DZTR researcher documentation

Estimate reservoir operating targets, calibrate them where observations exist, and apply frozen profiles in hydrological models. Start with the portable application guide; it needs no private experiment archives. The scientific assessment explains the evidence and limitations behind the available methods.

flowchart LR
    A[Local observations or station components] --> B[Normalize units, dates and boundaries]
    L[Levels and geometry with a declared reference] --> B
    B --> C[Usable-data route and source decisions]
    C --> D[Mixed calibration years: at least 65 percent]
    D --> E[Estimate targets and optional calibration]
    E --> F[Freeze a selected profile and its provenance]
    F --> G[Apply to inflow plus initial conditions]
    G --> H[Storage, release, spill and balance diagnostics]

Task	Guide
Install and reproduce three public applications	Portable application
Understand commands, clocks and file contracts	Public interface
Prepare a new reservoir	New-reservoir checklist
Choose a data-availability method	Four-route workflow
Understand level datum and geometry choices	Shared geometry workflow
Understand wet/dry/normal year selection	Representative-year calibration
Choose release, storage or weighted objectives	Full-data scalar calibration
Interpret roots, failures and water conservation	Root policies, interval simulation
Review performance and computational budgets	Scientific assessment
Download, cite and verify a release	Release and citation

The stable command is dztr (also python -m dztr). Versioned run managers remain the scientific backends. Earlier guides describe frozen experiments and can refer to private outputs/ directories; those paths are archive references, not prerequisites for the portable quick start.

The GitHub Pages outlet builds from these Markdown sources after the public Ubuntu and Windows application checks pass. Version 0.3.1 uses the MIT license. The release guide records the v0.3.1 software DOI; the all-versions DOI identifies the software across releases. The published model paper has its own separate DOI.