Operations Concept for MintPy pipeline

Introduction

The MintPy pipeline is a portion of the overall Standard Product S1-GUNW Processing Pipeline that focuses solely on the use of S1-GUNWs and related products for the use of monitoring volcanic areas of interest.

High-Level Overview of pipeline

The basic steps involved in the MintPy volcanic anomaly detection pipeline are:

S1-GUNW within defined volcanic polygon is published to S3.
This triggers the MintPy PGE to run, which outputs the displacement time-series.
The time-series is run through the volcanic anomaly detection ML PGE, which outputs a volcano anomaly map.

Once an AOI is defined, the system will also perform forward keep-up production of S1-GUNW products - this AOI is larger than the volcano polygon, and will include the enumeration strategy. Information for how to perform bulk processing and keep-up production can be found on the Bulk Processing and Keep-Up Ops Overview wiki page.

Code/information used for each step:

S1-GUNW generation: Standard Product S1-GUNW Processing Pipeline
MintPy PGE: Displacement Time-Series using MintPy and Developer Information
Volcanic Anomaly Detection ML PGE: Volcanic Anomaly Detection

Use cases

Use Case 1: System continues forward “keep-up” production of S1-GUNW for volcano.

User adds an AOI track for the system to keep up processing on, and sets the type of AOI as “volcano”.
Note to Developer: AOIs should be input by developers, not users, at this time. Need to figure out how to combine ‘volcano’ type with ‘monitoring’ capabilities - that is, we want to do forward keep-up production of S1-GUNW products (i.e. what happens for a ‘monitoring’-type AOI) but we also want this specific type to trigger the MintPy time series PGE and following pipeline.
Science story: AOIs will encompass volcanic areas to be monitored for anomalies.
User defines desired N-pairings for S1-GUNWs.
System automatically pulls ancillary data for AOI (L1 SLCs, orbit data, calibration data, etc.).
System automatically processes from the L1 SLC to L2 S1-GUNW.
Note to Developer: System will perform automatic forward “keep-up” production of S1-GUNW for each defined AOI. The system will produce the standard product of the nearest 2-neighbors, and will also produce annual and seasonal pairs.
Science story: Annual and seasonal pairs will account for/work around low coherence data from winter seasons (due to snow, etc).

Use Case 2: System updates MintPy time-series for volcano.

User defines frequency of displacement time-series processing for defined AOI track (automatically upon new data acquisition, monthly, every N months, etc.) (do we want this capability to be in the initial build? Or for later development? Talk to Alex Dunn)
System updates previous time-series for defined AOI with new acquisition.
Note to developer: Science team has specified a preference for output of raw displacement time-series (i.e. without atmospheric correction), rather than filtered data, rolling means, or velocity maps.
User logs in to system to download time-series.
Note to Developer: Will need to add facet to allow for user search of displacement time-series.
Science story: Displacement time-series generation will allow scientists to quickly check on behavior and status of volcanoes of interest.

Use Case 3: Machine learning applied to output of MintPy for potential anomaly detection.

System applies ML to detect potential anomalies in most recently published displacement time-series, and publishes detected results back to GRQ catalog.
User logs into system to browse potential anomalies.
Science story: Anomaly detection will inform scientists of possible volcanoes to monitor closely.

User Interfaces (who is a user/operator? Developer already on our team or an outside person?)

User actions:

Defines polygon for volcano, including start time, end time, and separate, smaller bbox.
- If monitoring two or more volcanoes separated by a water body, add two separate AOI tracks with identical information except for different bboxes/polygons.
Creates a trigger rule for desired N-pairings.
Logs into system and downloads most recent time-series.
Logs into system and browses potential anomalies.

System actions:

Produces S1-GUNWs for defined N-pairings for AOI track.
Runs complete AOI track through MintPy PGE and calculates and publishes displacement time-series.
Runs most updated time-series through volcanic anomaly detection ML PGE, publishes volcanic anomaly data product.

User Guide

(for operator to run time series processing; to be filled in upon completion of MintPy PGE)