Development Goals

The current release of KrigR is stable, but it is still undergoing development. Any and all feedback you can give us is more than welcome.

I am currently in the progress of closing out my PhD Project and am thus unable to develop KrigR right now. If you, or someone you know, would happen to be interested in hiring me to do so after my PhD, I’d be more than happy for that opportunity.

Data Objects to be included in KrigR:

In the future, we hope to ship KrigR with a few in-built data sets for easier demoing of the package:

• ERA5-Land data
• GMTED 2010 data

To improve user-friendliness of the download_ERA() experience, we plan to ship KrigR with metadata files for each data set containing:

• Name of dataset
• Dataset spatial extent
• Time-window of dataset
• Subsets of dataset (e.g. monthly averaged reanalysis, vs. monthly averaged reanalysis by hour of the day)
• Temporal resolution of dataset and subsets
• Variables contained in dataset
• Variable-dependant issues (i.e. missing values for starting intervals of time series, cumulative records)
• Obligatory and optional arguments for API call
• Citation key
• Link to data set

These metadata files will be used for sanity checking of download request before execution, automatic fixes of first-hour issues and the like, easier request building and piping through ecmwfr behind the scenes, and will be possible to query to give users an overview of datasets.

Function Changes

KrigR works as intended, but there are some changes to the core functions of KrigR which we have set our eyes on in an effort to improve efficiency and user-friendliness:

General Function Changes

• Try terra functions instead of raster functions for a potential increase in computational speed
• Check if tidyverse is loaded and multiple cores are specified. Unload tidyverse packages at the beginning of KrigR functions and reload them when code exits with on.exit()
• Enable saving of raster in non-NETCDF formats

Download Function

• Support additional ECMWF products (e.g.: CMIP6 climate projections)
• Download by hours of date, too
• Export NETCDFS with layer names according to desired time-window and temporal resolution
• Make temporal aggregation dependant on dataset metadata file
• Ensemble spread download for data sets which offer ensembles for uncertainty metrics
• Optional argument to return extracted values for points in data frame with arguments passed on to raster::extract() function
• Temporal aggregation time windows and entire time series match check before downloading
• Investigate ways to reduce the two month data download for a single month of cumulative records
• Toggle SingularDL on intelligently depending on download specification by default
• Multi-core temporal aggregation
• Make it recognise tibbles or spatialpoints for points
• Query to CDS download queue?
• Cumulative back calculation function
• Check for cumulatively stored variables and cumulative back-calculation

Covariate Function

• Include query to HWSD and Soil Property covariates
• Aggregation and preparation of user-provided covariates (provided at target resolution)

Kriging Function

• Kriging recommendation table
• Covariates that vary over time
• Duration timer into call list

Shiny / Web-Client Application

To further reduce barrier of entry to the use of KrigR, we intend to, in the future, ship KrigR with a shiny app with which to create download scripts from a more visual interface.

• Browser-based creation of download script for KrigR
• Time-line with coloured sections according to chosen time-window and temporal resolution
• Map for Visualisation of dataset extent
• Map for Extent selection
• Map for Polygon shapefile creation
• Map for Point selection by clicking or typing coordinates out