Skip to contents

Run Seasonal Analysis Engine

Source: R/analysis_engine.R
wapor_run_seasonal_analysis.Rd

Standalone function that performs the seasonal analysis workflow.

Usage

wapor_run_seasonal_analysis(
  config,
  crop_params,
  rasters,
  aoi_region = NULL,
  progress_callback = NULL
)

Arguments

config

List of configuration parameters. Besides the variable codes, period, data_source, folder and indicators, it accepts:

  • processing: "auto" (default), "memory", "stream" or "tiled". See wapor_plan_processing().

  • min_coverage: share of season days a pixel must have data for (default 1, every dekad). Pixels below it are NA rather than having missing dekads counted as zero.

  • keep_intermediates: keep dekadal_stacks and season_weights in the result. Default FALSE (they are large; indicator steps that need them still get them). The run logs an estimate of the disk space it writes and warns when the output or temporary folder has less free space (options(Rwapor.disk_check = FALSE) turns the check off).

  • output_dir: folder for stream and tiled outputs (default: a folder under tempdir()).

  • cache_dir: with data_source = "api", a folder in which the source rasters this run needs are saved first with wapor_download() (only the missing ones) and then read locally. Later runs for the same area read the saved files and work without internet. One folder per area; not available with l3_mode = "mosaic_all".

  • reference_layer, resampling_method: target grid and per-layer resampling (see Details).

crop_params

data.frame of crop class parameters (Kc, HI, etc.).

rasters

List of terra::SpatRaster objects (mask, start, end).

aoi_region

Optional region (bbox, vector path or sf). When NULL, the extent of rasters$crop_mask (or the season rasters) is used.

progress_callback

Optional function(value, detail) for updates.

Value

List of results (SpatRaster objects and summary tables), including coverage (per-variable share of season days with data) and processing (the wapor_plan_processing() plan that ran).

Details

Alignment uses config$reference_layer ("aeti" default, also "crop_mask", "ret", "pcp", "npp", "template").

Seasonal and monthly totals are summed at each source's native resolution and then resampled once onto the analysis grid. With the default nearest-neighbour resampling every output value is a raw server value or a sum of raw values. Set config$resampling_method (for example c(ret = "bilinear")) to interpolate instead. Crop mask and season rasters always use nearest neighbour.