calc_cloudcov calculates the aoi cloud cover and optionally saves raster cloud masks, all based on preview images. The previews are requested through get_previews. You may call get_previews before calc_cloudcov. In this case the previews will be reloaded. If one or more records have been processed in calc_cloudcov and in the same dir_out before they will be reloaded.

calc_cloudcov(
  records,
  max_deviation = 2,
  aoi = NULL,
  write_records = TRUE,
  write_cloud_masks = TRUE,
  dir_out = NULL,
  username = NULL,
  password = NULL,
  as_sf = TRUE,
  verbose = TRUE,
  ...
)

Arguments

records

data.frame, one or multiple records (each represented by one row), as it is returned by get_records.

max_deviation

numeric, the maximum allowed deviation of calculated scene cloud cover from the provided scene cloud cover. Use 100 if you do not like to consider the cloud cover % given by the data distributor. Default is 2.

aoi

sfc_POLYGON or SpatialPolygons or matrix, representing a single multi-point (at least three points) polygon of your area-of-interest (AOI). If it is a matrix, it has to have two columns (longitude and latitude) and at least three rows (each row representing one #corner coordinate). If its projection is not +proj=longlat +datum=WGS84 +no_defs, it is reprojected to the latter. Use set_aoi instead to once define an AOI globally for all queries within the running session. If aoi is undefined, the AOI that has been set using set_aoi is used.

write_records

logical specifies if the records (row by row) shall be written.

write_cloud_masks

logical specifies if the cloud mask tifs shall be written.

dir_out

character. If dir_out is not NULL the given cloud mask rasters and a record file for each record will be saved in dir_out. If it is NULL, the session dir_out is used. If no session dir_out is set through set_archive an error is thrown.

username

character, a valid user name to the ESA Copernicus Open Access Hub. If NULL (default), the session-wide login credentials are used (see login_CopHub for details on registration).

password

character, the password to the specified user account. If NULL (default) and no seesion-wide password is defined, it is asked interactively ((see login_CopHub for details on registration).

as_sf

logical, whether records should be returned as sf data.frame or a simple data.frame. In both cases, spatial geometries are stored in column footprint.

verbose

logical, if TRUE, details on the function's progress will be visibile on the console. Default is TRUE.

...

further arguments that can be passed to write_records for writing record files. Can be: driver, append.

Value

records data.frame with three added columns:

  1. cloud_mask_file: character path to the cloud mask file of the record

  2. aoi_HOT_cloudcov_percent: numeric percentage of the calculated aoi cloud cover.

  3. scene_HOT_cloudcov_percent: numeric percentage of the calculated scene cloud cover.

Details

Using the Haze-optimal transformation (HOT), the cloud cover estimation is done on the red and blue information of the input RGB. HOT procedure is applied to the red and blue bands [1-3]. Originally, the base computation was introduced by Zhang et al. (2002) [2]. The computation done in calc_cloudcov includes the following steps:

  1. Binning: extract low red values and their highest blue values

  2. Regression: calculate linear regression of these values

  3. HOT layer: compute haze-optimal transformation cloud likelihood layer

  4. Iterative thresholding: Find a HOT threshold by iterative comparison with the provider scene cloud cover.

  5. Aoi cloud cover calculation: Calculate the aoi cloud cover from the binary cloud mask.

HOT separates clear-sky pixels first from a threshold, calculates a linear regression from these pixels and exposes cloud pixels by the deviation of all pixels from this clear-sky line.

References

[1] Chen, S, Chen, X., Chen, J., Jia, P., 2015. An Iterative Haze Optimized Transformation for Automatic Cloud/Haze Detection of Landsat Imagery. IEEE Transactions on Geoscience and Remote Sensing 54 (5), 2682-2694.

[2] Zhang, Y., Guindon, B., Cihlar, J., 2002. An image transform to characterize and compensate for spatial variations in thin cloud contamination of Landsat images. Remote Sensing of Environment 82 (2-3), 173-187.

[3] Zhu, X., Helmer, E.H., 2018. An automatic method for screening clouds and cloud shadows in opticalsatellite image time series in cloudy regions. Remote Sensing of Environment 214 (2018), 135-153.