-
For other nowcasting applications, such as precipitation, the training data is usually sampled so that rainfall exists or is above some threshold in every, or nearly every input sample. This ensures t…
-
-
## Detailed Description
https://arxiv.org/abs/2112.13776
This paper tries to get uncertainty estimation by using stochastic attention, so a slightly different way than the skillful nowcasting …
-
## Detailed Description
Issue #30 documents an issue in non-HRV Zarr data whereby there's a 'rectangle of zeros'. We should re-compute the means and standard deviations, ignoring northern latitudes …
-
## Detailed Description
Maybe we should _always_ validated the on-disk batches?
(Let's wait for PR #300 to be merged before working on this)
-
I believe that in line 243 of the dfm.m file the EMstep function should return V_0_new and not V_0. Due to this typo, the initial state covariance matrix is not updated in the algorithm. It does not s…
-
-
For consistency, all calls to forecast endpoints should ideally return the `forecastCreationTime` and also `gspID` for each `targetTime`.
This will help with interpreting the data and also with parsi…
-
## Detailed Description
Would be useful to have a method to drop 'padded' pv systems. These are padding out zero so that the dataset can be save in an efficient way. However for plotting its useful…
-
## Detailed Description
We want a model that can predict both satellite imagery and pv yield.
## Context
While we ultimately just care about the PV output of the model, predicting where the …