ARD resampling for pyCCD

environment.yaml

name: resample
channels:
    - conda-forge
dependencies:
    # pyccd deps
    - numpy
    - scipy
    - scikit-learn
    - click
    - click-plugins
    - cachetools
    - pyyaml
    # read / resample data
    - xarray
    # notebook and visualize
    - jupyter
    - notebook
    - matplotlib
    - pip:
        - git+https://github.com/USGS-EROS/lcmap-pyccd.git

h03v09_ard_example.nc

resample.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Resampling ARD"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "One approach to achieving spatial-temporal consistency when using ARD is to resample the data to some standard (or regular) temporal interval before running anything.\n",
    "\n",
    "Resampling, often referred to as compositing within remote sensing literature, is the process of grouping observations by time and reducing this group according to some criteria. One commonly used example of this is generating monthly \"maximum NDVI\" composites, often at an annual or monthly time intervals. Even more sophisticated approaches, like the \"Best Available Pixel\" approach from Wulder _et al_, can be described by this logic.\n",
    "\n",
    "The resample operation can be considered to have the following parameters:\n",
    "\n",
    "* Dimension (in this example, the time dimension)\n",
    "* Interval\n",
    "* Resampling criteria (usually a band or index)\n",
    "* Reduction function (e.g., \"max\" or \"min\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime as dt\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import toolz\n",
    "import xarray as xr\n",
    "\n",
    "import ccd\n",
    "\n",
    "plt.style.use('ggplot')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<xarray.Dataset>\n",
       "Dimensions:  (time: 2916, x: 10, y: 10)\n",
       "Coordinates:\n",
       "  * y        (y) float64 1.935e+06 1.935e+06 1.935e+06 1.935e+06 1.935e+06 ...\n",
       "  * x        (x) float64 -2.11e+06 -2.11e+06 -2.11e+06 -2.109e+06 -2.109e+06 ...\n",
       "    crs      int32 ...\n",
       "  * time     (time) datetime64[ns] 1982-11-15 1982-11-24 1982-12-01 ...\n",
       "Data variables:\n",
       "    blue     (time, y, x) float64 ...\n",
       "    green    (time, y, x) float64 ...\n",
       "    red      (time, y, x) float64 ...\n",
       "    nir      (time, y, x) float64 ...\n",
       "    swir1    (time, y, x) float64 ...\n",
       "    swir2    (time, y, x) float64 ...\n",
       "    bt       (time, y, x) float64 ...\n",
       "    cfmask   (time, y, x) float64 ...\n",
       "Attributes:\n",
       "    crs:          +datum=WGS84 +lat_0=23 +lat_1=29.5 +lat_2=45.5 +lon_0=-96 +...\n",
       "    crs_wkt:      PROJCS[\"unnamed\",GEOGCS[\"WGS 84\",DATUM[\"WGS_1984\",SPHEROID[...\n",
       "    Conventions:  CF-1.6\n",
       "    spatial_ref:  PROJCS[\"unnamed\",GEOGCS[\"WGS 84\",DATUM[\"WGS_1984\",SPHEROID[..."
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_res

resample_example.pdf