Implement Reduced-Space Ensemble Kalman Filter Approach for Blending

[RubenImhoff]

Next to the already implemented STEPS blending approach (Bowler et al., 2006; Seed et al., 2013 and our paper Imhoff et al., 2023), we thought it was getting time to introduce another blending method based on Daniele's Monthly Weather Review paper from 2019. This blending approach uses a Bayesian blending approach, by using a reduced-space ensemble Kalman filter that is flow-dependent to combine extrapolation-based nowcasts and NWP.

@m-rempel (and colleagues) has continued with this approach in his research to make it ready for operations. This seems like an excellent time to include this approach in pysteps, I've discussed with @m-rempel to get started with this in the coming months. @m-rempel will make a new branch in which we can start working. We can use this issue (among others) for discussion about the approach.

[ladc]

Hi Ruben, Yes indeed, and it is also a good moment to start thinking about issues to tackle or finish during the pysteps Hackathon 😊 @sidekock suggested to create a project board already, which I think is a good idea.