Release v4.0.0
2023-09-26
Version description of the v4.0.0 release of ICESat-2 SlideRule. This document also captures functionality added in versions v3.5.0, v3.6.0, and v3.7.0.
Breaking Changes
This version contains a number of backward-incompatible changes, specifically to the names of the fields being returned by the atl03s and atl06 APIs, and the Python client function APIs. These changes were made to standardize the downstream processing of the photon and elevation data, and also to bring the names of the fields being returned by SlideRule closer to the ICESat-2 Standard Data Products.
For
atl03sandatl03spAPIs:distanceis nowx_atcAcross track distance is now
y_atc
For
atl06andatl06pAPIs:distanceis nowx_atcAcross track distance is now
y_atcdh_fit_dyhas been removedlonis nowlongitudelatis nowlatitude
The compact ATL06 record
atl06rec-compacthas been removed entirely and is no longer supported.The
assetparameter has been removed from the Python client functions and is now managed internally by the client. Users can still override the asset by supplying anassetkey in the request parameters; but going forward users should not generally need to worry about which assets the servers are using.
Major Changes
Added new capability to subset rasters. The
subsetsendpoint subsets rasters and returns the data back to the user. It works in conjunction with theraster.subsetPython API.Added new
opendataplugin that supports sampling the ESA World Cover 10m dataset.The Python client supports returning a GeoDataFrame with 3D point geometry. The user must supply a “height” column name in the API call in order to receive back a GeoDataFrame with the 3D point geometry; note there is a performance impact when selected. (See #272).
User’s can supply their own PROJ pipeline in raster sampling requests. To use this feature, add a “proj_pipeline” string to your request dictionary:
proj_pipeline = "+proj=pipeline +step +proj=axisswap +order=2,1 +step +proj=unitconvert +xy_in=deg +xy_out=rad +step +proj=stere +lat_0=-90 +lat_ts=-71 +lon_0=70 +x_0=6000000 +y_0=6000000 +ellps=WGS84"
parms = { "poly": region['poly'],
"cnf": "atl03_high",
"ats": 5.0,
"cnt": 5,
"len": 20.0,
"res": 10.0,
"maxi": 1,
"samples": {"mosaic": {"asset": "rema-mosaic", "radius": 10.0, "zonal_stats": True, "proj_pipeline": proj_pipeline}} }
gdf = icesat2.atl06p(parms, asset=asset, resources=[resource])
Localization of the CRS transform when GDAL is trying to select the best transform to use (did I say that right???). If what I said is confusing, please reach out to Eric. For non-mosiac datasets (i.e. arcticdem-strips, rema-strips, 3dep, landsat), the code will automatically provide a bounding box of the raster being sampled to GDAL when it does the coordinate transformation. For mosaic datasets, you can now provide your own bounding box with the request. To do so, add the “aoi_bbox” parameter to your request dictionary:
region = sliderule.toregion('grandmesa.geojson')
bbox = raster.poly2bbox(region["poly"])
parms = { "poly": region['poly'],
"cnf": "atl03_high",
"ats": 5.0,
"cnt": 5,
"len": 20.0,
"res": 10.0,
"maxi": 1,
"samples": {"gedi": {"asset": "gedil3-elevation", "aoi_bbox": bbox}}
}
Overhaul of how ancillary fields are handled by the server-side code. A new system-wide record object type has been added called a “container record”. A container record holds inside of it other records that are grouped and processed together. Ancillary data associated with an ATL03 extent or ATL06 elevation is now placed in a container record along with the corresponding ATL03 or ATL06 record.
Parquet file generation has been optimized and the gzip compression has been replaced with snappy compression. Also the ability to generate a non-geo parquet file has been fixed, where the latitude and longitude are present as normal columns and there is no geometry column.
Default ICESat-2 products used now set to version 006.
The ATL03 subsetting code now operates on one spot per thread (it used to be one pair track per thread). This change is made in conjunction with a change in the base AWS instance type to an instance that has 8 cores. The additional subsetting threads will therefore be able to utilize the greater number of cores when subsetting and speed up processing requests when only one subsetting operation on a given node is being processed.
When making raster sample requests, the user no longer needs to perform a separate earthdata.* call to generate a catalog for the 3DEP and Landsat raster datasets, but the client is not smart enough to know to generate the catalog on its own if the user does not supply it.
Development Updates
SlideRule Python bindings (srpy) have been updated to support building them inside a conda managed OS package environment.
The Python bindings for H5Coro are superseded by the new pure Python implementation effort: h5coro
Major rewrite of GDAL raster sampling code: new class structure has GdalRaster as a stand-alone class, with RasterObject parenting a GeoRaster and a GeoIndexedRaster for mosaic and strip datasets respectively.
The
latesttagged docker images for the sliderule compute node, the intelligent load balancer, the proxy, and the monitor, are now all built automatically on every push tomainby a “Build and Deploy” GitHub Action workflow. The new images are used when the automated Pytests are kicked off for thedeveloperscluster.
Issues Resolved
#182 - Across track slope not calculated for ATL06
#202 - Feature request: export x_atc and y_atc
#226 - Migrate ps-web documentation to provisioning system
#233 - Create performance PyTests
#235 - Need PyTests for raster sampling of strips
#246 - Add PyTest for file_directory raster code
#262 - Replace “nodata” value with NaN as sampled value in GeoRaster
#268 - Mismatch in pyH5Coro declarations
#270 - Using H5Coro in stand alone mode
#280 - Make default number of ATL06 iterations 6
#283 - feature request: report y_atc in ATL06-SR
#284 - Feature idea : extract arbitrary fields from ATL03 to ATL06-SR
#293 - Add ESA WorldCover plugin
#295 - Requests to TNM for 3DEP 1m is timing out
#299 - Errors reading ArcticDEM mosaic
#300 - Put querying the catalog for raster datasets inside the API functions in the client
#309 - GeoUserRaster causes use after free
#312 - Bitmask/flags rasters should only be read when flags param is set
Getting This Release
https://github.com/ICESat2-SlideRule/sliderule/releases/tag/v4.0.0
Benchmarks
atl06_aoi
output: 571661 x 15 elements
total: 37.352343 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 17.967412 secs
todataframe: 0.683151 secs
merge: 0.000002 secs
flatten: 2.631864 secs
atl06p: 37.137683 secs
atl06_ancillary
output: 1180 x 16 elements
total: 2.789055 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 0.584189 secs
todataframe: 0.006167 secs
merge: 0.018061 secs
flatten: 0.030961 secs
atl06p: 2.669654 secs
atl03_ancillary
output: 1180 x 16 elements
total: 3.039502 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 0.204709 secs
todataframe: 0.006145 secs
merge: 0.017784 secs
flatten: 0.030781 secs
atl06p: 3.038152 secs
atl06_parquet
output: 1577 x 16 elements
total: 2.704171 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 0.208408 secs
todataframe: 0.006145 secs
merge: 0.017784 secs
flatten: 0.096728 secs
atl06p: 2.703798 secs
atl03_parquet
output: 22833 x 23 elements
total: 1.926006 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 0.011052 secs
todataframe: 0.006145 secs
merge: 0.017784 secs
flatten: 0.096728 secs
atl06p: 2.703798 secs
atl03sp: 1.908603 secs
atl06_sample_landsat
output: 914 x 19 elements
total: 12.250066 secs
gdf2poly: 0.000495 secs
toregion: 0.187156 secs
__parse_native: 0.387278 secs
todataframe: 0.006386 secs
merge: 0.015527 secs
flatten: 0.034550 secs
atl06p: 4.577109 secs
atl03sp: 1.908603 secs
atl06_sample_zonal_arcticdem
output: 1651 x 26 elements
total: 8.280844 secs
gdf2poly: 0.000337 secs
toregion: 0.005492 secs
__parse_native: 0.309996 secs
todataframe: 0.007301 secs
merge: 0.025982 secs
flatten: 0.050954 secs
atl06p: 8.273125 secs
atl03sp: 1.908603 secs
atl06_sample_nn_arcticdem
output: 1651 x 19 elements
total: 4.964582 secs
gdf2poly: 0.000344 secs
toregion: 0.005554 secs
__parse_native: 0.085348 secs
todataframe: 0.006952 secs
merge: 0.023732 secs
flatten: 0.043986 secs
atl06p: 4.957544 secs
atl03sp: 1.908603 secs
atl06_msample_nn_arcticdem
output: 1526622 x 19 elements
total: 151.426132 secs
gdf2poly: 0.000327 secs
toregion: 0.005378 secs
__parse_native: 81.071236 secs
todataframe: 2.486024 secs
merge: 17.094814 secs
flatten: 30.848135 secs
atl06p: 150.431377 secs
atl03sp: 1.908603 secs
atl06_no_sample_arcticdem
output: 1526622 x 15 elements
total: 62.244200 secs
gdf2poly: 0.000369 secs
toregion: 0.006280 secs
__parse_native: 46.702178 secs
todataframe: 1.159332 secs
merge: 0.000003 secs
flatten: 6.242216 secs
atl06p: 61.391358 secs
atl03sp: 1.908603 secs
atl03_rasterized_subset
output: 51968 x 21 elements
total: 16.987414 secs
gdf2poly: 0.000369 secs
toregion: 0.006280 secs
__parse_native: 1.420302 secs
todataframe: 0.029791 secs
merge: 0.000003 secs
flatten: 0.278818 secs
atl06p: 61.391358 secs
atl03sp: 16.692929 secs
atl03_polygon_subset
output: 50799 x 21 elements
total: 2.615179 secs
gdf2poly: 0.000369 secs
toregion: 0.006280 secs
__parse_native: 1.153502 secs
todataframe: 0.030567 secs
merge: 0.000003 secs
flatten: 0.276284 secs
atl06p: 61.391358 secs
atl03sp: 2.595944 secs