variogram() supports stars (raster) objects, benefiting from them being griddedimport sftime; modify krigeST() variogram functions to accept sftime objects for data (as alternative to STI or STIDF), and stars or sftime objects for newdata; #108 with great help from @henningte
fix Gaussian cosimulation, probably introduced in 2016, #106
NA as estimate when prediction/simulation fails; #80use multiple cores in variogramST, using pkg future; #63 by @sigmafelix
fix bug with conditional simulation using stars target grid and nsim=1, #58
fix bug in support for sf objects; #46
fix krigeTg for the case when data or newdata are of class sf or sfc; #51
try to fix CRS issue found on OSX
check for temporal unit in krigeST opposed to covariance function and assign the temporal unit found in krigeST to the spatio-temporal variogram for consistency in case none has been provided by the model; add warning when ST variogram doesn't have a temporal unit attribute; #42
add plot.variogramModel; #40
support sf and stars objects, for vector and raster data, respectively; #39
memcpy for overlapping regions error, found in valgrind checks by Brian Ripleyfit.variogram to choose initial values for range, sill and nugget before fitting, and fit over a range of model typesNA values in vgmMakevars, remove configureai-geostats.orgfit.StVariogram example in dontrun