=========================================
INGV 5-minute forecast of Sym-H index
=========================================

.. contents:: The tiny table of contents

SYM-H is forecast using two different artificial neural network (ANN) models. One is based on the long short-term memory (LSTM) and the other on the convolutional neural network (CNN). Both requires as input 5-minute averages of IMF Bz, By^2 and B^2. The former model needs also SymH values. Prediction is performed 1-hour in advance. The product consists in a Python software to run the models and two sets of coefficients (two different files containing the two networks configurations).  

**Providers**:

- **INGV** - source:

  - Magnetometer


| **Product Type**: software

| **Status**: Active

| **Creation date**: 2022-04-29T12:00:00+02:00

| **Last Update**: 2023-10-01T12:00:00+02:00

Data Description
~~~~~~~~~~~~~~~~

| **Data type**: timeseries

| **Data formats**: - json


| **Sample**: 8_AQU_sample.json

Capabilities
~~~~~~~~~~~~

- int2130.1: Product ingestion
- int2130.4: Scheduled computation
- int2320.1: Data conversion
- ext2220.1: Standard Visualisation: Web GUI
- ext2220.2: Standard Query: Web GUI
- ext2220.3: Download: Web GUI
- ext2220.4: NRT Visualisation: Web GUI
- ext2230.1: Advanced Visualisation: ASPIS.py
- ext2230.2: Advanced Query: ASPIS.py
- ext2230.3: Download: ASPIS.py

