mapDisturbanceL8S1: Map forest disturbance using Landsat 8 and Sentinel-1
In dulvrq/ddl8s1ts: Map forest disturbance using Landsat 8 and Sentinel-1

Description Usage Arguments Details Value Note

Detect and map forest disturbance using Landsat 8 and Sentinel-1 time series.

mapDisturbanceL8S1(
  ls_l8,
  ls_s1,
  l8_doys,
  s1_doys,
  dt_ref,
  ls_dem = NULL,
  dir_save,
  VI,
  rf_model = NULL,
  startDOY,
  endDOY,
  mmu = NULL,
  only_rf = F,
  max_cores = 20,
  threshold = 0.5
)

`ls_l8`	a file list of Landsat 8 with 6 bands. All files should have the same extent.
`ls_s1`	a file list of Sentinel-1 with 2 bands (VV and VH, in this order). If this is used with Landsat, all files should have the same extent as Landsat (i.e. same origin, crs and resolution)
`l8_doys`	a list of julian day of Landsat 8 (numeric). This should be the same order and length as `ls_l8`.
`s1_doys`	a list of julian day of Sentinel-1 (numeric). This should be the same order and length as `ls_s1`.
`dt_ref`	a dataframe of reference data for RF. This should contain column 'x', 'y', and 'date'. x and y should indicate locations in the same crs as Landsat or Sentinel. 'date' should indicate the timing of disturbance. If there is no disturbance at that location, use NA.
`ls_dem`	a file name of DEM with 2 bands (elevation and slope in this order). This is used for RF model. If this is not to be used for model, set NULL.
`dir_save`	a directory for save the results.
`VI`	names of spectral index used for Landsat.
`rf_model`	a list of two RF models (as list) to skip RF model. Set NULL to build RF model.
`startDOY`	a julian day of the initial timing of disturbance detection (e.g. 2016-02-01 should be 2016.08767)
`endDOY`	a julian day of the terminate day of disturbance detection.
`mmu`	numeric. minimum mapping unit (pixel) for final maps. New tif image will be additionally generated.
`only_rf`	logical. If TRUE, only build RF models. If FALSE, build RF models and map disturbance detection.
`max_cores`	numeric. Maximum numbers of cores used for parallel processing.
`threshold`	numeric from 0 to 1. A threshold to detect disturbance from time series disturbance probabilities.

This function can be mainly used for three ways. 1. build RF models and then detect & map disturbance. 2. build RF models (no mapping nor detection of disturbance) 3. detect & map disturbance (mandatory with RF models)

The use of both Landsat 8 and Sentinel-1 is full implementation of this algorithm. However, you can use either Landsat 8 or Sentinle-1 if you want. The process takes very long time. This use parallel process to save time.

a filename of mapped disturbance.

Only 50% of sample ID is used for RF modeling.Subfolders are automatically genereted for saving RF models and mapping temporally tiles of a disturbance map.

dulvrq/ddl8s1ts documentation built on May 2, 2021, 5:18 a.m.