open-eo / fusets Goto Github PK

Due: 30/09/2023

Basic whitakker functionality is available in fusets and can be documented with sample notebooks

Due: 30/09/2022

Due: 30/03/2023

From SoW REQ 5: Define User Requirements:

For the User Requirements a set of 3-5 use cases shall be defined. The selection of use cases shall be guided and warranted by the respective potential for improved information retrieval when employing the here developed data fusion and time series analytics methods.

At least one of these use cases shall be specifically focussing on agriculture, addressing challenging target variables (e.g. crop yield, land management events, crop damage, etc.). Other major land domains shall be included such as forest, urban, grassland or water.

Each use case shall define an end to end processing workflow starting with EO and ancillary source data, data fusion into fused data stream(s), time series modelling, land surface characterisation and future state predictions. Each use case shall be developed in coordination with a domain expert (e.g. for agriculture, forestry, etc.).

Develop and compile the User Requirements Document to guide the further development.

Due: 30/12/2022

Now functions are directly part of jupyter notebooks.

Create better (but simple) module structure, following python guidelines, with one or two algorithms that are now in notebooks.
Move notebooks to separate folder (don't have to be part of the eventual wheel).
Communicate to project colleagues, or write project guidelines.

Library also has to be usable without openEO!

• Add setuptools files
• Github actions for building/testing/packaging
• Publish into artifactory for dev builds, pypi for prod?

As a first PoC, I'd like to have an XArray based signature for whittaker next to the current numpy version.
Something like:

def whittaker(array:DataArray, smoothing_lambda, days,time_dimension="t"):

This should be the basis for use in a UDF, so array can have dimensions time, x, y, bands, where bands, x, y are optional.

Due: 30/06/2022

Due: 30/06/2023

Under #44 I already set up the basic package files (setup.cfg, setup.py, Jenkinsfile, .github/workflows) for CI and (dev snapshot) packaging.

Once the package is more ready for general usage, we should also publish "official" releases to pypi

I'm working through whittaker as a first example of a time series smoothing function. I would also like to add some initial documentation, to make it already like a real library. This documentation should describe things for a user who does not know what whittaker is, without being overly long.

Some things to cover:
What does it do? => Smoothing, gap filling?
Performance? => it's relatively fast
What does it return? (Timeseries sampled at equidistant steps?)
Explain lambda parameter

Inventory of within-consortium available TS processing and fusion algorithms and collection of public additional algorithms

• subtask A
• subtask B
• subtask C

Provide basic structure within the repo for documenting the library (e.g. using markdown/Sphinx)
If possible, use github actions to publish as github pages.

Website can also use static generator mechanism, to be investigated.

Due: 30/03/2022

It is time to put together some concrete survey which can be provided to the stakeholders. This survey will serve as a basis for sending it out to a wider community on LPS22 in May. This was decided in the Progress Meeting no. 1, meaning that the results of the survey for the wider community (and potentially the updated survey) can be sent to ESA later on as an addendum.

If I understand correctly, Google Forms are most widely used for such occasions, but I remember some survey-like functionality being available also in MS Teams? Can someone more knowledgeable about the topic please kindly share their experiences?

Based on the conversations with experienced project managers it was suggested to structure the survey in the following manner:

• a generic Earth Observation / Remote Sensing part
• a use case specific part (showcased use cases from the AI4FOOD project are used here)

Questions in the second part focus in depth on the use case specific aspects. Questions can be repeated across sections, if they touch common dilemmas, since (I imagine) some survey partakers might skip some sections entirely.

TODO:

• the first objective is to discuss regarding the survey platform
• the second objective is to gather generic and use-case specific questions
• the third objective is to organize the questions and text into a proper survey

Feedback or suggestions are appreciated.