IDV now includes a runGemini feature, shown as the Gemini Agentic Vision dialog, that sends a weather-analysis request to Google Gemini from inside the application. User must enter their own Gemini API key; IDV does not ship a shared key. A prepaid key can be created in Google AI Studio, then billed through Gemini API prepaid billing. After the key is stored, the user can attach archived weather maps and analysis files, choose a master prompt, reuse or clear prior instructions, and type an Instruction for Gemini Agent plus optional Knowledge Context. A Reset Gemini Key control replaces the stored credential. After OK, those inputs and images are sent with the user’s key so the model can interpret the maps and return an analysis without leaving IDV.

The IDV Data Choosers tab now includes an ObjectStore URL interface under General, so a cloud object can be specified the same way as an OpenDAP URL. The form lets you set the data source type, choose a host such as Amazon, Google, Microsoft, optionally supply a profile and custom endpoint, then pick a bucket and object key. Pressing Add Source loads that object into IDV without downloading the archive first.

IDV can now browse Amazon S3 object stores directly from the Data Catalog Chooser, so users no longer need a local copy of the archive. Enter a bucket URL such as https://noaa-mrms-pds.s3.amazonaws.com?delimiter=/ or https://noaa-goes19.s3.amazonaws.com/?prefix=ABI-L2-CMIPF/&delimiter=/, or https://unidata-nexrad-level2.s3.amazonaws.com/?&delimiter=/, and IDV lists the bucket as a folder tree: S3 common prefixes appear as expandable directories and objects appear as files. Large listings are paged, folders can be shown newest-first or oldest-first, and a selected NetCDF or GRIB object can be opened in place through the cdms3: path, including use in NcML time aggregations for satellite, NOAA MRMS and level2 radar dataset. This puts public cloud holdings such as NOAA HRRR and GOES ABI on the same path as a local file, from browse to display.

The version of the netCDF-Java library currently distributed with the IDV is 5.11.0. See the netCDF-Java Library for more details information.

IDV now includes Jython functions that locate local maxima and minima on a grid and return them as point observations. findLocalMaxMinAsPointObs2D(grid, radius, function) scans a 2-D field, and findLocalMaxMinAsPointObs3D(grid, level, radius, function) does the same on a selected level of a 3-D field. Each grid point is compared with neighbors inside the given radius so that only true local highs or lows are kept, which is the usual way to mark H and L centers on surface pressure. The result is a set of points that can be plotted on the map, so closed-contour highs and lows can be labeled automatically instead of being picked by eye.

• Fixed a bug related to ECMWF GRIB2 DRS 42 (CCSDS compression).
• Fixed a bug related to the TDS radar server catalog.

For a list of outstanding-known problems, see the Known Problems page.