Migrate from raster to terra

[ramarty]

No description provided.

[alexyshr]

If you will use terra::extract to update the function bm_extract. You can use points, points with radius (buffer), polygons, or polygons with radius (buffer) to extract the values. In addition, setting the parameters exact and weights to TRUE helps create a smoother time series (those parameters can be part of bm_extract).

   ...
   ...
   buffer <- 0.1

   #the geometry to extract time series (SF of points or polygons)
   the_geometry 

   #this variable is created from HDF5 files using terra::
   ntl_raster_terra
   ... 
   ...


  #check if the buffer is different from 0
  if (buffer != 0) {
    #before to use st_buffer we need set sf_use_s2 to FALSE
    temp <- sf_use_s2()
    sf_use_s2(FALSE)
    #make a buffer from initial geometry
    the_geometry <- sf::st_buffer(the_geometry, buffer)
    #return sf_use_s2 to the original value
    sf_use_s2(temp)
  } 


  #extract the values with terra
  values_df <- terra::extract(ntl_raster_terra,
    the_geometry, #this can be points or polygons
    fun = mean,
    exact = FALSE, #TRUE, exact fraction of cells
    weights = FALSE, #TRUE, approximate fraction of cells
    method = "simple",
    na.rm = TRUE,
    ID = FALSE
  )
  
#convert the values_df to long format
...
...

[alexyshr]

Hi @ramarty

I have been working Black Marble data in HDF5 format with stars; it is faster than terra. I can share the code with you to evaluate if stars is a better option. Another option is that you implement functions using terra and I can contribute with the functions using stars, then after some use of the tools you can decide. Best

[ramarty]

@alexyshr If you have some code to share that'd be great -- I think stars could be a good option for this, even with a set of parallel functions that use a stars approach

[alexyshr]

If already implemented different functions. I want to check the latest version you posted and I would love to contribute to the project. I can suggest (push-commit) small changes or new functionality and you can check and decide if leaving them as part of the package. Please let me know if we can work collaboratively. Thanks

[ramarty]

@alexyshr This sounds great - and absolutely welcome contributions / would be great to work collaboratively here. The goal is to make the package as useful as possible to different workflows, so think your contributions would be particularly useful here.