Multi output gaussian process regression

Sentinel-1 and Sentinel-2 data fusion through Multi-output Gaussian process regression (MOGPR)

This service is designed to enable multi-output regression analysis using Gaussian Process Regression (GPR) on geospatial data. It provides a powerful tool for understanding and predicting spatiotemporal phenomena by filling gaps based on other correlated indicators. This service focuses on fusing Sentinel-1 and Sentinel-2 data, allowing the user to select one of the predefined data sources.

Parameters

The fusets_mogpr_s1s2 service requires the following parameters:

Name	Description	Type	Default
spatial_extent	Polygon representing the AOI on which to apply the data fusion	GeoJSON
temporal_extent	Date range for which to apply the data fusion	Array
s1_collection	S1 data collection to use for the fusion	Text	RVI
s2_collection	S2 data collection to use for fusing the data	Text	NDVI

Supported collections

Sentinel-1

• RVI
• GRD

Sentinel-2

• NDVI
• FAPAR
• LAI
• FCOVER
• EVI
• CCC
• CWC

Limitations

The spatial extent is limited to a maximum size equal to a Sentinel-2 MGRS tile (100 km x 100 km).

Dependencies

In addition to various Python libraries, the workflow utilizes the following libraries included in the User-Defined Function (UDF):

• Biopar: The biopar package retrieves biophysical parameters like FAPAR, FCOVER, and more, that were passed as the S2_collection. The biopar package is a Python package that calculates biophysical parameters from Sentinel-2 satellite images as described here. The fusets_mogpr udp directly uses the biopar udp shared in the APEX Algorithms repository.

• FuseTS: The fusets library was developed to facilitate data fusion and time-series analytics using AI/ML to extract insights about land environments. It functions as a Time Series & Data Fusion toolbox integrated with openEO. For additional information, please refer to the FuseTS documentation.

Output

This User-Defined-Process (UDP) produces a datacube that contains a gap-filled time series for all pixels within the specified temporal and spatial range. This datacube can be seamlessly integrated with other openEO processes.