List view
The package is supposed to support the calculation of a variety of indicators for biodiversity monitoring. The following requirements should be fulfilled: - an indicator function receives a single polygon as its first argument - the next arguments are the required resources for that indicator, it should expect raster as a spatRaster and vectors as sf object - additional arguments are allowed, but where possible should be retrieved from the attributes of a portfolio in a generic indicator calculation function - the output of the indicator function should be a tibble, where possible with a single row (time is an example for multiple rows) - the column names should be unique between indicators - an indicator function should not take care of parallel processing, the best strategy for parallel processing will be determined by the generic indicator function - the name of the indicator and their required inputs are backlogged in a package internal function - an indicator should expect an "engine" argument, from that argument it should be derived which implementation for the calculation to use, e.g. via terra, exactextract etc.
Overdue by 4 year(s)•Due by February 28, 2022•1/1 issues closedResources that are needed for indicator calculations should be automatically downloaded on the portfolio level. The following requirements should hold for any download function: - a download function receives the bounding box of the portfolio as its first argument - additional arguments are allowed, but whenever possible they should be derived from the portfolio objects attributes (e.g. the timeframe of the analysis) - the name of the resources, its format (raster/vector) and additional arguments should be backlogged in a package internal function - a download function should check the values of additional arguments for correctness and give informing error messages if they are wrong - a download function receives an argument 'rundir' where any files written to disk shold be saved to - for global datasets (raster, vector), a download function crops the downloaded global data to bbox and removes the global dataset to free-up space - a download function returns a character vector to all output files located in rundir relevant for the resource
Overdue by 4 year(s)•Due by February 18, 2022•2/2 issues closed**S3 implementation has been put on hold** The basic framework consists of an S3 portfolio-class build on-top of the sf class. It has the following characteristics: - tidy tabular format (just like sf): one row is an observation, one column represents a variable - arbitrary metadata from the user-side is allowed and preserved - during object initiation, attributes describing the timeframe of the portfolio and locally available resources are added - get_resource functions take a portfolio object as input and download a single resource, the local path to the resource is added to the portfolio - get_indicator functions take a portfolio object, check if the required resources are available and automatically defines an efficient parallel computation strategy based on the composition of the portfolio - due to different sizes (i.e. number of rows/columns) between indicators, the results are added to the portfolio object as nested columns, users should be able to unnest single indicators resulting in an sf object for further analysis - I/O of portfolios should include writing a portfolio object to disk as a Geopackage with indicators saved into different tables. Reading such an object back into the R session should return a nested portfolio object.
Overdue by 4 year(s)•Due by February 18, 2022