoch 47/50: 100%|██████████| 120/120 [00:07<00:00, 16.16it/s, nll=3.1]\n",
      "train, epoch 48/50: 100%|██████████| 120/120 [00:04<00:00, 26.06it/s, loss=2.79, nll=2.79]\n",
      "test, epoch 48/50: 100%|██████████| 120/120 [00:07<00:00, 16.82it/s, nll=3.1]\n",
      "train, epoch 49/50: 100%|██████████| 120/120 [00:04<00:00, 26.74it/s, loss=2.79, nll=2.79]\n",
      "test, epoch 49/50: 100%|██████████| 120/120 [00:07<00:00, 16.25it/s, nll=3.1]\n",
      "train, epoch 50/50: 100%|██████████| 120/120 [00:04<00:00, 27.53it/s, loss=2.78, nll=2.78]\n",
      "test, epoch 50/50: 100%|██████████| 120/120 [00:07<00:00, 15.83it/s, nll=3.1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE: 1.882798671722412 bias: 0.2587953805923462\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 3240x432 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fold 2/3\n",
      "2400 training samples, 2400 test samples, 8x16 -> 32x64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "train, epoch 1/50:   8%|▊         | 20/240 [00:14<02:44,  1.34it/s, loss=7.12, nll=7.12]\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-8-07b9dec1061d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     29\u001b[0m     \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mVariationalModel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mencoder\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnormal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptimizer\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mAdam\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlr\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1.0E-4\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutput_shape\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mht_hi\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mwt_hi\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     30\u001b[0m     model.fit(train_ds.repeat(n_epochs), epochs=n_epochs, steps_per_epoch=train_steps,\n\u001b[0;32m---> 31\u001b[0;31m              validation_data=test_ds.repeat(n_epochs), validation_steps=test_steps)\n\u001b[0m\u001b[1;32m     32\u001b[0m     \u001b[0mx_true\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     33\u001b[0m     \u001b[0my_true\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/generative-downscaling/normalizing_flows/normalizing_flows/models/variational.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, data, epochs, steps_per_epoch, validation_data, validation_steps, **flow_kwargs)\u001b[0m\n\u001b[1;32m    102\u001b[0m             \u001b[0mtrain_hist\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    103\u001b[0m             \u001b[0;32mwith\u001b[0m \u001b[0mtqdm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtotal\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdesc\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34mf'train, epoch {epoch+1}/{epochs}'\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mprog\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 104\u001b[0;31m                 \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32min\u001b[0m \u001b[0menumerate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtake\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    105\u001b[0m                     \u001b[0mloss\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnll\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_batch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mflow_kwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    106\u001b[0m                     \u001b[0mutils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate_metrics\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_hist\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mloss\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnll\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mnll\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/srv/conda/envs/notebook/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m__next__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    620\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    621\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m__next__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# For Python 3 compatibility\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 622\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    623\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    624\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_next_internal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/srv/conda/envs/notebook/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36mnext\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    664\u001b[0m     \u001b[0;34m\"\"\"Returns a nested structure of `Tensor`s containing the next element.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    665\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 666\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_next_internal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    667\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOutOfRangeError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    668\u001b[0m       \u001b[0;32mraise\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/srv/conda/envs/notebook/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m_next_internal\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    649\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_iterator_resource\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    650\u001b[0m             \u001b[0moutput_types\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flat_output_types\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 651\u001b[0;31m             output_shapes=self._flat_output_shapes)\n\u001b[0m\u001b[1;32m    652\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    653\u001b[0m       \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/srv/conda/envs/notebook/lib/python3.7/site-packages/tensorflow_core/python/ops/gen_dataset_ops.py\u001b[0m in \u001b[0;36miterator_get_next_sync\u001b[0;34m(iterator, output_types, output_shapes, name)\u001b[0m\n\u001b[1;32m   2657\u001b[0m         \u001b[0m_ctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_context_handle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_ctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_thread_local_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdevice_name\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2658\u001b[0m         \u001b[0;34m\"IteratorGetNextSync\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_ctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_post_execution_callbacks\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0miterator\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2659\u001b[0;31m         \"output_types\", output_types, \"output_shapes\", output_shapes)\n\u001b[0m\u001b[1;32m   2660\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0m_result\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2661\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0m_core\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_FallbackException\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "sample_batch_size = 10\n",
    "load_batch_size = 1200\n",
    "n_epochs = 50\n",
    "for i, ((train_lo, train_hi), (test_lo, test_hi)) in enumerate(folds):\n",
    "    print(f'Fold {i+1}/{len(folds)}')\n",
    "    N_train, N_test = train_lo.Time.size, train_lo.Time.size\n",
    "    (ht, wt), (ht_hi, wt_hi) = train_lo.shape[1:3], train_hi.shape[1:3]\n",
    "    print('{} training samples, {} test samples, {}x{} -> {}x{}'.format(N_train, N_test, ht, wt, ht_hi, wt_hi))\n",
    "    #batch_multiplier = data_util.calculate_n_subimages(train_lo, k, stride)\n",
    "    train_steps = data_util.num_batches(N_train, sample_batch_size)\n",
    "    test_steps = data_util.num_batches(N_test, sample_batch_size)\n",
    "    train_lo, monthly_means_lo = remove_monthly_means(train_lo)\n",
    "    train_hi, monthly_means_hi = remove_monthly_means(train_hi)\n",
    "    test_lo,_ = remove_monthly_means(test_lo, monthly_means_lo)\n",
    "    test_hi,_ = remove_monthly_means(test_hi, monthly_means_hi)\n",
    "    train_lo_ds = data_util.xr_to_tf_dataset(train_lo, load_batch_size)\n",
    "    test_lo_ds = data_util.xr_to_tf_dataset(test_lo, load_batch_size)\n",
    "    train_hi_ds = data_util.xr_to_tf_dataset(train_hi, load_batch_size)\n",
    "    test_hi_ds = data_util.xr_to_tf_dataset(test_hi, load_batch_size)\n",
    "    train_ds = preprocess_vds(train_lo_ds, train_hi_ds, batch_size=sample_batch_size)\n",
    "    test_ds = preprocess_vds(test_lo_ds, test_hi_ds, batch_size=sample_batch_size)\n",
    "    #vdsrcnn = dscnn.create_vdsrcnn(scale=wt_hi // wt, c_in=1, c_out=2, module_layers=2, n_layers=5, output_init='zeros')\n",
    "    from tensorflow.keras import backend as K\n",
    "    def rmse(y_true, y_pred):\n",
    "        return K.sqrt(K.mean(K.square(y_true - y_pred)))\n",
    "    encoder = dscnn.create_bmd_cnn10(ht, wt, scale=wt_hi // wt, c_out=2)\n",
    "    #encoder.compile(loss='mse', optimizer=Adam(lr=1.0E-4), metrics=[rmse])\n",
    "    #encoder.fit(train_ds.repeat(n_epochs), epochs=n_epochs, steps_per_epoch=train_steps, validation_data=test_ds.repeat(n_epochs), validation_steps=test_steps)\n",
    "    model = VariationalModel(encoder, normal(), optimizer=Adam(lr=1.0E-4), output_shape=(None,ht_hi,wt_hi,1))\n",
    "    model.fit(train_ds.repeat(n_epochs), epochs=n_epochs, steps_per_epoch=train_steps,\n",
    "             validation_data=test_ds.repeat(n_epochs), validation_steps=test_steps)\n",
    "    x_true = []\n",
    "    y_true = []\n",
    "    y_mean = []\n",
    "    y_median = []\n",
    "    y_samples = []\n",
    "    for x, y in test_ds:\n",
    "        x_true.append(x)\n",
    "        y_true.append(y)\n",
    "        y_mean.append(model.mean(x))\n",
    "        y_median.append(model.quantile(x, 0.5))\n",
    "        y_samples.append(model.sample(x, sample_fn=lambda d: d.loc + d.scale*tf.random.normal((1,))))\n",
    "    x_true = tf.concat(x_true, axis=0)\n",
    "    y_true = tf.concat(y_true, axis=0)\n",
    "    y_mean = tf.concat(y_mean, axis=0)\n",
    "    y_median = tf.concat(y_median, axis=0)\n",
    "    y_samples = tf.concat(y_samples, axis=0)\n",
    "    rmse = tf.math.reduce_mean(tf.math.sqrt(tf.math.reduce_mean((y_mean - y_true)**2, axis=0)))\n",
    "    bias = tf.math.reduce_mean(tf.math.reduce_mean(y_mean - y_true, axis=0))\n",
    "    print(f'RMSE: {rmse.numpy()}', f'bias: {bias.numpy()}')\n",
    "    plt.figure(figsize=(5*9,6))\n",
    "    plt.subplot(1,5,1)\n",
    "    plt.imshow(x_true[0].numpy().squeeze(), origin='lower')\n",
    "    plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.subplot(1,5,2)\n",
    "    plt.imshow(y_true[0].numpy().squeeze(), origin='lower')\n",
    "    plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.subplot(1,5,3)\n",
    "    plt.imshow(y_mean[0].numpy().squeeze(), origin='lower')\n",
    "    plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.subplot(1,5,4)\n",
    "    plt.imshow(y_median[0].numpy().squeeze(), origin='lower')\n",
    "    plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.subplot(1,5,5)\n",
    "    plt.imshow(y_samples[0].numpy().squeeze(), origin='lower')\n",
    "    plt.colorbar(pad=0.04, shrink=0.5)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def download_bmd(run_id, out_dir, epoch=50):\n",
    "    from tensorflow.keras.models import load_model\n",
    "    import os\n",
    "    import shutil\n",
    "    os.makedirs(out_dir, exist_ok=True)\n",
    "    filename = f'bmd-epoch{epoch}.h5'\n",
    "    with gcs.open(f'gs://generative-downscaling-artifact-store/bmd-final/{run_id}/artifacts/model/{filename}') as src:\n",
    "        with open(f'{out_dir}/{filename}', 'wb') as dst:\n",
    "            shutil.copyfileobj(src, dst)\n",
    "    return load_model(f'{out_dir}/{filename}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_train_test_splits(data_lo, data_hi, region_fn, scale, cv=False):\n",
    "    data_lo = region_fn(data_lo)\n",
    "    data_hi = region_fn(data_hi, scale_factor=scale)\n",
    "    if cv:\n",
    "        split_fn = create_time_series_train_test_generator_v2(n_splits=5, test_size=146)\n",
    "        folds = list(split_fn(data_lo, data_hi))\n",
    "        return folds\n",
    "    else:\n",
    "        lr_train = data_lo.isel(Time=slice(0,data_lo.Time.size-2*365))\n",
    "        lr_test = data_lo.isel(Time=slice(data_lo.Time.size-2*365, data_lo.Time.size+1))\n",
    "        hr_train = data_hi.isel(Time=slice(0,data_lo.Time.size-2*365))\n",
    "        hr_test = data_hi.isel(Time=slice(data_lo.Time.size-2*365, data_lo.Time.size+1))\n",
    "        return lr_train, lr_test, hr_train, hr_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import experiments.prcp_experiment_base as prcp\n",
    "from experiments.common import upsample\n",
    "erai_deg1_arr = prcp.preprocess_dataset(erai_deg1[['PRCP']], erai_deg1['PRCP'].shape)\n",
    "erai_deg1_arr = erai_deg1_arr.to_array('chan').transpose('Time','lat','lon','chan')\n",
    "erai_deg1_arr = erai_deg1_arr.where(erai_deg1_arr > 1.0, erai_deg1_arr, 0.0)\n",
    "ras_deg14_arr = prcp.preprocess_dataset(ras_deg14[['PRCP']], ras_deg14['PRCP'].shape)\n",
    "ras_deg14_arr = ras_deg14_arr.to_array('chan').transpose('Time','lat','lon','chan')\n",
    "ras_deg14_arr = ras_deg14_arr.where(ras_deg14_arr > 1.0, ras_deg14_arr, 0.0)\n",
    "lr_train, lr_test, hr_train, hr_test = get_train_test_splits(erai_deg1_arr, ras_deg14_arr, southeast_us, scale=4)\n",
    "data_fold = prcp.preprocess_fold_prcp(((lr_train, hr_train),(lr_test, hr_test)))\n",
    "train_lo_prcp, train_hi_prcp = data_fold.train\n",
    "test_lo_prcp, test_hi_prcp = data_fold.test\n",
    "N_train, N_test = train_lo_prcp.Time.size, test_lo_prcp.Time.size\n",
    "(wt, ht), (wt_hi, ht_hi) = train_lo_prcp.shape[1:3], train_hi_prcp.shape[1:3]\n",
    "test_ds_prcp = data_fold.test_dataset(batch_size=100, buffer_size=1000,\n",
    "                                      mode='test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:No training configuration found in save file: the model was *not* compiled. Compile it manually.\n",
      "WARNING:tensorflow:No training configuration found in save file: the model was *not* compiled. Compile it manually.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/srv/conda/envs/notebook/lib/python3.7/site-packages/tensorflow_core/python/training/tracking/data_structures.py:669: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated since Python 3.3,and in 3.9 it will stop working\n",
      "  if not isinstance(wrapped_dict, collections.Mapping):\n"
     ]
    }
   ],
   "source": [
    "encoder = download_bmd('b22951acf857496eb4d63899479c9797', '/tmp/bmd-cnn')\n",
    "model = VariationalModel(encoder, bernoulli_gamma(), output_shape=(None,ht_hi,wt_hi,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/8 [00:00<?, ?it/s]/srv/conda/envs/notebook/lib/python3.7/site-packages/dask/core.py:119: RuntimeWarning: invalid value encountered in greater\n",
      "  return func(*args2)\n",
      "100%|██████████| 8/8 [00:05<00:00,  1.49it/s]\n"
     ]
    }
   ],
   "source": [
    "# Store true values and predictions in memory\n",
    "x_true = []\n",
    "y_true = []\n",
    "y_pred = []\n",
    "y_sample = []\n",
    "for x, y in tqdm(test_ds_prcp, total=hr_test.Time.size//100+1):\n",
    "    y_ = model.mean(x)\n",
    "    y_s = model.sample(x)\n",
    "    y_pred.append(y_)\n",
    "    y_sample.append(y_s)\n",
    "    x_true.append(x)\n",
    "    y_true.append(y)\n",
    "x_true = tf.concat(x_true, axis=0)\n",
    "y_true = tf.concat(y_true, axis=0)\n",
    "y_pred = tf.concat(y_pred, axis=0)\n",
    "y_sample = tf.concat(y_sample, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_true = x_true**3\n",
    "y_true = y_true**3\n",
    "y_pred = y_pred**3\n",
    "y_sample = y_sample**3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x504 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from utils.plot import image_map_factory, prcp_cmap\n",
    "fig, axs, plot_fn = image_map_factory(2, 2, figsize=(6,3.5))\n",
    "plot_fn(axs[0,0], x_true[229].numpy().squeeze(), lr_test.lat, lr_test.lon, title='ERA-I', cmap=prcp_cmap())\n",
    "plot_fn(axs[0,1], y_true[229].numpy().squeeze(), hr_test.lat, hr_test.lon, title='WRF-4', cmap=prcp_cmap())\n",
    "plot_fn(axs[1,0], y_pred[229].numpy().squeeze(), hr_test.lat, hr_test.lon, title='BMD-CNN predicted', cmap=prcp_cmap())\n",
    "cs = plot_fn(axs[1,1], y_sample[229].numpy().squeeze(), hr_test.lat, hr_test.lon, title='BMD-CNN sample', cmap=prcp_cmap())\n",
    "fig.colorbar(cs, ax=axs.ravel().tolist(), orientation='vertical', shrink=0.75, pad=0.02).set_label('daily precipitation (mm)')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}