[ANN] GeoStats.jl - Geospatial Data Science and Geostatistical Modeling in Julia

Package Announcements
, , , ,
41

Quick announcements:

  • GeoStats.jl v0.82 is out with full support for PrettyTables.jl v3.
  • New tutorial by Olav Moyner with GeoStats.jl + JutulDarcy.jl for petroleum reservoir simulation: Quarter-five-spot example | JutulDarcy.jl
  • New tutorials by Marcos da Silva coming up soon in the context of agriculture and forestry:

We are starting a new research project in O&G that will give us the opportunity to work on exciting new features. Stay tuned.

42

Quick updates:

  • ZScore and StdFeats now support missing values
  • GHC clustering now accepts multiple k to cut the tree
  • NN now picks the nearest non-missing value
  • isconvex now implemented for Hexahedron
  • EPSG{29902} added for Ireland region
  • New ccdf for Ensemble of realizations
43

GeoStats.jl v0.83

We are actively working on an improved interface for geostatistical learning. The Learn transform has been refactored to handle labeled geotables produced with the new label function:

image
image831×702 58.1 KB

The columns of the geospatial table are now colored according to their role in geostatistical modeling. Predictor variables are colored in sage, target variables are colored in red, and the geometry column is colored in teal.

More updates

  • DropLocalLowHigh is a new transform that drops local extremes with a moving neighborhood. It can be useful to remove data acquisition artifacts in various applications.
  • All EPSG and ESRI codes for UTM coordinates have been mapped to our native CRS types with WGS84 and SIRGAS2000 datum. That should cover a lot of datasets in industry.
44

GeoStats.jl v0.84

One of our favorite releases this year! New features include faster and more accurate geostatistics with non-Euclidean geometry. Update and enjoy it!

More info: [ANN] Meshes.jl - Computational Geometry in Julia - #58 by juliohm

45

GeoStats.jl v0.85

Very important updates, including full support for differential and integral calculus:

and removal of external dependencies in GeoIO.jl:

We are also seeing publications from the Python community using GeoStats.jl as the reference implementation, which is a very good sign of project maturity:

https://rmets.onlinelibrary.wiley.com/doi/10.1002/gdj3.70064

46

Quick updates:

Replaced LScene by Axis3 with hints provided by the latest Makie.jl release:

viewer
viewer1149×797 115 KB

The axis labels are automatically set based on the CRS of the domain.

New publication for those interested in geostatistical modeling at scale:

47

Quick updates:

cbar is much improved and we added support for all kinds of scientific data, including distributional, compositional and unitful data. The viewer (and viz) recognize all objects supported by Colorfy.jl and use the transparency channel to codify uncertainty, etc.

Screenshot_20260619_102451
Screenshot_20260619_102451829×458 165 KB

Please update the stack to get these new features.

More publications:

https://www.sciencedirect.com/science/article/abs/pii/S0098300426000981

48

Quick updates:

funplot and surfplot reviewed and improved to handle different assumptions about geostatistical functions (e.g., variogram, transiogram):

Screenshot_20260624_155556
Screenshot_20260624_1555561049×1124 701 KB

Added precompilation workloads to reduce time-to-first-plot.

More publications: