# Time Series Index AIKP

This AIKP is a concrete implementation of the `ocli.aikp.timeseries` base module. It calculates spectral indices (e.g., NDVI) over time and stores them in a Zarr store.

## Features
- **Inheritance**: Inherits state-machine and filter logic from the base module.
- **Spectral Indices**: Default implementation for NDVI using Sentinel-2.

## Configuration
In addition to base parameters:
- `index_kind`: Name of the index to calculate (`NDVI`, `NDWI`, `NDMI`).

## Usage
1. `ocli task template setup` - Fetch product list.
2. `ocli task template calculation` - Process products and build Zarr. Check the data `ocli task template calculation-info`.
3. `ocli task template publication` - Finalize task state. Check the data `ocli task template publication-info`.

## Publication
This AIKP uses custom strategies for metadata generation and upload:
1. `ocli ai makegeojson` - Generates a `TimeSeries` document metadata. During this step, the command:
   - Calculates the global minimum (`domain_min`) and global maximum (`domain_max`) across all time slices (excluding non-finite values) and includes them under `properties.zarr`.
   - Computes slice-by-slice statistics (`min`, `max`, `std`, `mean`, `median`) for each time slice and stores them as parallel arrays of float values inside `properties.zarr.stats`.
2. `ocli ai upload` - Uploads the Zarr store and metadata to Object Storage.
3. `ocli ai publish post` - Registers the metadata in the Document Database.