inference_demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "### import lib and helper functions\n",
    "\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from einops import rearrange, repeat, reduce\n",
    "import os\n",
    "import yaml, shutil\n",
    "\n",
    "import wandb\n",
    "from torch.utils.data import DataLoader\n",
    "from weather_transformer import CaFAEPD\n",
    "from dynamics_training_loop import configure_epd_models, configure_train_dataset_and_loader, configure_residual_stat, epd_predict\n",
    "from utils.loss_utils import latitude_weighted_rmse\n",
    "from utils.training_utils import dict2namespace\n",
    "import gc\n",
    "import xarray as xr\n",
    "from datetime import datetime,timedelta                                                                                                                                 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "### set the device， it is recommended to use GPU\n",
    "device = 'cuda:0'\n",
    "\n",
    "### load the config which was used to train the model\n",
    "with open('configs/era5-EPD-6432-stage2.yml', 'r') as f:\n",
    "    config = yaml.safe_load(f)\n",
    "config = dict2namespace(config)\n",
    "model = configure_epd_models(config)    # load model\n",
    "\n",
    "\n",
    "### load checkpoint, change the ckpt path correspondingly\n",
    "ckpt_path = 'experiments/pretrained_ckpt/ema_last_6432.pth'\n",
    "model.load_state_dict(torch.load(ckpt_path)['model'])\n",
    "model.to(device)\n",
    "model.eval()\n",
    "print('loaded successfully')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 37 years of data, ranging from 1979 to 2015\n",
      "There is a total of 13514 days\n",
      "Filter out years not in the range: [1990, 2015]\n",
      "Original # of years: 37, Filtered # of years: 26\n"
     ]
    }
   ],
   "source": [
    "### prepare dataset's normalization statistics\n",
    "train_dataset, _ = configure_train_dataset_and_loader(2, 1, config)    # the train dataset is just used to get the normalization statistics\n",
    "normalizer = train_dataset.datareader.normalizer\n",
    "\n",
    "# get the constant feature\n",
    "constant_names = config.data.constant_names\n",
    "constants = torch.from_numpy(train_dataset.datareader.constants).float().unsqueeze(0)\n",
    "\n",
    "### load residual normalizer, which is used to de-normalize output of the model\n",
    "residual_normalizer = configure_residual_stat(config)\n",
    "# to device\n",
    "residual_normalizer = residual_normalizer.to(device) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "### fetch data point from xarray dataset of WeatherBench 2\n",
    "def fetch_and_process(dat_path,\n",
    "                      init_time_coord,     # datetime of the initial condition\n",
    "                      ):\n",
    "    ds = xr.open_zarr(dat_path)\n",
    "    # a slice of time coords\n",
    "    time_coord = slice(init_time_coord, init_time_coord + timedelta(hours=6)*28)   # hardcode the forecast length to a week\n",
    "    ds = ds.sel(time=time_coord)   # select the initial condition\n",
    "\n",
    "    variable_groups = config.data.variable_groups\n",
    "    # flatten the variable groups\n",
    "    feature_names = [item for sublist in variable_groups for item in sublist]\n",
    "\n",
    "    # get the data by variable name\n",
    "    input_feat_dict = {}\n",
    "    out_feat_dict = {}\n",
    "    for var_name in feature_names:\n",
    "        # transpose the latitude and longitude\n",
    "        feat = ds[var_name].values\n",
    "        if len(feat.shape) == 4:  # t, levels, nlon, nlat -> t, nlat, nlon, levels\n",
    "            feat = np.transpose(feat, (0, 3, 2, 1))\n",
    "        elif len(feat.shape) == 3:  # t, nlon, nlat -> t, nlat, nlon\n",
    "            feat = np.transpose(feat, (0, 2, 1))\n",
    "\n",
    "        input_feat_dict[var_name] = torch.from_numpy(feat[0]).float().unsqueeze(0)   # only use the first time step as input\n",
    "        out_feat_dict[var_name] = torch.from_numpy(feat[1:]).float().unsqueeze(0) \n",
    "\n",
    "    return input_feat_dict, out_feat_dict\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "### try prediction on 2018-01-01 00:00:00\n",
    "init_time = datetime(2018, 1, 1, 0, 0, 0)\n",
    "### specify the path to the zarr file\n",
    "base_path = # your path\n",
    "dat_path = base_path + '1959-2022-6h-64x32_equiangular_with_poles_conservative.zarr'\n",
    "### get the input and output features\n",
    "input_dict, target_dict = fetch_and_process(dat_path, init_time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([1, 28, 32, 64])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_dict['2m_temperature'].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "### use epd_predict to predict the output\n",
    "with torch.no_grad():\n",
    "    torch.cuda.empty_cache()\n",
    "    # model = torch.compile(model, reduce_overhead=True)\n",
    "    loss_dict, pred_dict = epd_predict(model, [1, 4, 12, 20, 28], input_dict, target_dict, constants,\n",
    "                                    normalizer,\n",
    "                                    residual_normalizer,\n",
    "                                    latitude_weighted_rmse,\n",
    "                                    config=config,\n",
    "                                    device=device,\n",
    "                                    return_pred=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0.3263, 0.5097, 0.8088, 1.2526, 1.8833]], device='cuda:0',\n",
       "       dtype=torch.float64)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### checkout the loss at different lead time: 6hr 1day 3day 5day 7day\n",
    "loss_dict['2m_temperature']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### visualize the prediction and target\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "fig, axs = plt.subplots(1, 2, figsize=(10, 5))\n",
    "# prediction\n",
    "axs[0].imshow(pred_dict['2m_temperature'][0, 0].cpu().numpy(), cmap='RdBu_r')\n",
    "axs[0].set_title('Prediction')\n",
    "# target\n",
    "axs[1].imshow(target_dict['2m_temperature'][0, 0].cpu().numpy(), cmap='RdBu_r')\n",
    "axs[1].set_title('Target')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(1, 2, figsize=(10, 5))\n",
    "# prediction\n",
    "axs[0].imshow(pred_dict['2m_temperature'][0, -1].cpu().numpy(), cmap='RdBu_r')\n",
    "axs[0].set_title('Prediction')\n",
    "# target\n",
    "axs[1].imshow(target_dict['2m_temperature'][0, -1].cpu().numpy(), cmap='RdBu_r')\n",
    "axs[1].set_title('Target')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "cafa",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}