===========
Get started
===========



.. _get-forest-cover-change-from-tmf:

Get forest cover change from TMF
--------------------------------

The function ``.get_fcc_loss()`` can be used to download forest cover change
from the Tropical Moist Forest product. We will use the Reunion Island
(isocode “REU”) as a case study.

.. code:: python

    from pathlib import Path

    import ee
    import geefcc
    import rioxarray

.. code:: python

    # Initialize GEE
    ee.Initialize(project="deforisk", opt_url="https://earthengine-highvolume.googleapis.com")

.. code:: python

    # Download data from GEE
    out_dir = Path("out_tmf")
    forest_file = out_dir / "forest_tmf.tif"
    if not forest_file.is_file():
        geefcc.get_fcc_loss(
            aoi="REU",
            years=[2000, 2010, 2020],
            source="tmf",
            parallel=False,
            crop_to_aoi=True,
            tile_size=0.5,
            output_file=forest_file,
        )

::

    get_fcc running, 3 tiles ....


We transform the forest raster with three bands into a single-band forest cover change raster.

.. code:: python

    fcc_file = out_dir / "fcc_tmf.tif"
    if not fcc_file.is_file():
        geefcc.sum_raster_bands(
            input_file=forest_file,
            output_file=fcc_file,
            verbose=False,
        )

We plot the forest cover change map.

.. code:: python

    # Plot
    years = [2000, 2010, 2020]
    geefcc.plot_fcc_loss(
        input_file=fcc_file,
        years=years,
        output_file="tmf.png",
        title="Forest cover change 2000-2010-2020, TMF",
        dpi=100,
    )

.. image:: tmf.png
    :width: 800
    :align: center

.. _compare-with-forest-cover-change-from-gfc:

Compare with forest cover change from GFC
-----------------------------------------

.. code:: python

    # Get data from GEE
    out_dir = Path("out_gfc_50")
    forest_file = out_dir / "forest_gfc_50.tif"
    if not forest_file.is_file():
        geefcc.get_fcc_loss(
            aoi="REU",
            years=[2001, 2010, 2020],  # Here, first year must be 2001 (1st Jan)
            source="gfc",
            perc=50,
            parallel=False,
            crop_to_aoi=True,
            tile_size=0.5,
            output_file=forest_file,
        )

::

    get_fcc running, 3 tiles ....


.. code:: python

    # Sum bands to get a single-band fcc raster
    fcc_file = out_dir / "fcc_gfc_50.tif"
    if not fcc_file.is_file():
        geefcc.sum_raster_bands(
            input_file=forest_file,
            output_file=fcc_file,
            verbose=False,
        )

.. code:: python

    # Plot
    years_gfc = [2001, 2010, 2020]
    geefcc.plot_fcc_loss(
        input_file=fcc_file,
        years=years_gfc,
        output_file="gfc.png",
        title="Forest cover change 2001-2010-2020, GFC",
        dpi=100,
    )

.. image:: gfc.png
    :width: 800
    :align: center

.. _comparing-forest-cover-in-2020-between-tmf-and-gfc:

Comparing forest cover in 2020 between TMF and GFC
--------------------------------------------------

.. code:: python

    import matplotlib.patches as mpatches
    import matplotlib.pyplot as plt
    from matplotlib.colors import ListedColormap

    # Computing difference and sum
    forest_tmf = rioxarray.open_rasterio(Path("out_tmf") / "forest_tmf.tif").astype("int")
    forest_gfc = rioxarray.open_rasterio(Path("out_gfc_50") / "forest_gfc_50.tif").astype("int")
    forest_diff = forest_tmf.sel(band=3) - forest_gfc.sel(band=3)
    forest_sum = forest_tmf.sel(band=3) + forest_gfc.sel(band=3)
    forest_diff = forest_diff.where(forest_sum != 0, -2)

.. code:: python

    # Colors
    cols=[(10, 10, 150, 255), (34, 139, 34, 255), (200, 200, 0, 255)]
    colors = [(1, 1, 1, 0)]  # transparent white for -2
    cmax = 255.0
    for col in cols:
        col_class = tuple([i / cmax for i in col])
        colors.append(col_class)
    color_map = ListedColormap(colors)

    # Labels
    labels = {0: "non-forest tmf, non-forest gfc", 1: "non-forest tmf / forest gfc",
              2: "forest tmf / forest gfc", 3: "forest tmf, non-forest gfc"}
    patches = [mpatches.Patch(facecolor=col, edgecolor="black",
                              label=labels[i]) for (i, col) in enumerate(colors)]

    # Plot
    fig = plt.figure()
    ax = plt.subplot(111)
    ds = forest_diff.squeeze()
    nrow, ncol = ds.shape
    xmin = float(ds.x.min())
    xmax = float(ds.x.max())
    ymin = float(ds.y.min())
    ymax = float(ds.y.max())
    xres = (xmax - xmin) / (ncol - 1)
    yres = (ymax - ymin) / (nrow - 1)
    extent = [xmin - xres / 2, xmax + xres / 2,
              ymin - yres / 2, ymax + yres / 2]
    ax.imshow(ds.values, cmap=color_map, extent=extent, resample=False)
    ax.set_aspect("equal")
    plt.title("Difference between TMF and GFC for forest cover in 2020")
    plt.legend(handles=patches, bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
    fig.savefig("comp.png", bbox_inches="tight", dpi=100)
    plt.close(fig)

.. image:: comp.png
    :width: 800
    :align: center

Differences are quite important between the two data-sets. This might
change depending on the tree cover threshold we select for defining
forest with the GFC dataset.

.. _download-data-from-an-extent:

Download data from an extent
----------------------------

We will use the following extent which corresponds to a region around
the Analamazaotra special reserve in Madagascar.

.. code:: python

    out_dir = Path("out_tmf_extent")
    forest_file = out_dir / "forest_tmf_extent.tif"
    if not forest_file.is_file():
        geefcc.get_fcc_loss(
            aoi=(48.4, -19.0, 48.6, -18.8),
            years=[2000, 2010, 2020],
            source="tmf",
            tile_size=0.2,
            output_file=forest_file,
        )

::

    get_fcc running, 4 tiles .....


.. code:: python

    # Sum bands to get a single-band fcc raster
    fcc_file = out_dir / "fcc_tmf_extent.tif"
    if not fcc_file.is_file():
        geefcc.sum_raster_bands(
            input_file=forest_file,
            output_file=fcc_file,
            verbose=False,
        )

.. code:: python

    # Plot
    geefcc.plot_fcc_loss(
        input_file=fcc_file,
        years=[2000, 2010, 2020],
        output_file="extent.png",
        title="Forest cover change 2000-2010-2020, TMF",
        dpi=100,
    )

.. image:: extent.png
    :width: 700
    :align: center