v1.4.0

Highlights

Network support

Compare observations with 1D network models. mikeio1d reads the result file.

• Install with pip install modelskill[network]. It requires mikeio1d[network] >= 1.4.0.
• NetworkModelResult reads .res1d, .res11 and EPANET .res files.
• NodeObservation locates data by node name or by (reach, position). ReachObservation locates data by reach name.
• from_multiple creates several observations in one call.

User guide: Network.

Vertical skill

Compare vertical profiles with 3D layered dfsu models.

• VerticalObservation holds a profile at a fixed x, y.
• Comparer.sel(z=...) selects by depth.
• Comparer.vertical.skill(bins=...) computes metrics per depth bin.
• Comparer.vertical.plot.profile() and .hovmoller() plot profiles and depth against time.

Breaking changes

• Python 3.12 or later is required (#674).
• match() accepts only Observation and ModelResult objects. Build them with ms.observation() and ms.model_result() (#533).
• Scatter plot quantile points use quantiles 0–1. Previously they were percentiles (#644).
• spatial_overview draws tracks in one colour (#565).

Fixes

• gridded_skill(binsize=...) keeps data at the domain edges (#543).
• DfsuModelResult accepts an xarray.Dataset (#605).
• Input DataFrames are no longer modified (#562).
• import modelskill takes half as long (#662).