ps: persistent scatterers identification

moraine.cli.ps 的命令（源码）；同样的函数在 Python 里以 mc.<name>(...) 调用（import moraine.cli as mc）。下面的帮助由 docstring 生成，与 moraine COMMAND --help 相同（docstring 为英文）。

命令	摘要
amp-disp	calculation the amplitude dispersion index from SLC stack.

amp-disp

calculation the amplitude dispersion index from SLC stack.

moraine amp-disp --rslc STR --adi STR [--chunks INT INT] [--out_chunks INT INT] [--cuda | --no-cuda] [--processes VALUE] [--n_workers VALUE] [--threads_per_worker VALUE]

参数	类型	默认值	说明
--rslc	str	必需	input rslc stack, shape (nlines, width, nimages)
--adi	str	必需	output amplitude dispersion index, shape (nlines, width), float32
--chunks	tuple[int, int]	None	data processing chunk size, same as rslc by default
--out_chunks	tuple[int, int]	None	output data chunk size, same as chunks by default
--cuda	bool	False	if use cuda for processing, false by default
--processes		None	use processes (True) or threads (False) for the workers, only for cpu processing. Default: False
--n_workers		None	number of workers. Default: 1 for cpu, one per GPU for cuda
--threads_per_worker		None	tasks a worker runs at the same time. Default: 1

全局选项：--json、--traceback、--log FILE、-q / --quiet（约定）。

在 pipeline 文件中（examples/02_ps.toml，步骤 adi）：

[[step]]
name = "adi"
run = "amp-disp"
rslc = "raw/rslc.zarr"
adi = "ps/ras_adi.zarr"
cuda = false                      # reads the stack once and reduces it: the CPU is faster than starting a GPU worker

在 Python 中：

import moraine.cli as mc
mc.amp_disp(rslc, adi, chunks=None, out_chunks=None, cuda=False, processes=None, n_workers=None, threads_per_worker=None)

数组函数 mr.amp_disp 对内存中的 numpy / cupy 数组做同样的事。