Python_Demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "toc": true
   },
   "source": [
    "<h1>Table of Contents<span class=\"tocSkip\"></span></h1>\n",
    "<div class=\"toc\"><ul class=\"toc-item\"><li><span><a href=\"#Installation\" data-toc-modified-id=\"Installation-1\"><span class=\"toc-item-num\">1&nbsp;&nbsp;</span>Installation</a></span></li><li><span><a href=\"#Remote-access\" data-toc-modified-id=\"Remote-access-2\"><span class=\"toc-item-num\">2&nbsp;&nbsp;</span>Remote access</a></span></li><li><span><a href=\"#The-basics\" data-toc-modified-id=\"The-basics-3\"><span class=\"toc-item-num\">3&nbsp;&nbsp;</span>The basics</a></span><ul class=\"toc-item\"><li><span><a href=\"#Open/close-with-netCDF4.Dataset\" data-toc-modified-id=\"Open/close-with-netCDF4.Dataset-3.1\"><span class=\"toc-item-num\">3.1&nbsp;&nbsp;</span>Open/close with <code>netCDF4.Dataset</code></a></span></li><li><span><a href=\"#Add-dimensions,-variables,-attributes\" data-toc-modified-id=\"Add-dimensions,-variables,-attributes-3.2\"><span class=\"toc-item-num\">3.2&nbsp;&nbsp;</span>Add dimensions, variables, attributes</a></span><ul class=\"toc-item\"><li><span><a href=\"#netCDF4.Dataset.createDimension\" data-toc-modified-id=\"netCDF4.Dataset.createDimension-3.2.1\"><span class=\"toc-item-num\">3.2.1&nbsp;&nbsp;</span><code>netCDF4.Dataset.createDimension</code></a></span></li><li><span><a href=\"#netCDF4.Dataset.createVariable\" data-toc-modified-id=\"netCDF4.Dataset.createVariable-3.2.2\"><span class=\"toc-item-num\">3.2.2&nbsp;&nbsp;</span><code>netCDF4.Dataset.createVariable</code></a></span></li></ul></li></ul></li><li><span><a href=\"#Write-a-netCDF-from-a-set-of-GeoTIFFs\" data-toc-modified-id=\"Write-a-netCDF-from-a-set-of-GeoTIFFs-4\"><span class=\"toc-item-num\">4&nbsp;&nbsp;</span>Write a netCDF from a set of GeoTIFFs</a></span><ul class=\"toc-item\"><li><span><a href=\"#Requirements\" data-toc-modified-id=\"Requirements-4.1\"><span class=\"toc-item-num\">4.1&nbsp;&nbsp;</span>Requirements</a></span></li><li><span><a href=\"#Coordinates\" data-toc-modified-id=\"Coordinates-4.2\"><span class=\"toc-item-num\">4.2&nbsp;&nbsp;</span>Coordinates</a></span><ul class=\"toc-item\"><li><span><a href=\"#Time\" data-toc-modified-id=\"Time-4.2.1\"><span class=\"toc-item-num\">4.2.1&nbsp;&nbsp;</span>Time</a></span></li><li><span><a href=\"#Spatial-characteristics\" data-toc-modified-id=\"Spatial-characteristics-4.2.2\"><span class=\"toc-item-num\">4.2.2&nbsp;&nbsp;</span>Spatial characteristics</a></span></li></ul></li><li><span><a href=\"#Open-a-netCDF-for-writing\" data-toc-modified-id=\"Open-a-netCDF-for-writing-4.3\"><span class=\"toc-item-num\">4.3&nbsp;&nbsp;</span>Open a netCDF for writing</a></span></li><li><span><a href=\"#Add-dimensions\" data-toc-modified-id=\"Add-dimensions-4.4\"><span class=\"toc-item-num\">4.4&nbsp;&nbsp;</span>Add dimensions</a></span></li><li><span><a href=\"#Add-coordinate-variables\" data-toc-modified-id=\"Add-coordinate-variables-4.5\"><span class=\"toc-item-num\">4.5&nbsp;&nbsp;</span>Add coordinate variables</a></span><ul class=\"toc-item\"><li><span><a href=\"#time\" data-toc-modified-id=\"time-4.5.1\"><span class=\"toc-item-num\">4.5.1&nbsp;&nbsp;</span>time</a></span></li><li><span><a href=\"#time_bnds\" data-toc-modified-id=\"time_bnds-4.5.2\"><span class=\"toc-item-num\">4.5.2&nbsp;&nbsp;</span>time_bnds</a></span></li><li><span><a href=\"#x-and-y\" data-toc-modified-id=\"x-and-y-4.5.3\"><span class=\"toc-item-num\">4.5.3&nbsp;&nbsp;</span>x and y</a></span></li><li><span><a href=\"#lon-and-lat\" data-toc-modified-id=\"lon-and-lat-4.5.4\"><span class=\"toc-item-num\">4.5.4&nbsp;&nbsp;</span>lon and lat</a></span></li></ul></li><li><span><a href=\"#Add-crs-variable\" data-toc-modified-id=\"Add-crs-variable-4.6\"><span class=\"toc-item-num\">4.6&nbsp;&nbsp;</span>Add crs variable</a></span></li><li><span><a href=\"#Add-data-variable\" data-toc-modified-id=\"Add-data-variable-4.7\"><span class=\"toc-item-num\">4.7&nbsp;&nbsp;</span>Add data variable</a></span></li><li><span><a href=\"#Add-global-attributes\" data-toc-modified-id=\"Add-global-attributes-4.8\"><span class=\"toc-item-num\">4.8&nbsp;&nbsp;</span>Add global attributes</a></span></li><li><span><a href=\"#Close-the-file-and-check\" data-toc-modified-id=\"Close-the-file-and-check-4.9\"><span class=\"toc-item-num\">4.9&nbsp;&nbsp;</span>Close the file and check</a></span></li></ul></li><li><span><a href=\"#Write-the-netCDF-again-in-a-script!\" data-toc-modified-id=\"Write-the-netCDF-again-in-a-script!-5\"><span class=\"toc-item-num\">5&nbsp;&nbsp;</span>Write the netCDF again in a script!</a></span></li></ul></div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# netCDF in Python\n",
    "\n",
    "## Installation\n",
    "\n",
    "The Python interface to the netCDF4 library is called [`netCDF4`](https://unidata.github.io/netcdf4-python/netCDF4/index.html) (maintained by Unidata). Visit the documentation at the link to read more.\n",
    "\n",
    "Install with pip:\n",
    "```shell\n",
    "pip install netCDF4\n",
    "```\n",
    "\n",
    "Install with conda:\n",
    "```shell\n",
    "conda install netCDF4\n",
    "```\n",
    "\n",
    "## Remote access\n",
    "\n",
    "Remember those awesome features of netCDF that Michele mentioned? They're all available to you through Python!\n",
    "\n",
    "This is the published version of the file from our example data that is equivalent to the one that Michele just made in her demo. It's accessible through THREDDS, and **we can read the data directly from the web via OPeNDAP**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:17.503535Z",
     "start_time": "2019-09-09T15:41:17.091087Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Dataset'>\n",
       "root group (NETCDF3_CLASSIC data model, file format DAP2):\n",
       "    _NCProperties: version=1|netcdflibversion=4.4.1|hdf5libversion=1.8.17\n",
       "    Conventions: CF-1.6\n",
       "    institution: University of Washington\n",
       "    references: Mahoney, P., Liston, G., Mangipane, B., Prugh, L.\n",
       "    project: Arctic-Boreal Vulnerability Experiment (ABoVE)\n",
       "    contact: pmahoney29@gmail.com\n",
       "    title: Navigating Snowscapes: Scale-Dependent Responses of Mountain Sheep to Snowpack Properties\n",
       "    DODS.strlen: 0\n",
       "    dimensions(sizes): maxStrlen64(64), nv(2), time(365), x(160), y(240)\n",
       "    variables(dimensions): |S1 \u001b[4malbers_conical_equal_area\u001b[0m(maxStrlen64), float64 \u001b[4mtime\u001b[0m(time), int32 \u001b[4mtime_bnds\u001b[0m(time,nv), float64 \u001b[4mx\u001b[0m(x), int32 \u001b[4my\u001b[0m(y), float32 \u001b[4mSnowDepth_500m\u001b[0m(time,y,x), float64 \u001b[4mlat\u001b[0m(y,x), float64 \u001b[4mlon\u001b[0m(y,x)\n",
       "    groups: "
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import netCDF4 as nc4\n",
    "\n",
    "# Open remote url to the file published on THREDDS.\n",
    "publishedds = nc4.Dataset( \n",
    "    'https://thredds.daac.ornl.gov/thredds/dodsC/'\n",
    "    'ornldaac/1602/snow_depth_00500m_2006.nc4' ) \n",
    "\n",
    "publishedds # Display the header."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that the staff member that was responsible for this dataset made a few different choices in the structuring of the final copy of the netCDF. This drives home the point that netCDF is very flexible and powerful. So be a responsible user of netCDF and remember to format your files in a well-documented, consistent manner!\n",
    "\n",
    "**Pull the data for the first timestep:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:17.614518Z",
     "start_time": "2019-09-09T15:41:17.507454Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "masked_array(\n",
       "  data=[[ 8.66,  8.65,  8.71, ..., 77.86, 79.  , 79.89],\n",
       "        [ 8.51,  8.51,  8.55, ..., 81.88, 78.94, 81.5 ],\n",
       "        [23.18,  8.41,  8.42, ..., 84.29, 92.59, 84.35],\n",
       "        ...,\n",
       "        [14.78, 14.94, 15.27, ..., 74.52, 79.63, 74.94],\n",
       "        [15.3 , 13.95,  4.61, ..., 69.66, 65.08, 60.02],\n",
       "        [ 4.14,  4.93,  5.84, ..., 67.21, 66.26, 89.39]],\n",
       "  mask=False,\n",
       "  fill_value=1e+20,\n",
       "  dtype=float32)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "publishedarr = publishedds.variables[\"SnowDepth_500m\"][0]\n",
    "\n",
    "publishedarr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**and plot with `matplotlib`:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.657227Z",
     "start_time": "2019-09-09T15:41:17.617661Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x242666b5688>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fig = plt.figure(figsize=(5, 10))\n",
    "plt.imshow(publishedarr, cmap=\"jet\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Close the dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.672696Z",
     "start_time": "2019-09-09T15:41:18.662260Z"
    }
   },
   "outputs": [],
   "source": [
    "publishedds.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The basics\n",
    "\n",
    "### Open/close with `netCDF4.Dataset`\n",
    "\n",
    "Calling the `Dataset` constructor with a file path and a keyword argument `mode=<'r', 'w' ['r+', 'a']>` opens a netCDF for reading or writing. Here I've omitted the path argument (giving `None` instead) because I don't plan to write this example to disk:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.697288Z",
     "start_time": "2019-09-09T15:41:18.679083Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Dataset'>\n",
       "root group (NETCDF4 data model, file format HDF5):\n",
       "    dimensions(sizes): \n",
       "    variables(dimensions): \n",
       "    groups: "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nc = nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\")  \n",
    "\n",
    "nc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The empty `netCDF4.Dataset` has no dimensions, variables, or groups.\n",
    "\n",
    "You should always remember to close any open datasets when you are finished working:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.709227Z",
     "start_time": "2019-09-09T15:41:18.699991Z"
    }
   },
   "outputs": [],
   "source": [
    "nc.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And the dataset is no longer accessible:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.726003Z",
     "start_time": "2019-09-09T15:41:18.712511Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Yep, this threw an error!\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    print(nc)\n",
    "except:\n",
    "    print(\"Yep, this threw an error!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Context is managed using the `with` statement, so you can do all of your work inside `with` and the file will close automatically:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.742366Z",
     "start_time": "2019-09-09T15:41:18.732127Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'netCDF4._netCDF4.Dataset'>\n",
      "root group (NETCDF4 data model, file format HDF5):\n",
      "    dimensions(sizes): \n",
      "    variables(dimensions): \n",
      "    groups: \n",
      "\n"
     ]
    }
   ],
   "source": [
    "with nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\") as nc:\n",
    "    print(nc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And the dataset is no longer accessible, so the following call to `netCDF.Dataset.close()` will throw an error:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.757722Z",
     "start_time": "2019-09-09T15:41:18.748619Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Yep, this threw another error!\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    nc.close()\n",
    "except:\n",
    "    print(\"Yep, this threw another error!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The rest of these examples will take place inside the `with` statement so we don't have to worry about opening/closing manually.\n",
    "\n",
    "### Add dimensions, variables, attributes\n",
    "\n",
    "Dimensions and variables are created by calling the [`createDimension`](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset.createDimension) and [`createVariable`](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset.createVariable) methods of the [`netCDF4.Dataset`](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset) constructor. \n",
    "\n",
    "Say we have this simple 4x4 raster:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.777104Z",
     "start_time": "2019-09-09T15:41:18.765377Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0,  1,  2,  3],\n",
       "       [ 4,  5,  6,  7],\n",
       "       [ 8,  9, 10, 11],\n",
       "       [12, 13, 14, 15]])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "np.arange(16).reshape(4,4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The netCDF that stores this raster must have (at least) two dimensions of size=4. \n",
    "\n",
    "#### `netCDF4.Dataset.createDimension`\n",
    "\n",
    "Here is how you would open the file and add the two dimensions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.804666Z",
     "start_time": "2019-09-09T15:41:18.783865Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('y', <class 'netCDF4._netCDF4.Dimension'>: name = 'y', size = 4),\n",
       "             ('x',\n",
       "              <class 'netCDF4._netCDF4.Dimension'>: name = 'x', size = 4)])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nc = nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\") # Open a netCDF for writing.\n",
    "\n",
    "nc.createDimension(\"y\", 4)                # Add a y dimension, size 4.\n",
    "nc.createDimension(\"x\", 4)                # Add an x dimension, size 4.\n",
    "    \n",
    "nc.dimensions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note the two dimensions named `x` and `y` in resulting file. The [`netCDF.Dataset.createDimension`](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset.createDimension) method takes two arguments:\n",
    "* dimension name\n",
    "* dimension size\n",
    "\n",
    "*Note: If you do not specify a size for the dimension, netCDF will assume that the dimension is UNLIMITED when it writes the file.* \n",
    "\n",
    "#### `netCDF4.Dataset.createVariable`\n",
    "\n",
    "The netCDF created in the last step can accomodate the 4x4 array, so make a new variable inside the dataset using [`netCDF.Dataset.createVariable`](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset.createVariable) and add the raster to this new variable.\n",
    "\n",
    "Here we add `MyVariable`, which stores 16-bit integers on y and x:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.825031Z",
     "start_time": "2019-09-09T15:41:18.807547Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "int16 MyVariable(y, x)\n",
       "unlimited dimensions: \n",
       "current shape = (4, 4)\n",
       "filling on, default _FillValue of -32767 used"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var = nc.createVariable(\"MyVariable\", \"i2\", (\"y\", \"x\")) \n",
    "var"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, the dataset now how a variable called *MyVariable* (argument 1: `\"MyVariable\"`) that stores integers (argument 2: `\"i2\"`) mapped to the dimensions *x* and *y* (argument 3: `(\"y\", \"x\")`). The `createVariable` method has several other optional arguments that provide important functionality like assigning the fill value and setting compression level, among other things, so please check out the [documentation](https://unidata.github.io/netcdf4-python/netCDF4/index.html#netCDF4.Dataset.createVariable) for more info.\n",
    "\n",
    "But, we didn't add the data last time! Add the array from before: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.845485Z",
     "start_time": "2019-09-09T15:41:18.830447Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0  1  2  3]\n",
      " [ 4  5  6  7]\n",
      " [ 8  9 10 11]\n",
      " [12 13 14 15]]\n"
     ]
    }
   ],
   "source": [
    "var[:,:] = np.arange(16).reshape(4,4) # Add array to 'MyVariable'.\n",
    "print(var[:,:])                       # Print 'MyVariable's array."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The indices of the Python variable `var` determine the destination of the input data.\n",
    "\n",
    "I created the variable and assigned its reference to the Python variable `var`. Then I set the values of `var` using the two colons separated by a comma, which indicates the incoming data values span the full length of each dimension of the variable. \n",
    "\n",
    "I can modify the data at a specific position using the same indexing method:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.861468Z",
     "start_time": "2019-09-09T15:41:18.849708Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[    0     1 -9999     3]\n",
      " [    4     5 -9999     7]\n",
      " [    8     9 -9999    11]\n",
      " [   12    13 -9999    15]]\n"
     ]
    }
   ],
   "source": [
    "var[:,2] = -9999                      # Set third col's values to -9999.\n",
    "print(var[:,:])                       # Print the array again."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the previous cell we set all of the data values at position 2 along the `x` dimension to -9999. We must have identified some error associated with the data at these positions and decided to assign a fill value instead of storing bad data. \n",
    "\n",
    "netCDF allows for a fill value to be specified for each variable. Setting the fill value provides a few conveniences such as automated masking and reduced storage volume. Fill values are specified when you call `createVariable`. So, we need to close the netCDF dataset we were just working with, and open a new one with `MyVariable`'s fill_value argument set to `-9999`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.888669Z",
     "start_time": "2019-09-09T15:41:18.868089Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0 1 -- 3]\n",
      " [4 5 -- 7]\n",
      " [8 9 -- 11]\n",
      " [12 13 -- 15]]\n"
     ]
    }
   ],
   "source": [
    "nc.close() # Close the dataset.\n",
    "\n",
    "# Open a new netCDF dataset and apply the routine above.\n",
    "nc = nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\")\n",
    "nc.createDimension(\"y\", 4)\n",
    "nc.createDimension(\"x\", 4)\n",
    "\n",
    "# But this time, pass -9999 to createVariable's fill_value argument.\n",
    "var = nc.createVariable(\"MyVariable\", \"i2\", (\"y\", \"x\"), fill_value=-9999)\n",
    "var[:,:] = np.arange(16).reshape(4,4)\n",
    "var[:,2] = -9999\n",
    "\n",
    "print(var[:,:]) # Print the array again."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "See the values for `x` position 2 are automatically masked! Setting attributes for our new variable is as easy as assigning to the variable like any other Python attribute. \n",
    "\n",
    "Check out what I mean in the next cell (**and don't forget to close the file**):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:18.995624Z",
     "start_time": "2019-09-09T15:41:18.892131Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "netcdf MyFirstNetCDF {\n",
      "dimensions:\n",
      "\ty = 4 ;\n",
      "\tx = 4 ;\n",
      "variables:\n",
      "\tshort MyVariable(y, x) ;\n",
      "\t\tMyVariable:_FillValue = -9999s ;\n",
      "\t\tMyVariable:MyAttribute = \"MyVariable! I made this!\" ;\n",
      "data:\n",
      "\n",
      " MyVariable =\n",
      "  0, 1, _, 3,\n",
      "  4, 5, _, 7,\n",
      "  8, 9, _, 11,\n",
      "  12, 13, _, 15 ;\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "var.MyAttribute = \"MyVariable! I made this!\" # Add an attribute.\n",
    "nc.close()                                   # Close the dataset.\n",
    "\n",
    "# Print the entire closed file with ncdump.\n",
    "!ncdump MyFirstNetCDF.nc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now you're ready to convert a time series of GeoTIFFs to netCDF!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Write a netCDF from a set of GeoTIFFs\n",
    "\n",
    "This example uses the same time series of GeoTIFFs from the ABoVE dataset that Michele introduced.\n",
    "\n",
    "### Requirements\n",
    "\n",
    "This workflow uses only open software packages maintained by OGC and Unidata + packages from the Python standard library:\n",
    "\n",
    "* [`pyproj`](https://pyproj4.github.io/pyproj/stable/),\n",
    "\n",
    "Pyproj is the Python interface to the PROJ library for cartographic projection and coordinate transformation. PROJ is maintained by the Open Geospatial Consortium (OGC) and is a go-to for geospatial data power users.\n",
    "\n",
    "* [`gdal`,`osr`](https://gdal.org/)\n",
    "\n",
    "GDAL and OSR are both part of the Python interface to the Geospatial Data Abstraction Library, also maintained by OSGeo.\n",
    "\n",
    "* [`netCDF4`](https://unidata.github.io/netcdf4-python/netCDF4/index.html)\n",
    "* [`numpy`](https://numpy.org/)\n",
    "\n",
    "In case you're new to Python, `numpy` is the fundamental package for scientific computing in Python. It provides the array  structures behind the Python packages interfaces to both GDAL and netCDF4. NumPy is highly optimized. You’ll see that these examples will run much faster than the equivalent workflow using the NCO tools.\n",
    "\n",
    "Importing GDAL's Python interface is a bit unusual. Install using:\n",
    "```shell\n",
    "pip install gdal          # OR\n",
    "conda install gdal        # OR\n",
    "sudo apt-get install python-gdal\n",
    "```\n",
    "BUT *import* using:\n",
    "```python\n",
    "import osgeo.gdal         # OR\n",
    "from osgeo import gdal\n",
    "```\n",
    "\n",
    "Import all requirements and glob for the GeoTIFFs that you saw in Michele's example. Print the first five:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.093706Z",
     "start_time": "2019-09-09T15:41:18.999220Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GeoTIFF COUNT: 365 | FIRST FIVE:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['singleband\\\\SnowModel_DALL_2006_01_01.tif',\n",
       " 'singleband\\\\SnowModel_DALL_2006_01_02.tif',\n",
       " 'singleband\\\\SnowModel_DALL_2006_01_03.tif',\n",
       " 'singleband\\\\SnowModel_DALL_2006_01_04.tif',\n",
       " 'singleband\\\\SnowModel_DALL_2006_01_05.tif']"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pyproj import Proj, transform\n",
    "from osgeo import gdal, osr\n",
    "import netCDF4 as nc4\n",
    "import numpy as np\n",
    "\n",
    "import datetime as dt\n",
    "import glob\n",
    "import os\n",
    "\n",
    "# Get a list of files and print the first five.\n",
    "tifs = glob.glob(\"singleband/*2006*.tif\")\n",
    "print(\"GeoTIFF COUNT: {0} | FIRST FIVE:\".format(len(tifs))); tifs[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We learn a few basic steps before writing a netCDF.\n",
    "\n",
    "### Coordinates\n",
    "\n",
    "#### Time\n",
    "\n",
    "Notice that the filenames contain dates for the time series. Convert the first five filenames to a list of dates (with `datetime` from Python's standard library) in three steps:\n",
    "\n",
    "1. Split the filename by the string `__DALL__`. Yields a list:\n",
    "```python\n",
    ">>> x.split(\"_DALL_\")\n",
    "['data\\\\SnowModel', '2006_01_01.tif']\n",
    "```\n",
    "2. Remove the extension from the second item in the resulting list. Yields another list:\n",
    "```python\n",
    ">>> os.path.splitext(x[1])\n",
    "['2006_01_01', '.tif']\n",
    "```\n",
    "3. Split the first item, the date string, by underscores to get its component *yyyy, mm, dd* as a list:\n",
    "```python\n",
    ">>> x[0].split(\"_\")\n",
    "['2006', '01', '01']\n",
    "```\n",
    "\n",
    "**Try it out:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.109222Z",
     "start_time": "2019-09-09T15:41:19.097254Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2006-01-01\n",
      "2006-01-02\n",
      "2006-01-03\n",
      "2006-01-04\n",
      "2006-01-05\n",
      "\n",
      "Date objects are instances of: <class 'datetime.date'>\n"
     ]
    }
   ],
   "source": [
    "# Loop over the tifs.\n",
    "for t in tifs[:5]:\n",
    "    \n",
    "    # Split by \"_DALL_\" string, remove the extension, split by underscore.\n",
    "    yyyy, mm, dd = os.path.splitext(t.split(\"_DALL_\")[1])[0].split(\"_\")\n",
    "    \n",
    "    # Make a datetime date. Don't forget to convert strings to integers.\n",
    "    date = dt.date(int(yyyy), int(mm), int(dd))\n",
    "    \n",
    "    # Print the date.\n",
    "    print(date)\n",
    "    \n",
    "# Check the last date's type.\n",
    "print(\"\\nDate objects are instances of: %s\" % type(date))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "CF also specifies that we need to provide time values relative to some origin time, typically since epoch (1970-01-01 00:00:00), but we will start from 2006 for consistency with Michele's example. \n",
    "\n",
    "The routine below computes the time since origin using the datetime object for the last date that was printed above (2006-01-05). It also shifts the value by one half day so that it complies with the CF mandate that time values represent the center of the observation period. Other than subtracting the datetimes, these steps are just basic arithmetic."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.120746Z",
     "start_time": "2019-09-09T15:41:19.113223Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time:\t\t4.5\n",
      "time_bnds:\t[4, 5]\n"
     ]
    }
   ],
   "source": [
    "origin = dt.date(2006, 1, 1)                  # Create origin datetime.\n",
    "since_origin = (date - origin).days           # Subtract origin from date 5.\n",
    "time_v = since_origin + 0.5                   # Get fractional days.\n",
    "time_b = [since_origin, since_origin + 1]     # Get time_bnds.\n",
    "\n",
    "print(\"time:\\t\\t%s\\ntime_bnds:\\t%s\" % (time_v, time_b))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Spatial characteristics\n",
    "\n",
    "Now we collect some important spatial characteristics about the rasters. We will open the first raster in the time series and collect the following information:\n",
    "* number of columns\n",
    "* number of rows\n",
    "* position of the raster origin in native coordinate system (via a tuple describing raster's geotransform)\n",
    "\n",
    "**Try it out:**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.144767Z",
     "start_time": "2019-09-09T15:41:19.124127Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cols:\t\t160\n",
      "Rows:\t\t240\n",
      "Transform:\t(-28000.0, 500.0, 0.0, 1230000.0, 0.0, -500.0)\n"
     ]
    }
   ],
   "source": [
    "ras  = gdal.Open(tifs[0])           # Open first geotiff.\n",
    "\n",
    "cols = ras.RasterXSize              # Get the column count.\n",
    "rows = ras.RasterYSize              # Get the row count.\n",
    "gt   = ras.GetGeoTransform()        # Get the geotransform.\n",
    "\n",
    "print(\"Cols:\\t\\t{0}\\nRows:\\t\\t{1}\\nTransform:\\t{2}\".format(cols, rows, gt))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**What is the GeoTransform?**\n",
    "\n",
    "The geotransform (a six item tuple in Python) is GDAL's way of describing the positioning of the raster onto the plane of the its native coordinate system. \n",
    "\n",
    "The coordinates for each pixel are calculated like:\n",
    "```\n",
    "GT = (  -28000.0,    <-   0 x minimum (top left; GeoTIFF origin)\n",
    "           500.0,    <-   1 x resolution\n",
    "             0.0,    <-   2 x rotation\n",
    "       1230000.0,    <-   3 y maximum (top left; GeoTIFF origin)\n",
    "             0.0,    <-   4 y rotation\n",
    "          -500.0  )  <-   5 y resolution\n",
    "```\n",
    "\n",
    "We need to use it to generate one-dimensional arrays of x and y coordinates to write to the netCDF. This is fairly simple with list comprehension:\n",
    "```python\n",
    ">>> [i for i in range(10)]\n",
    "[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]\n",
    "```\n",
    "We use list comprehension to generate the sequences of coordinates based on their positions relative to the raster origin. Get each set of coordinates as a [`numpy.array`](#) and display the first coordinate pair:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.161778Z",
     "start_time": "2019-09-09T15:41:19.148074Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FIRST PIXEL (TL)  : -27750.0, 1229750.0\n",
      "LAST PIXEL  (BR)  : 51750.0, 1110250.0\n",
      "\n",
      "FIRST FIVE  (x, y):\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(array([-27750., -27250., -26750., -26250., -25750.]),\n",
       " array([1229750., 1229250., 1228750., 1228250., 1227750.]))"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the geotransform as its component parts.\n",
    "xmin, xres, xrot, ymax, yrot, yres = gt\n",
    "\n",
    "# Get arrays of x and y coordinates.\n",
    "xarr = np.array([xmin + xres*0.5 + i*xres for i in range(0,cols)])\n",
    "yarr = np.array([ymax + yres*0.5 + i*yres for i in range(0,rows)])\n",
    "\n",
    "\n",
    "### THIS IS ALL PRINTING. YOU DONT NEED THIS. ------> \n",
    "# Print the first and last coordinate pairs.\n",
    "print(\"FIRST PIXEL (TL)  : %s, %s\" % (xarr[0], yarr[0]))\n",
    "print(\"LAST PIXEL  (BR)  : %s, %s\" % (xarr[-1], yarr[-1]))\n",
    "print(\"\\nFIRST FIVE  (x, y):\"); xarr[:5], yarr[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**We also need to calculate *two-dimensional* arrays of latitude and longitude coordinates that represented the geographic positions of the permuted x and y arrays.**\n",
    "\n",
    "We use the `transform` method of `Proj` from the `pyproj` module. \n",
    "```python\n",
    "DEV NOTE: This requires more explanation.\n",
    "```\n",
    "\n",
    "Test for the first pixel's x,y pair and then calculate all of them in a loop:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.215237Z",
     "start_time": "2019-09-09T15:41:19.165513Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FIRST PIXEL (TL)  : -154.5155916837612, 61.049237618268045\n",
      "LAST PIXEL  (BR)  : -153.06963977550794, 59.97850631454614\n"
     ]
    }
   ],
   "source": [
    "srs = osr.SpatialReference()                   # Create an osr.SRS object.\n",
    "srs.ImportFromWkt(ras.GetProjection())         # Import SRS from open GeoTIFF.\n",
    "proj4 = srs.ExportToProj4()                    # Get the proj4 of the SRS. \n",
    "\n",
    "inproj = Proj(proj4)                           # Create Proj for input SRS.\n",
    "outproj = Proj(init=\"epsg:4326\")               # Create Proj for output SRS.\n",
    "\n",
    "# ----------------------------------------------------------------------------\n",
    "#   Permute the one dimensional x and y arrays to create two two dimensional \n",
    "#   arrays that represent all possible combinations of the two. Then, flatten\n",
    "#   the two arrays to one dimension and apply the transform over the paired\n",
    "#   arrays.\n",
    "# ----------------------------------------------------------------------------\n",
    "\n",
    "xarr2d, yarr2d = np.meshgrid(xarr, yarr)       # Permute arrays.  \n",
    "\n",
    "lonarr, latarr = transform(\n",
    "    inproj,                                    # Input raster SRS.\n",
    "    outproj,                                   # Output geographic SRS.\n",
    "    xarr2d.flatten(),                          # Flattened 2d array of x.\n",
    "    yarr2d.flatten())                          # Flattened 2d array of y.\n",
    "\n",
    "lonarr2d = lonarr.reshape(xarr2d.shape)        # Reshape lons back to 2d.\n",
    "latarr2d = latarr.reshape(yarr2d.shape)        # Reshape lats back to 2d.\n",
    "\n",
    "# Check the result.\n",
    "print(\"FIRST PIXEL (TL)  : %s, %s\" % (lonarr2d[0,0], latarr2d[0,0]))\n",
    "print(\"LAST PIXEL  (BR)  : %s, %s\" % (lonarr2d[-1,-1], latarr2d[-1,-1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Don't forget to \"dereference\" the open GeoTIFF dataset:*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.234992Z",
     "start_time": "2019-09-09T15:41:19.224582Z"
    }
   },
   "outputs": [],
   "source": [
    "ras = None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now use all the same methods that were introduced in section 1 (**The basics**) to open a netCDF and write the essential metadata elements, then the data.\n",
    "\n",
    "### Open a netCDF for writing\n",
    "\n",
    "Open a netCDF with the `Dataset` constructor:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.252766Z",
     "start_time": "2019-09-09T15:41:19.239279Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Dataset'>\n",
       "root group (NETCDF4 data model, file format HDF5):\n",
       "    dimensions(sizes): \n",
       "    variables(dimensions): \n",
       "    groups: "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nc = nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\")\n",
    "nc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-03T16:41:30.751852Z",
     "start_time": "2019-09-03T16:41:30.748892Z"
    }
   },
   "source": [
    "### Add dimensions\n",
    "\n",
    "Dimensions are written to the netCDF4 using the createDimension method of netCDF4.Dataset.\n",
    "* `time` is the 'UNLIMITED' dimension (no size).\n",
    "* `x` size comes from the input GeoTIFFs' column count.\n",
    "* `y` from the row count.\n",
    "* `nv` makes space for a special CF variable called `time_bnds`. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.268617Z",
     "start_time": "2019-09-09T15:41:19.256980Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('time',\n",
       "              <class 'netCDF4._netCDF4.Dimension'> (unlimited): name = 'time', size = 0),\n",
       "             ('nv',\n",
       "              <class 'netCDF4._netCDF4.Dimension'>: name = 'nv', size = 2),\n",
       "             ('y',\n",
       "              <class 'netCDF4._netCDF4.Dimension'>: name = 'y', size = 240),\n",
       "             ('x',\n",
       "              <class 'netCDF4._netCDF4.Dimension'>: name = 'x', size = 160)])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nc.createDimension(\"time\")\n",
    "nc.createDimension(\"nv\", 2)\n",
    "nc.createDimension(\"y\", rows)\n",
    "nc.createDimension(\"x\", cols)\n",
    "\n",
    "nc.dimensions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-03T16:45:02.865918Z",
     "start_time": "2019-09-03T16:45:02.861912Z"
    }
   },
   "source": [
    "### Add coordinate variables\n",
    "\n",
    "Variables are written to the netCDF4 using the `createVariable` method of `netCDF4.Dataset`.\n",
    "\n",
    "When I say 'coordinate variables' I mean the variables that play some special role in describing the positions of the data arranged along dimensions (axes) that we created above.\n",
    "\n",
    "#### time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.291212Z",
     "start_time": "2019-09-09T15:41:19.272505Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "float32 time(time)\n",
       "    standard_name: time\n",
       "    long_name: time\n",
       "    calendar: standard\n",
       "    units: days since 2006-01-01 00:00:00 UTC\n",
       "    bounds: time_bnds\n",
       "unlimited dimensions: time\n",
       "current shape = (0,)\n",
       "filling on, default _FillValue of 9.969209968386869e+36 used"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "time = nc.createVariable(\"time\", \"f4\", (\"time\",))\n",
    "time.standard_name = \"time\"\n",
    "time.long_name = \"time\"\n",
    "time.calendar = \"standard\"\n",
    "time.units = \"days since 2006-01-01 00:00:00 UTC\"\n",
    "time.bounds = \"time_bnds\"\n",
    "\n",
    "nc.variables[\"time\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### time_bnds\n",
    "\n",
    "`time` describes the center of the time interval. We need to define another CF-prescribed variable 'time_bnds' that describes the limits of of the observation period."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.307564Z",
     "start_time": "2019-09-09T15:41:19.295621Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "float32 time_bnds(time, nv)\n",
       "    units: days since 2006-01-01 00:00:00 UTC\n",
       "unlimited dimensions: time\n",
       "current shape = (0, 2)\n",
       "filling on, default _FillValue of 9.969209968386869e+36 used"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "time_bnds = nc.createVariable(\"time_bnds\", \"f4\", (\"time\",\"nv\",))\n",
    "time_bnds.units = \"days since 2006-01-01 00:00:00 UTC\" \n",
    "\n",
    "nc.variables[\"time_bnds\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now add the data for the `time` and `time_bnds` variables in a loop using the routine from section 3.2.1:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.492608Z",
     "start_time": "2019-09-09T15:41:19.310564Z"
    }
   },
   "outputs": [],
   "source": [
    "origin = dt.date(2006, 1, 1)\n",
    "\n",
    "for i, t in enumerate(tifs): \n",
    "    \n",
    "    # Get the day from filename.\n",
    "    yyyy, mm, dd = os.path.splitext(t.split(\"_DALL_\")[1])[0].split(\"_\")\n",
    "    date = dt.date(int(yyyy), int(mm), int(dd))\n",
    "\n",
    "    # Get the number of days since the desired origin.\n",
    "    since_origin = (date - origin).days + 1\n",
    "    time[i] = since_origin + 0.5\n",
    "    time_bnds[i] = [since_origin, since_origin + 1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### x and y\n",
    "\n",
    "`x` and `y` are both one dimensional. We already made the array for these variables in section 3.2.2. Add them to the variables:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.517031Z",
     "start_time": "2019-09-09T15:41:19.495608Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "float32 x(x)\n",
       "    units: m\n",
       "    long_name: x coordinate of projection\n",
       "    standard_name: projection_x_coordinate\n",
       "unlimited dimensions: \n",
       "current shape = (160,)\n",
       "filling on, default _FillValue of 9.969209968386869e+36 used"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create x variable.\n",
    "x = nc.createVariable(\"x\", \"f4\", (\"x\",))\n",
    "x.units = \"m\"\n",
    "x.long_name = \"x coordinate of projection\"\n",
    "x.standard_name = \"projection_x_coordinate\"\n",
    "x[:] = xarr                                  # Add the x array from 3.2.2.\n",
    "\n",
    "# Create y variable.\n",
    "y = nc.createVariable(\"y\", \"f4\", (\"y\",))\n",
    "y.units = \"m\"\n",
    "y.long_name = \"y coordinate of projection\"\n",
    "y.standard_name = \"projection_y_coordinate\"\n",
    "y[:] = yarr                                  # Add the y array from 3.2.2.\n",
    "\n",
    "\n",
    "nc.variables[\"x\"]                            # Print the x variable."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### lon and lat\n",
    "\n",
    "`lon` and `lat` are both two dimensional. We already made the array for these variables in section 3.2.2. Add them to the variables:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.543816Z",
     "start_time": "2019-09-09T15:41:19.521898Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "float32 lon(y, x)\n",
       "    units: degrees_east\n",
       "    long_name: longitude coordinate\n",
       "    standard_name: longitude\n",
       "unlimited dimensions: \n",
       "current shape = (240, 160)\n",
       "filling on, default _FillValue of 9.969209968386869e+36 used"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create longitude variable.\n",
    "lon = nc.createVariable(\"lon\", \"f4\", (\"y\", \"x\",))\n",
    "lon.units = \"degrees_east\"\n",
    "lon.long_name = \"longitude coordinate\"\n",
    "lon.standard_name = \"longitude\"\n",
    "lon[:,:] = lonarr2d                          # Add lon array from 3.2.2.\n",
    "\n",
    "# Create latitude variable.\n",
    "lat = nc.createVariable(\"lat\", \"f4\", (\"y\", \"x\",))\n",
    "lat.units = \"degrees_north\"\n",
    "lat.long_name = \"latitude coordinate\"\n",
    "lat.standard_name = \"latitude\"\n",
    "lat[:,:] = latarr2d                          # Add lat array from 3.2.2.\n",
    "\n",
    "\n",
    "nc.variables[\"lon\"]                          # Print the lon variable."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add crs variable\n",
    "\n",
    "The crs variable is another special CF variable that describes the `grid_mapping` according to the CF convention for projected data. We get this information from via one of a gdal.Dataset's proj methods. The crs variable does not store any data - only attributes.\n",
    "\n",
    "Add the `crs` variable to the dataset and pull all of the projection parameters from the SRS object that was created in section 3.3.2. Add the parameters to the variable as attributes:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.569363Z",
     "start_time": "2019-09-09T15:41:19.550224Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "int16 albers_conical_equal_area()\n",
       "    grid_mapping_name: albers_conical_equal_area\n",
       "    longitude_of_central_meridian: -154.0\n",
       "    latitude_of_projection_origin: 50.0\n",
       "    standard_parallel: [55. 65.]\n",
       "    false_easting: 0.0\n",
       "    false_northing: 0.0\n",
       "    semi_major_axis: 6378137.0\n",
       "    inverse_flattening: 298.2572221010042\n",
       "    crs_wkt: PROJCS[\"unnamed\",GEOGCS[\"NAD83\",DATUM[\"North_American_Datum_1983\",SPHEROID[\"GRS 1980\",6378137,298.2572221010042,AUTHORITY[\"EPSG\",\"7019\"]],TOWGS84[0,0,0,0,0,0,0],AUTHORITY[\"EPSG\",\"6269\"]],PRIMEM[\"Greenwich\",0],UNIT[\"degree\",0.0174532925199433],AUTHORITY[\"EPSG\",\"4269\"]],PROJECTION[\"Albers_Conic_Equal_Area\"],PARAMETER[\"standard_parallel_1\",55],PARAMETER[\"standard_parallel_2\",65],PARAMETER[\"latitude_of_center\",50],PARAMETER[\"longitude_of_center\",-154],PARAMETER[\"false_easting\",0],PARAMETER[\"false_northing\",0],UNIT[\"metre\",1,AUTHORITY[\"EPSG\",\"9001\"]]]\n",
       "unlimited dimensions: \n",
       "current shape = ()\n",
       "filling on, default _FillValue of -32767 used"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create the CRS variable. \n",
    "grid_mapping_name = \"albers_conical_equal_area\"  # This is a CF standard.\n",
    "crs = nc.createVariable(grid_mapping_name, \"i2\")\n",
    "\n",
    "# Populate the attributes using the osr.SRS. Some required by GMN.\n",
    "crs.grid_mapping_name = grid_mapping_name\n",
    "crs.longitude_of_central_meridian = srs.GetProjParm(\"longitude_of_center\")\n",
    "crs.latitude_of_projection_origin = srs.GetProjParm(\"latitude_of_origin\")\n",
    "crs.standard_parallel = [srs.GetProjParm(\"standard_parallel_1\"),\n",
    "                         srs.GetProjParm(\"standard_parallel_2\") ]\n",
    "crs.false_easting = srs.GetProjParm(\"false_easting\")\n",
    "crs.false_northing = srs.GetProjParm(\"false_northing\")\n",
    "crs.semi_major_axis = srs.GetSemiMajor()\n",
    "crs.inverse_flattening = srs.GetInvFlattening()\n",
    "crs.crs_wkt = srs.ExportToWkt()\n",
    "\n",
    "\n",
    "crs # Print the crs variable."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add data variable\n",
    "\n",
    "This example only merges one variable from the time series of GeoTIFFs, so we'll only write one data variable to the output netCDF. Data will be arranged over `time`, `y`, and `x` dimensions. Don't forget to check the CF standard names table for appropriate standard_name attribute.\n",
    "\n",
    "Add a variable called `SnowDepth` to the output file, and give it some attributes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:19.589712Z",
     "start_time": "2019-09-09T15:41:19.572974Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Variable'>\n",
       "float32 SnowDepth_500m(time, y, x)\n",
       "    _FillValue: -9999.0\n",
       "    grid_mapping: albers_conical_equal_area\n",
       "    long_name: SnowModel snow depth\n",
       "    units: cm\n",
       "unlimited dimensions: time\n",
       "current shape = (365, 240, 160)\n",
       "filling on"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = nc.createVariable(\n",
    "    \"SnowDepth_500m\",            # Specify variable name.\n",
    "    \"f4\",                        # Specify data type.\n",
    "    (\"time\", \"y\", \"x\"),          # Specify dimensions.\n",
    "    zlib=True,                   # Specify compression with zlib.\n",
    "    complevel=4,                 # Specify compression level.\n",
    "    fill_value=-9999.)           # Specify the fill value.\n",
    "\n",
    "# Set a few attributes.\n",
    "\n",
    "data.grid_mapping = grid_mapping_name            # CF required.\n",
    "#data.standard_name = None                       # CF strongly encouraged.\n",
    "data.long_name = \"SnowModel snow depth\"          # CF encouraged.\n",
    "data.units = \"cm\"                                # CF required.\n",
    "\n",
    "nc.variables[\"SnowDepth_500m\"]   # Print the data variable."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, add the data from each GeoTIFF in a loop."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:23.265316Z",
     "start_time": "2019-09-09T15:41:19.592250Z"
    }
   },
   "outputs": [],
   "source": [
    "for i, t in enumerate(tifs):     # Loop over GeoTIFFs.\n",
    "    ras  = gdal.Open(t)          # Open it.\n",
    "    band = ras.GetRasterBand(1)  # Get the first band.\n",
    "    arr  = band.ReadAsArray()    # Get data as a 2d numpy.ndarray.\n",
    "    ras  = None                  # Dereference the open GeoTIFF.\n",
    "\n",
    "    arr[np.isnan(arr)] = -9999.  # Set missing values to -9999.\n",
    "    data[i,:,:] = arr            # Add the array to the current timestep.\n",
    "    nc.sync()                    # Write changes to disk."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add global attributes\n",
    "Finally, write some global attributes to the file.\n",
    "\n",
    "Think of the global attributes as file-level attributes that describe the dataset in a way that is similar to a header in a table's header. Some attributes are CF standard attributes. Any information that will be helpful to your record keeping, or potentially useful to an end user, can go into a global attribute."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:23.285781Z",
     "start_time": "2019-09-09T15:41:23.270021Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<class 'netCDF4._netCDF4.Dataset'>\n",
       "root group (NETCDF4 data model, file format HDF5):\n",
       "    Conventions: CF-1.6\n",
       "    archive: ORNL DAAC\n",
       "    dimensions(sizes): time(365), nv(2), y(240), x(160)\n",
       "    variables(dimensions): float32 \u001b[4mtime\u001b[0m(time), float32 \u001b[4mtime_bnds\u001b[0m(time,nv), float32 \u001b[4mx\u001b[0m(x), float32 \u001b[4my\u001b[0m(y), float32 \u001b[4mlon\u001b[0m(y,x), float32 \u001b[4mlat\u001b[0m(y,x), int16 \u001b[4malbers_conical_equal_area\u001b[0m(), float32 \u001b[4mSnowDepth_500m\u001b[0m(time,y,x)\n",
       "    groups: "
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nc.Conventions = \"CF-1.6\"\n",
    "nc.archive     = \"ORNL DAAC\"\n",
    "nc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Close the file and check\n",
    "\n",
    "Close the file, then admire your work by printing the header with `ncdump`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:23.448263Z",
     "start_time": "2019-09-09T15:41:23.291861Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "netcdf MyFirstNetCDF {\n",
      "dimensions:\n",
      "\ttime = UNLIMITED ; // (365 currently)\n",
      "\tnv = 2 ;\n",
      "\ty = 240 ;\n",
      "\tx = 160 ;\n",
      "variables:\n",
      "\tfloat time(time) ;\n",
      "\t\ttime:standard_name = \"time\" ;\n",
      "\t\ttime:long_name = \"time\" ;\n",
      "\t\ttime:calendar = \"standard\" ;\n",
      "\t\ttime:units = \"days since 2006-01-01 00:00:00 UTC\" ;\n",
      "\t\ttime:bounds = \"time_bnds\" ;\n",
      "\tfloat time_bnds(time, nv) ;\n",
      "\t\ttime_bnds:units = \"days since 2006-01-01 00:00:00 UTC\" ;\n",
      "\tfloat x(x) ;\n",
      "\t\tx:units = \"m\" ;\n",
      "\t\tx:long_name = \"x coordinate of projection\" ;\n",
      "\t\tx:standard_name = \"projection_x_coordinate\" ;\n",
      "\tfloat y(y) ;\n",
      "\t\ty:units = \"m\" ;\n",
      "\t\ty:long_name = \"y coordinate of projection\" ;\n",
      "\t\ty:standard_name = \"projection_y_coordinate\" ;\n",
      "\tfloat lon(y, x) ;\n",
      "\t\tlon:units = \"degrees_east\" ;\n",
      "\t\tlon:long_name = \"longitude coordinate\" ;\n",
      "\t\tlon:standard_name = \"longitude\" ;\n",
      "\tfloat lat(y, x) ;\n",
      "\t\tlat:units = \"degrees_north\" ;\n",
      "\t\tlat:long_name = \"latitude coordinate\" ;\n",
      "\t\tlat:standard_name = \"latitude\" ;\n",
      "\tshort albers_conical_equal_area ;\n",
      "\t\talbers_conical_equal_area:grid_mapping_name = \"albers_conical_equal_area\" ;\n",
      "\t\talbers_conical_equal_area:longitude_of_central_meridian = -154. ;\n",
      "\t\talbers_conical_equal_area:latitude_of_projection_origin = 50. ;\n",
      "\t\talbers_conical_equal_area:standard_parallel = 55., 65. ;\n",
      "\t\talbers_conical_equal_area:false_easting = 0. ;\n",
      "\t\talbers_conical_equal_area:false_northing = 0. ;\n",
      "\t\talbers_conical_equal_area:semi_major_axis = 6378137. ;\n",
      "\t\talbers_conical_equal_area:inverse_flattening = 298.257222101004 ;\n",
      "\t\talbers_conical_equal_area:crs_wkt = \"PROJCS[\\\"unnamed\\\",GEOGCS[\\\"NAD83\\\",DATUM[\\\"North_American_Datum_1983\\\",SPHEROID[\\\"GRS 1980\\\",6378137,298.2572221010042,AUTHORITY[\\\"EPSG\\\",\\\"7019\\\"]],TOWGS84[0,0,0,0,0,0,0],AUTHORITY[\\\"EPSG\\\",\\\"6269\\\"]],PRIMEM[\\\"Greenwich\\\",0],UNIT[\\\"degree\\\",0.0174532925199433],AUTHORITY[\\\"EPSG\\\",\\\"4269\\\"]],PROJECTION[\\\"Albers_Conic_Equal_Area\\\"],PARAMETER[\\\"standard_parallel_1\\\",55],PARAMETER[\\\"standard_parallel_2\\\",65],PARAMETER[\\\"latitude_of_center\\\",50],PARAMETER[\\\"longitude_of_center\\\",-154],PARAMETER[\\\"false_easting\\\",0],PARAMETER[\\\"false_northing\\\",0],UNIT[\\\"metre\\\",1,AUTHORITY[\\\"EPSG\\\",\\\"9001\\\"]]]\" ;\n",
      "\tfloat SnowDepth_500m(time, y, x) ;\n",
      "\t\tSnowDepth_500m:_FillValue = -9999.f ;\n",
      "\t\tSnowDepth_500m:grid_mapping = \"albers_conical_equal_area\" ;\n",
      "\t\tSnowDepth_500m:long_name = \"SnowModel snow depth\" ;\n",
      "\t\tSnowDepth_500m:units = \"cm\" ;\n",
      "\n",
      "// global attributes:\n",
      "\t\t:Conventions = \"CF-1.6\" ;\n",
      "\t\t:archive = \"ORNL DAAC\" ;\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "nc.close()\n",
    "\n",
    "!ncdump -h MyFirstNetCDF.nc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Write the netCDF again in a script!\n",
    "\n",
    "The cell below does everything we learned above:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-09T15:41:27.974619Z",
     "start_time": "2019-09-09T15:41:23.451472Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "netcdf MyFirstNetCDF {\n",
      "dimensions:\n",
      "\ttime = UNLIMITED ; // (365 currently)\n",
      "\tnv = 2 ;\n",
      "\ty = 240 ;\n",
      "\tx = 160 ;\n",
      "variables:\n",
      "\tfloat time(time) ;\n",
      "\t\ttime:standard_name = \"time\" ;\n",
      "\t\ttime:long_name = \"time\" ;\n",
      "\t\ttime:calendar = \"standard\" ;\n",
      "\t\ttime:units = \"days since 1970-01-01 00:00:00 UTC\" ;\n",
      "\t\ttime:bounds = \"time_bnds\" ;\n",
      "\tfloat time_bnds(time, nv) ;\n",
      "\t\ttime_bnds:units = \"days since 1970-01-01 00:00:00 UTC\" ;\n",
      "\tfloat x(x) ;\n",
      "\t\tx:units = \"m\" ;\n",
      "\t\tx:long_name = \"x coordinate of projection\" ;\n",
      "\t\tx:standard_name = \"projection_x_coordinate\" ;\n",
      "\tfloat y(y) ;\n",
      "\t\ty:units = \"m\" ;\n",
      "\t\ty:long_name = \"y coordinate of projection\" ;\n",
      "\t\ty:standard_name = \"projection_y_coordinate\" ;\n",
      "\tfloat lon(y, x) ;\n",
      "\t\tlon:units = \"degrees_east\" ;\n",
      "\t\tlon:long_name = \"longitude coordinate\" ;\n",
      "\t\tlon:standard_name = \"longitude\" ;\n",
      "\tfloat lat(y, x) ;\n",
      "\t\tlat:units = \"degrees_north\" ;\n",
      "\t\tlat:long_name = \"latitude coordinate\" ;\n",
      "\t\tlat:standard_name = \"latitude\" ;\n",
      "\tshort albers_conical_equal_area ;\n",
      "\t\talbers_conical_equal_area:grid_mapping_name = \"albers_conical_equal_area\" ;\n",
      "\t\talbers_conical_equal_area:longitude_of_central_meridian = -154. ;\n",
      "\t\talbers_conical_equal_area:latitude_of_projection_origin = 50. ;\n",
      "\t\talbers_conical_equal_area:standard_parallel = 55., 65. ;\n",
      "\t\talbers_conical_equal_area:false_easting = 0. ;\n",
      "\t\talbers_conical_equal_area:false_northing = 0. ;\n",
      "\t\talbers_conical_equal_area:semi_major_axis = 6378137. ;\n",
      "\t\talbers_conical_equal_area:inverse_flattening = 298.257222101004 ;\n",
      "\t\talbers_conical_equal_area:crs_wkt = \"PROJCS[\\\"unnamed\\\",GEOGCS[\\\"NAD83\\\",DATUM[\\\"North_American_Datum_1983\\\",SPHEROID[\\\"GRS 1980\\\",6378137,298.2572221010042,AUTHORITY[\\\"EPSG\\\",\\\"7019\\\"]],TOWGS84[0,0,0,0,0,0,0],AUTHORITY[\\\"EPSG\\\",\\\"6269\\\"]],PRIMEM[\\\"Greenwich\\\",0],UNIT[\\\"degree\\\",0.0174532925199433],AUTHORITY[\\\"EPSG\\\",\\\"4269\\\"]],PROJECTION[\\\"Albers_Conic_Equal_Area\\\"],PARAMETER[\\\"standard_parallel_1\\\",55],PARAMETER[\\\"standard_parallel_2\\\",65],PARAMETER[\\\"latitude_of_center\\\",50],PARAMETER[\\\"longitude_of_center\\\",-154],PARAMETER[\\\"false_easting\\\",0],PARAMETER[\\\"false_northing\\\",0],UNIT[\\\"metre\\\",1,AUTHORITY[\\\"EPSG\\\",\\\"9001\\\"]]]\" ;\n",
      "\tfloat SnowDepth_500m(time, y, x) ;\n",
      "\t\tSnowDepth_500m:_FillValue = -9999.f ;\n",
      "\t\tSnowDepth_500m:grid_mapping = \"albers_conical_equal_area\" ;\n",
      "\t\tSnowDepth_500m:standard_name = \"SOME_STANDARD_NAME\" ;\n",
      "\t\tSnowDepth_500m:long_name = \"SnowModel snow depth\" ;\n",
      "\t\tSnowDepth_500m:units = \"cm\" ;\n",
      "\n",
      "// global attributes:\n",
      "\t\t:Conventions = \"CF-1.6\" ;\n",
      "\t\t:archive = \"ORNL DAAC\" ;\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "# Open a netCDF4.Dataset in write mode.\n",
    "with nc4.Dataset(\"MyFirstNetCDF.nc\", \"w\") as nc:\n",
    "\n",
    "    ##########################################################################\n",
    "    #                  Create time, x, y, and nv dimensions.                  \n",
    "    ##########################################################################\n",
    "    #\n",
    "    #   Dimensions are written to the netCDF4 using the createDimension method\n",
    "    #   of netCDF4.Dataset.\n",
    "    #\n",
    "    #  * 'time' is the 'UNLIMITED' dimension (no size).\n",
    "    #  * 'x' size comes from the input GeoTIFFs' column count.\n",
    "    #  * 'y' from the row count.\n",
    "    #  * 'nv' makes space for a special CF variable called 'time_bnds'. \n",
    "    #\n",
    "    # ------------------------------------------------------------------------\n",
    "    \n",
    "    nc.createDimension(\"time\")\n",
    "    nc.createDimension(\"nv\", 2)\n",
    "    nc.createDimension(\"y\", rows)\n",
    "    nc.createDimension(\"x\", cols)\n",
    "    \n",
    "    ##########################################################################\n",
    "    #                     Create 'coordinate' variables.                     \n",
    "    ##########################################################################\n",
    "    #\n",
    "    #   Variables are written to the netCDF4 using the createVariable method\n",
    "    #   of netCDF4.Dataset.\n",
    "    #\n",
    "    #   I'm not sure if 'coordinate' variable is the term used by CF, but \n",
    "    #   when I say coordinate variables I mean the variables that play some \n",
    "    #   special role in describing the positions of the data arranged along \n",
    "    #   dimensions (axes) that we created above.\n",
    "    #\n",
    "    # ------------------------------------------------------------------------\n",
    "    \n",
    "    # Create time variable.\n",
    "    \n",
    "    time = nc.createVariable(\"time\", \"f4\", (\"time\",))\n",
    "    time.standard_name = \"time\"\n",
    "    time.long_name = \"time\"\n",
    "    time.calendar = \"standard\"\n",
    "    time.units = \"days since 1970-01-01 00:00:00 UTC\"\n",
    "    time.bounds = \"time_bnds\"\n",
    "    \n",
    "    # ------------------------------------------------------------------------\n",
    "    #   time_bnds: 'time' describes the center of the time interval. We need \n",
    "    #   to define another CF-prescribed variable 'time_bnds' that describes \n",
    "    #   the limits of of the observation period.\n",
    "    # ------------------------------------------------------------------------\n",
    "    \n",
    "    # Create time_bnds variable.\n",
    "    \n",
    "    time_bnds = nc.createVariable(\"time_bnds\", \"f4\", (\"time\",\"nv\",))\n",
    "    time_bnds.units = \"days since 1970-01-01 00:00:00 UTC\"   \n",
    "    \n",
    "    # Use routine from section 1.1.1 to write the time and time_bnds values.\n",
    "    \n",
    "    origin = dt.date(1970, 1, 1)\n",
    "    \n",
    "    for i, t in enumerate(tifs): \n",
    "        \n",
    "        yyyy, mm, dd = os.path.splitext(t.split(\"_DALL_\")[1])[0].split(\"_\")\n",
    "        date = dt.date(int(yyyy), int(mm), int(dd))\n",
    "        \n",
    "        since_origin = (date - origin).days\n",
    "        time[i] = since_origin + 0.5\n",
    "        time_bnds[i] = [since_origin, since_origin + 1]\n",
    "  \n",
    "    # ------------------------------------------------------------------------\n",
    "    #   'x' and 'y' are one dimensional. We already made these. Add variables.\n",
    "    # ------------------------------------------------------------------------\n",
    "    \n",
    "    # Create x variable.\n",
    "    \n",
    "    x = nc.createVariable(\"x\", \"f4\", (\"x\",))\n",
    "    x.units = \"m\"\n",
    "    x.long_name = \"x coordinate of projection\"\n",
    "    x.standard_name = \"projection_x_coordinate\"\n",
    "    x[:] = xarr                                  # Add the x array from 3.2.2.\n",
    "\n",
    "    # Create y variable.\n",
    "    \n",
    "    y = nc.createVariable(\"y\", \"f4\", (\"y\",))\n",
    "    y.units = \"m\"\n",
    "    y.long_name = \"y coordinate of projection\"\n",
    "    y.standard_name = \"projection_y_coordinate\"\n",
    "    y[:] = yarr                                  # Add the y array from 3.2.2.\n",
    "\n",
    "    # ------------------------------------------------------------------------\n",
    "    #   'lon' and 'lat' are two dimensional. Add variables.\n",
    "    # ------------------------------------------------------------------------\n",
    "    \n",
    "    # Create longitude variable.\n",
    "    \n",
    "    lon = nc.createVariable(\"lon\", \"f4\", (\"y\", \"x\",))\n",
    "    lon.units = \"degrees_east\"\n",
    "    lon.long_name = \"longitude coordinate\"\n",
    "    lon.standard_name = \"longitude\"\n",
    "    lon[:,:] = lonarr2d                          # Add lon array from 3.2.2.\n",
    "    \n",
    "    # Create latitude variable.\n",
    "    \n",
    "    lat = nc.createVariable(\"lat\", \"f4\", (\"y\", \"x\",))\n",
    "    lat.units = \"degrees_north\"\n",
    "    lat.long_name = \"latitude coordinate\"\n",
    "    lat.standard_name = \"latitude\"\n",
    "    lat[:,:] = latarr2d                          # Add lat array from 3.2.2.\n",
    "         \n",
    "    ##########################################################################\n",
    "    #                     Create 'crs' variable.                     \n",
    "    ##########################################################################\n",
    "    #\n",
    "    #   The crs variable is another special CF variable that describes the \n",
    "    #   'grid_mapping' according to the CF convention for projected data. We\n",
    "    #   get this information from via one of a gdal.Dataset's proj methods.\n",
    "    #\n",
    "    #   The crs variable does not store any data - only attributes.\n",
    "    #\n",
    "    # ------------------------------------------------------------------------\n",
    "\n",
    "    # Create the CRS variable. \n",
    "    \n",
    "    grid_mapping_name = \"albers_conical_equal_area\"  # This is a CF standard.\n",
    "    crs = nc.createVariable(grid_mapping_name, \"i2\")\n",
    "    \n",
    "    # Populate the attributes using the osr.SRS. Some required by GMN.\n",
    "    \n",
    "    crs.grid_mapping_name = grid_mapping_name\n",
    "    crs.longitude_of_central_meridian = srs.GetProjParm(\"longitude_of_center\")\n",
    "    crs.latitude_of_projection_origin = srs.GetProjParm(\"latitude_of_origin\")\n",
    "    crs.standard_parallel = [srs.GetProjParm(\"standard_parallel_1\"),\n",
    "                             srs.GetProjParm(\"standard_parallel_2\") ]\n",
    "    crs.false_easting = srs.GetProjParm(\"false_easting\")\n",
    "    crs.false_northing = srs.GetProjParm(\"false_northing\")\n",
    "    crs.semi_major_axis = srs.GetSemiMajor()\n",
    "    crs.inverse_flattening = srs.GetInvFlattening()\n",
    "    crs.crs_wkt = srs.ExportToWkt()\n",
    "\n",
    "    ##########################################################################\n",
    "    #                     Create the data variable.                     \n",
    "    ##########################################################################\n",
    "    #\n",
    "    #   This example only merges one variable time series, so we'll only write\n",
    "    #   one data variable to the output netCDF. Data will be arranged over \n",
    "    #   time, y, and x dimensions. Don't forget to check the CF standard names\n",
    "    #   table for appropriate standard_name attribute.\n",
    "    #\n",
    "    # ------------------------------------------------------------------------\n",
    "\n",
    "    # Create data variable.\n",
    "    \n",
    "    data = nc.createVariable(\n",
    "        \"SnowDepth_500m\",            # UPDATE NAME (Don't know these data.)\n",
    "        \"f4\",                        # Specify data type.\n",
    "        (\"time\", \"y\", \"x\"),          # Specify dimensions.\n",
    "        zlib=True,                   # Compress with zlib.\n",
    "        complevel=4,                 # Specify compression level.\n",
    "        fill_value=-9999.)           # Specify the fill value.\n",
    "    \n",
    "    # Set a few attributes.\n",
    "    \n",
    "    data.grid_mapping = grid_mapping_name            # CF required.\n",
    "    data.standard_name = \"SOME_STANDARD_NAME\"        # CF strongly encouraged.\n",
    "    data.long_name = \"SOME DESCRIPTIVE LONG NAME\"    # CF encouraged.\n",
    "    data.units = \"SOME UNITS\"                        # CF required.\n",
    "    \n",
    "    data.grid_mapping = grid_mapping_name            # CF required.\n",
    "    #data.standard_name = None                       # CF strongly encouraged.\n",
    "    data.long_name = \"SnowModel snow depth\"          # CF encouraged.\n",
    "    data.units = \"cm\"                                # CF required.\n",
    "    \n",
    "    # Add the data from each GeoTIFF in a loop.\n",
    "\n",
    "    for i, t in enumerate(tifs):     # Loop over GeoTIFFs.\n",
    "        ras  = gdal.Open(t)          # Open it.\n",
    "        band = ras.GetRasterBand(1)  # Get the first band.\n",
    "        arr  = band.ReadAsArray()    # Get data as a 2d numpy.ndarray.\n",
    "        ras  = None                  # Dereference the open GeoTIFF.\n",
    "        \n",
    "        arr[np.isnan(arr)] = -9999.  # Set missing values to -9999.\n",
    "        data[i,:,:] = arr            # Add the array to the current timestep.\n",
    "        nc.sync                      # Write changes to disk.\n",
    "        \n",
    "    ##########################################################################\n",
    "    #                     Write the global attributes.                     \n",
    "    ##########################################################################\n",
    "    #\n",
    "    #   Think of the global attributes as file-level attributes that describe\n",
    "    #   the dataset in a way that is similar to a header in a tabular dataset.\n",
    "    #   Some attributes are CF standard attributes. You can put what you like.\n",
    "    #   \n",
    "    # ------------------------------------------------------------------------\n",
    "\n",
    "    nc.Conventions = \"CF-1.6\"\n",
    "    nc.archive     = \"ORNL DAAC\"\n",
    "    \n",
    "\n",
    "# Display the result with ncdump.\n",
    "!ncdump -h MyFirstNetCDF.nc"
   ]
  },
  {
   "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": "3.7.4"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": false,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": true,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "165px"
   },
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}