- add argument for specifying temp directory for download files. Allows users to specify a specific location. (Thanks, @andrew-caudillo: #95)
- exposed ncpu argument so user can control. Defaults to 2 if more than 2 cores available. Thanks to @courtiol for finding this issue and the suggestion!
- dropped sp, rgdal and rgeos depends
- All sf
- Terra
- Still returns RasterLayer, but raster will be dropped in upcoming release.
- Add location clip example
- cleaned up examples (e.g. plot was throwing warning, now max.plot = 1)
- Switched to MIT from CCO
- clean up: typos, update citations for new CRAN req, hide one test (thanks @Nowosad)
- Updated documentation
- added startup message warning of switch to sf and terra in future releases.
- Changed API ERRORS to messages and return NA. Would bomb out runs when this would happen only occasionally. Also this should meet CRAN policy on failing gracefully.
- Switched long lat check from my homespun thing to st::sf_is_longlat
- Fixed EPQS API URL. Moved to a new one. Thanks @haas4726 for the catch.
- URL fixes in docs
- OpenTopography now requires an API Key. This can be acquired form OpenTopography. You need to set it with elevatr using elevatr::set_opentopo_key(). After a restart, elevatr will use this key.
- The "Introduction to elevatr" vignette has been updated to include chanage reflected in version 0.4.2 and also includes a new section on accessing data from OpenTopography and details on setting the API key.
- The epqs server was occasionally returning an empty response (see #29) and would error. If that happens now, elevatr will retry up to 5 times (which usually fixes the issue). If still an empty response after 5 tries, NA is returned and a warning is issued indicating what happened.
- Changing to future::plans was losing tempfiles on parallel downloads. Moved the change back to serial plan after creation of raster.
- Changed get_tile_xy. My math was messing up in areas near 180/-180 longitude were trying to grab non-existent tiles. Now using slippymath::lonlat_to_tilenum instead.
- NA's introduced with simultaneous gdal mosaic and project. Now uses two steps. Solves https://stackoverflow.com/questions/67839878/gridded-dot-artifacts-in-geom-raster-plot
- Previously errored on solaris. Skipped tests. A fix has been proposed within sp. Once that is available can re-test on solaris.
- Updated CITATION file to reference new version and new Zenodo DOI
- The prj argument relied on proj4string in the past. That support is going away in order to keep up with changes in PROJ and the rest of R Spatial ecosystem. Spatial reference will be pulled form "locations" if they have it defined or will be pulled from an acceptable SRS_string. EPSG codes in the form of "EPSG:4326" are a pretty safe bet. WKT strings work well on PROJ > 5.2.0.
- Empty rasters were failing as elevatr was using nrow(locations) to get number of features. It still does that as the default behavior, but if nrow(locations) returns a null, it uses length(locations) instead. Thanks for the catch, Gengping Zhu!
- Documentation fix on get_elev_raster, now correctly reports that the function returns a raster, not points. Thanks @AndyBunn!
- sfc objects getting missing in coercion in loc_check. Not anymore!
- epqs occasionally times out, but subsequent hits usually work fine. Added a second hit when that happens and if that second one doesn't work then it assigns elevation to NA and throws a warning, instead of erroring
- Was suppressing messages (and thus progress bar) on get_elev_pt src = "aws". Turned that off so progress of building the DEM is tracked.
- proj_expand was using buffers to expand. Not great for geographic projections.Now it adds the expansion to the max and subtracts from the min to expand the bbox by the expand value. For raster retrievals with a single point the resultant raster will be significant smaller than previous (approximate 1km by 1km). Multiple points should see no difference. Thanks to WithRegards on SO for helping me find this.
- In tests with spTransform, changed SRS_string to CRS(SRS_string=paste0("EPSG:", ll_prj$epsg)). Details in #56. Thanks to rsbivand and Fonteh-Bonaventure for helping me with this.
- Raster locations were not returning correctly, that is now fixed.
- Added access to OpenTopography Global Bathymetry SRTM15+ V2.1 with src = "srtm15plus"
- serial loop for get_epqs was taking a long time (API returns are slow), so use furrr::future_map_dbl to paralellize the gets. Defaults to 1 minus available cores.
- Added argument to get_eqps to control serial vs parallel API calls. Defaults to serial for 35 or fewer points, but can be set to TRUE for force serial.
- Added overwrite argument to get_elev_point() to check for existence of elevation and elev_units columns. If either exist and overwrite not TRUE then errors.
- Updated progress bars to use the progressr package.
- Converted all coordinate reference system handling to pull from locations or an SRS string. Should take care of running on older versions of PROJ
- Removed message that reported out CRS, was too verbose and not necessarily useful.
- Cleaned up message on units.
- updated tests to deal with build errors
- Parsing crs through st_crs()
- Fixed vignettes
- Rasters were not getting handled correctly by size estimation. Thanks to @tteo for the catch (#37)
- Single point requests to Open Topography were failing. Expands now to capture small area around single point
- Switched to using wkt instead of proj4. I think this will help with the more recent versions of PROJ...
- Added rgdal_show_exportToProj4_warning=thin to options on load.
- Fixed #38. Thanks to @cjcarlson, @ACheysson, and @jsta for helping track this one down.
- Added user agents to httr requests
- Added new OpenTopography API access to three global datasets, SRTM GL1 and 3, and the ALOS World 3D 30m. These are accessed via the src argument
- Added new argument to convert any negative values to NA.
- Zoom levels 1 and 0 were throwing errors becuase tile selction was overzealous and was selecting tiles that existed. Conditionals to check fo this. Also zoom 0 returns as "image/tif", not "image/tiff" that all other levls return. More robust checking on return type.
- Tiles for points that fall exactly on the equator were returning NA on
get_elev_point(). On tile selection inget_tilexy()a conditional was added to check for lat == 0 and projeciton being and acceptable proj4 alias of Lat/Long. If that is met a very small (~ 1meter) expansion to the bounding box is done. Thanks @willgearty for the bug report #25. - USGS epqs return -1000000 for areas without an elevation.
elevatrnow converts those values to NA. Thanks to George Moroz for the catch! #24 - Updated stale sp objects. Thank you Roger Bivand for making the update very easy!
- Updated links to mapzen documentation. Thanks to Alec Robitaille for the fix.
- Sped up merging and projecting via PR from Mike Johnson. Thanks for the contribution, Mike.
- Major change for this is dropping Mapzen support since Mapzen shutdown in January of 2018. Replacement services for terrain tiles exist at Nextzen, however; the published geotiff endpoints were not working. I opted to not include the Nextzen endpoints at this time. Rolled back to terrain tiles from AWS only.
- Added point elevations from AWS. Extract point elevations from a DEM obtained via
get_elev_raster(). Will likely be faster for cases with many points in a relatively small geographic area. - Added a clip argument to
get_elev_raster(). Default behavior of returning the full tiles is the same as in prior versions. The argument expands this by allowing users to clip the resultant DEM either by the bounding box of the input locations viaclip = "bbox"or by the locations themselves viaclip = "locations". Partly inspired by #13. Thanks to Michael Sumner (@mdsumner) for the inspiration. - Support for input simple features of the class
sfhas been added. This is supported by coercion of the inputsfclass to aSpatialXDataFrame. Ansfobject is also returned when used as the input locations forget_elev_point
- Updated Vignette to reflect new focus on AWS and USGS
- Updated Tests
- Updated README
- Added message to inform user of vertical units and CRS.
- Primary change with this released is fixing a bug with the return file type on the AWS and mapzen APIs. "tif" was changed to "tiff" and the check was stopping processing of the raster images. Details are on #17. Thanks to the following individuals for catching this: @yipcma, @TomBor, @jslingsby. And thanks to @vividbot for #18 which provided a fix.
- Thanks to @pascalfust for kicking me into gear to send fix to CRAN.
- Fixed NOTE on CRAN: Packages in Imports, not imported.
- Removed prettyunits
- moved rgdal to suggests
- Changed where ratelimitr getting called (was not in a function so couldn't be exported/called.
- Fixed travis build errors caused by change in elevation API that now requires a key.
- Added deprecation message to get_elev_point and get_elev_raster, due to pending shutdown of Mapzen :(
- There was a typo in building the mapzen api key. Was masked prior as a keyless access was allowed. It no longer is and get_elev_raster was failing. That has been fixed
- Tests also failing due to keyless access. Encripted key now pushed for use on travis. Tests not run on CRAN
- Thanks to @hrbrmstr for pointing me in the right direction on fixing the testing with an api key.
- Also thanks to @noamross and @ropensci for maintaing https://discuss.ropensci.org where I found https://discuss.ropensci.org/t/test-api-wrapping-r-packages-with-oauth-tokens/157. And thanks to @jennybc for wrapping all this up and provide great guidance on testing and vignettes that require a key. That info is here: https://rawgit.com/jennybc/googlesheets/master/vignettes/managing-auth-tokens.html#encrypting-tokens-for-hosted-continuous-integration
- inst/doc was inadvertently included in package. This verisons removes that and includes only vignettes.
- This is the initial CRAN release. Provides access to point elevation data from USGS and from Mapzen. Provides access to raster DEM from Mapzen Terrain Tiles and AWS Terrain Tiles.