cropmaps package

Submodules

cropmaps.clipper module

class cropmaps.clipper.Clipper

Bases: object

static clipByMask(image, shapefile, store=None, band=None, new=None, resize=False, method=None, ext='tif', verbose=False, compress=False, force_update=False)

Mask image based on a shapefile mask.

Parameters:

  • image (senimage) – senimage object

  • shapefile (str) – Path to shapefile mask

  • store (str, optional) – Path to store data. If None stores the masked data inside the default satellite data structure. Defaults to None

  • band (str, optional) – Band to be applied. Note that the band name must be the same with the object attribute name. Defaults to None

  • new (str, optional) – Piece of string added to the end of the new filename. Defaults to None. If None then it names the new data with the shapefile name

  • resize (bool, optional) – If True resizes the output resize_val times. Defaults to False

  • method (rasterio.enum.Resampling, optional) – Resampling method. Defaults to None

  • ext (str, optional) – Extention of the image. Defaults to ‘tif’

Raises:

  • PathError – Raises when cannot find a path to store the new image

  • BandNotFound – Raises when the image has no attribute band selected by the user

cropmaps.cube module

cropmaps.cube.convert(x)
cropmaps.cube.generate_cube_paths(eodata: sentimeseries, bands: list, mask: str | None = None) → list

Get all the paths from all the available images in a timeseries.

Parameters:

  • eodata (sentimeseries) – Sentinel 2 data timeseries

  • bands (list) – List of the bands to extract data

  • mask (str, optional) – If provided then searches for attribute with this specific name. Defaults to None

Returns:

List of paths

Return type:

list

cropmaps.cube.make_cube(listOfPaths: List[str], searchPath: str, newFilename: str, dtype: dtype, nodata: float = -9999, gap_fill: bool = True, harmonize: bool = True, alpha: float = 0.0001, beta: float = 0.0, force_new: bool = False, compress=False) → Tuple[List[str], Dict]

Stack satellite images (FROM DIFFERENT FILES) as timeseries cube, without loading them in memory. If there is a datetime field in filename, could enable sort=True, to sort cube layers by date, ascending. Also, if sort=True, dates are written at .txt file which will be saved with the same output name, as cube.

Parameters:

  • listOfPaths (List[str]) – Paths of images which will participate in cube.

  • searchPath (str) – Where the result will be saved. Fullpath, ending to dir.

  • newFilename (str) – Not a full path. Only the filename, without format ending.

  • dtype (np.dtype) – Destination datatype.

  • nodata (float, optional) – No data value. Defaults to -9999

  • gap_fill (bool, optional) – Fill original nodata values where cloud exists and in feasible. Defaults to True

  • harmonize (bool, optional) – Apply DN to reflectance convertion. User need to provide alpha and beta from alpha * x + beta. Defaults to True

  • alpha (float, optional) – Alpha component of alpha in alpha * x + beta. Defaults to 0.0001

  • beta (float, optional) – Beta component of beta in alpha * x + beta. Defaults to 0

  • force_new (bool, optional) – Update cube with force if exists. Defaults to False

Returns:

Bands description in stacked order, metadata of written cube.

Return type:

Tuple[List[str], Dict]

cropmaps.exceptions module

exception cropmaps.exceptions.BandNotFound

Bases: imageError

Error raised when an attribute is not found.

exception cropmaps.exceptions.GeometryError

Bases: Exception

Base class for exceptions in this module.

exception cropmaps.exceptions.MinMaxCloudBoundError

Bases: timeseriesError

Error raised when the provided maximum cloud coverage value is less than the minimum cloud coverage value.

exception cropmaps.exceptions.MinMaxDateError

Bases: timeseriesError

Error raised when the provided maximum date value is less or equal than the minimum date value.

exception cropmaps.exceptions.NameNotInTimeSeriesError

Bases: timeseriesError

Error raised when the provided name is not in the time series object.

exception cropmaps.exceptions.NoDataError

Bases: timeseriesError

Error raised when no satellite data was found.

exception cropmaps.exceptions.NotSameGeometryError

Bases: GeometryError

Error raised when two geometry objects should be the same and their not.

exception cropmaps.exceptions.PathError

Bases: imageError

Raise when the path is not correct.

exception cropmaps.exceptions.VegetationIndexNotInList

Bases: imageError

Error raised when the provided Vegetation Index name is not in the list.

exception cropmaps.exceptions.imageError

Bases: Exception

Base class for exceptions in this module.

exception cropmaps.exceptions.timeseriesError

Bases: Exception

Base class for exceptions in this module.

cropmaps.get_creodias module

cropmaps.logger module

cropmaps.logger.setup(name=None, level=20, propag=False)

Logger configuration for the project.

cropmaps.models module

cropmaps.models.LandCover_Masking(landcover_mask: str, predicted: str, results_to: str, fname: str = 'Predictions_Crops.tif')

Masks classification results with WorldCover product to keep only areas with crops.

Parameters:

  • landcover_mask (str) – Path to land cover map (returned from sts.sentimeseries.LandCover)

  • predicted (str) – Path to predicted image

  • results_to (str) – Store the results

  • fname (str, optional) – New image name. Defaults to “Predictions_Crops.tif”.

Returns:

Path of the new image

Return type:

str

cropmaps.models.fill_confMatrix(ct: DataFrame, labels: List) → DataFrame

Fill a confusion matrix with data.

Parameters:

  • ct (pd.DataFrame) – Cross-tabulate object

  • labels (List) – Labels of the categories

Returns:

Filled confusion matrix

Return type:

pd.DataFrame

cropmaps.models.importance(band_desc: list, feature_importance: ndarray, store_to: str | None = None)

Get the importance of the data per band and per date.

Parameters:

  • band_desc (list) – List of the bands

  • feature_importance (np.ndarray) – Feature importance from sklearn model.features_importance_

  • store_to (str, optional) – Folder to save results. If None then the importance is printed as information from the logger. Defaults to None

cropmaps.models.load_model(fname: str) → RandomForestClassifier | SVC

Loads a saved model to memory.

Parameters:

fname (str) – Path to model path

Returns:

Restored model

Return type:

Union[RandomForestClassifier, SVC]

cropmaps.models.random_forest_predict(cube_path: str, model: RandomForestClassifier, results_to: str, fname: str = 'Predictions.tif') → str

Use to make predictions using a Random Forest model.

Parameters:

  • cube_path (str) – Path to datacube

  • model (RandomForestClassifier) – The model

  • results_to (str) – Path to save results

  • fname (str, optional) – Name of the output predicted image. Defaults to “Predicts.tif”

Returns:

Path of the predicted image

Return type:

str

cropmaps.models.random_forest_predict_patches(cube_path: str, model: RandomForestClassifier, results_to: str, patch_size: tuple = (512, 512), fname: str = 'Predictions.tif') → str

Use to make predictions using a RandomForestClassifier model with image patches.

Parameters:

  • cube_path (str) – Path to datacube

  • model (RandomForestClassifier) – The model

  • results_to (str) – Path to save results

  • patch_size (tuple, optional) – Size of the image patches. Defaults to (512, 512).

  • fname (str, optional) – Name of the output predicted image. Defaults to “Predictions.tif”

Returns:

Path of the predicted image

Return type:

str

cropmaps.models.random_forest_train(cube_path: str, gt_fpath: str, results_to: str, test_size: float = 0.33, n_jobs=-1, gridsearch: bool = False, parameters: dict = {'n_estimators': 200}) → RandomForestClassifier

Train a model with Random Forest classifier.

Parameters:
Returns:

The model

Return type:

RandomForestClassifier

cropmaps.models.save_model(model, spath: str, save_as: str = 'model.save') → str

Save model to disk.

Parameters:

  • model (sklearn.object) – Model to be saved

  • spath (str) – Path to save the model

  • save_as (str, optional) – Name of the file with .save extension. Defaults to “model.save”

Returns:

Path of the saved file

Return type:

str

cropmaps.models.svm_predict(cube_path: str, model: SVC, results_to: str, fname: str = 'Predictions.tif') → str

Use to make predictions using a SVM model.

Parameters:

  • cube_path (str) – Path to datacube

  • model (RandomForestClassifier) – The model

  • results_to (str) – Path to save results

  • fname (str, optional) – Name of the output predicted image. Defaults to “Predicts.tif”

Returns:

Path of the predicted image

Return type:

str

cropmaps.models.svm_predict_patches(cube_path: str, model: SVC, results_to: str, patch_size: tuple = (512, 512), fname: str = 'Predictions.tif') → str

Use to make predictions using a SVM model with image patches.

Parameters:

  • cube_path (str) – Path to datacube

  • model (svm.SVC) – The model

  • results_to (str) – Path to save results

  • patch_size (tuple, optional) – Size of the image patches. Defaults to (512, 512)

  • fname (str, optional) – Name of the output predicted image. Defaults to “Predictions.tif”

Returns:

Path of the predicted image

Return type:

str

cropmaps.models.svm_train(cube_path: str, gt_fpath: str, results_to: str, test_size: float = 0.33, gridsearch: bool = False, parameters: dict = {'C': 1.0}) → str

Train a model with Random Forest classifier.

Parameters:
Returns:

The model

Return type:

svm.SVC

cropmaps.prepare_vector module

cropmaps.prepare_vector.buffer()
cropmaps.prepare_vector.burn(shapefile: str, classes: str, metadata: dict, classes_id: str | None = None, save_nomenclature: bool = True, save_to: str | None = None, outfname: str = 'gt.tif') → str

Burns vector ground truth data to raster.

Parameters:

  • shapefile (str) – Path to shapefile ground truth data

  • classes (str) – Column name with the categories

  • metadata (dict) – A dictionary with width, height, crs, transform (Can retrieved from a base image with rasterio.open().meta)

  • classes_id (str, optional) – Column name with ID for each class. If None then this function creates an ID for each unique category. Defaults to None

  • save_nomenclature (bool, optional) – Writes the nomenclature data. Defaults to True

  • save_to (str, optional) – Save path of the results. If None then the data will be written at the same path as the shapefile. Defaults to None

  • outfname (str, optional) – Name of the ground truth image. Defaults to “gt.tif”

Returns:

Path to ground truth image data

Return type:

str

cropmaps.sentinels module

cropmaps.sentinels.convert(x)
class cropmaps.sentinels.sentinel2(path, name)

Bases: object

A Sentinel 2 image.

apply_cloud_mask(band: str | None = None, store: str | None = None, subregion: str | None = None, resolution: str | None = None, new: str = 'CLOUDMASK', compress=False) → None

Apply default SCL mask to S2 images.

Parameters:

  • band (str, optional) – Band to apply SCL mask. If None then applies mask to all default bands. Defaults to None.

  • store (str, optional) – Path to store the new images. If None then the results are saved in the default S2 path. Defaults to None.

  • subregion (str, optional) – Apply mask to subregion. If subregion is None then applies the mask to the default raw images. Defaults to None.

  • resolution (str, optional) – Resolution to apply cloud mask. If None the applies to default resolution else tries to apply to the highest 10m resolution. Defaults to None.

  • new (str, optional) – New name extension. Defaults to “CLOUDMASK”.

calcVI(index, store=None, subregion=None, verbose: bool = False, compress=False)

Calculates a selected vegetation index (NDVI, NDBI, NDWI). :param index: Vegetation index to be calculated and saved. Currently only NDVI, NDBI, NDWI are supported :type index: str

getBands()

Finds all the available bands of an image and sets new attributes for each band.

getmetadata()

Searching for metadata (XML) files.

static reproj_match(image: str, base: str, to_file: bool = False, outfile: str = 'output.tif', resampling: Resampling = Resampling.nearest, compress=False) → None

Reprojects/Resamples an image to a base image. :param image: Path to input file to reproject/resample :type image: str :param base: Path to raster with desired shape and projection :type base: str :param outfile: Path to saving Geotiff :type outfile: str

static setResolution(band)

Getting band resolution for Sentinel 2. :param band: Band short name as string :type band: str

Returns:

Band resolution

Return type:

str

property show_metadata

Prints metadata using __dict__

upsample(band=None, store=None, new=None, subregion=None, method=None, ext='tif')

Upsample 20 meters spatial resolution bands to 10 meters.

Parameters:

  • store (str, optional) – Path to store data. If None then stores the results to default image path. Defaults to None

  • band (str, optional) – Band to apply upsample. If None then raises error. Defaults to None

  • new (str, optional) – Name extension. If None then adds new resolution and _Upsampled. Defaults to None

  • subregion (str, optional) – Perform upsample to subregion. Defaults to None.

  • method (Resampling, optional) – Resampling method. If None then applies Resampling.nearest. Defaults to None

  • ext (str, optional) – Image extension. Defaults to ‘tif’.

static writeResults(path: str, name: str, array: array, metadata: dict)

Writing a new image with the use of rasterio module. :param path: Path to image :type path: str :param name: Image name :type name: str :param array: Image numpy array :type array: np.ndarray :param metadata: Metadata dictionary :type metadata: dict

cropmaps.sts module

class cropmaps.sts.sentimeseries(name: str)

Bases: timeseries

Sentinel 2 time series.

LandCover(store, aoi: str | None = None, outname='LC_mosaic.tif') → str

Download and add WorldCover masking data to the object.

Parameters:

  • aoi (str) – Path to AOI

  • write (bool, optional) – Write result to disk. Defaults to False

Returns:

Path with LC data

Return type:

str

apply_SCL(image: sentinel2 | None = None, band: str | None = None, store: str | None = None, subregion: str | None = None, resolution: str | None = None)

Apply default SCL mask to S2 images.

Parameters:

  • image (sentinel2, optional) – Apply SCL mask to one image. If None then applies the mask to all the timeseries. Defaults to None.

  • band (str, optional) – Band to apply SCL mask. If None then applies mask to all default bands. Defaults to None.

  • store (str, optional) – Path to store the new images. If None then the results are saved in the default S2 path. Defaults to None.

  • subregion (str, optional) – Apply mask to subregion. If subregion is None then applies the mask to the default raw images. Defaults to None.

  • resolution (str, optional) – Resolution to apply cloud mask. If None the applies to default resolution else tries to apply to the highest 10m resolution. Defaults to None.

clipbyMask(shapefile, image=None, band=None, resize=False, method=None, new=None, store=None, force_update=False)

Masks an image or the complete time series with a shapefile.

Parameters:

  • shapefile (path-like, str) – Path to shapefile mask

  • image (senimage, optional) – Masks a specific image. Defaults to None

  • band (str, optional) – Masks a specific band. Defaults to None

  • resize (bool, optional) – Resize band. Defaults to False

  • method (rasterio.enums.Resampling) – Available resampling methods. If None the Nearest is used. Defaults to None

find(path: str, level: str = 'L2A')

Finds automatically all the available data in a provided path based on the S2 level product (L1C or L2A) that the user provides.

Parameters:

  • path (str, path-like) – Search path

  • level (str, optional) – Level of the S2 time series (L1C or L2A). Defaults to ‘L2A’.

Raises:

NoDataError – Raises when no data were found in the provided path

find_DIAS(DIAS_data: list)
getVI(index: str, store: str | None = None, image: sentinel2 | None = None, subregion=None, verbose=False)

Calculates a vegetation index for an image if the user provides an image or for all the time series.

Parameters:

  • index (str) – Vegetation index. Currently works only for NDVI, NDWI, NDBI

  • image (sentinel2) – If an sentinel2 object is provided calculates VI for this image only

remove_orbit(orbit)

Remove images with specific orbit.

Parameters:

orbit (str) – Number of orbit

upsample(image=None, band=None, method=None, new=None, store=None, subregion=None)

Upsample lower resolution bands to the highest available resolution.

Parameters:

  • image (sentinel2, optional) – Image to perform upsampling. If None, then performs upsample to all images in the timeseries. Defaults to None

  • band (str, optional) – Name of the band to upsample. If None, then performs upsample to all 20m bands. Defaults to None

  • method (Resampling, optional) – Upsampling method. Defaults to None

  • new (str, optional) – Name extension. If None the by default adds _Upsampled. Defaults to None

  • store (str, optional) – Path to store data. If None then stores the data to default image path. Defaults to None

  • subregion (str, optional) – Perform upsample to subregion. Defaults to None

cropmaps.ts module

class cropmaps.ts.timeseries(name)

Bases: object

Base class of time series objects.

filter_dates(max_date, min_date=None)

Removes from the list of data all the instances with date more than max_date and less than min_date. Date format: DDMMYYYY

Parameters:

  • max_date (str) – Maximum date in DDMMYYY format

  • min_date (str, optional) – Minimum date in DDMMYYY format. Defaults to None

Raises:

  • TypeError – Raises if max_date is not a str

  • ValueError – Raises if max_date is not 8 length str

  • TypeError – Raises if min_date is not a str

  • ValueError – Raises if min_date is not 8 length str

  • MinMaxDateError – Raises if maximum date is sooner than minimum date

keep_timerange(start_time, end_time)

Keeps all the data that are inside the range of a start_time (HHMMSS format) and an end time (HHMMSS format).

Parameters:

  • start_time (str) – Start time in HHMMSS format

  • end_time (str) – End time in HHMMSS format

remove_cloudy(max_cloud, min_cloud=0)

Removes from the list of data all the images with cloud coverage that overcomes the maximum and minimum bounds provided by the user.

Parameters:

  • max_cloud (int) – Maximum allowed cloud coverage

  • min_cloud (int, optional) – Mimimum allowed cloud coverage. Defaults to 0

Raises:

MinMaxCloudBoundError – Raises when maximum cloud coverage is less than the minimum coverage

remove_date(date)

Removes from the list of data all the instances with a specific date(s) provided by the user. Date format: DDMMYYYY

Parameters:

date (str, list) – Date(s) to remove in DDMMYYYY format

Raises:

  • TypeError – Raises if date is not a str or a list

  • ValueError – Raises if date value is not a 8 length str

show_metadata(name=None)

Prints all the metadata of all available data of the time series or of a selected image.

Parameters:

name (str, optional) – Name of the image to print its metadata. Defaults to None

Raises:

NameNotInTimeSeriesError – Raises when the provided name is not in the list of data

sort_images(cloud_coverage=False, date=False)

Sort images based on name. If cloud coverage is True then sorts based on cloud coverage. If date is True then sorts based on dates.

Parameters:

  • cloud_coverage (bool, optional) – Sort by cloud coverage. Defaults to False

  • date (bool, optional) – Sort by dates. Defaults to False

cropmaps.utils module

cropmaps.utils.worldcover(geometry: Polygon, savepath: str) → GeoDataFrame

Downloads landcover maps from worldcover project :param geometry: Path to AOI file to download data or a list with geometries :type geometry: shapely.geometry.polygon.Polygon :param savepath: Path to store data :type savepath: str

Returns:

Downloaded tiles

Return type:

gpd.GeoDataFrame

cropmaps.vi module

class cropmaps.vi.vi

Bases: object

A collection of functions for Vegetation Incices in Python. Created for use with Sentinel 2. Sources: http://www.sentinel-hub.com/eotaxonomy/indices

static ndbi(swir, nir)

Normalized Difference Buildings Index

General formula:
\[(SWIR — NIR)/(SWIR + NIR)\]
Sentinel 2:
\[(B11 - B08) / (B11 + B08)\]
Parameters:

  • swir (ndarray) – SWIR numpy array

  • nir (ndarray) – NIR numpy array

Returns:

NDBI numpy array

Return type:

ndarray

static ndvi(red, nir)

Normalized Difference Vegetation Index General formula:

\[(NIR — VIS)/(NIR + VIS)\]
Sentinel 2:
\[(B08 - B04) / (B08 + B04)\]
Parameters:

  • red (ndarray) – Red numpy array

  • nir (ndarray) – NIR numpy array

Returns:

NDVI numpy array

Return type:

ndarray

static ndwi(green, nir)

Normalized Difference Water Index

General formula:
\[(GREEN — NIR)/(GREEN + NIR)\]
Sentinel 2:
\[(B03 - B08) / (B03 + B08)\]
Parameters:

  • green (ndarray) – Green numpy array

  • nir (ndarray) – NIR numpy array

Returns:

NDWI numpy array

Return type:

ndarray

Module contents