1. Preserve the original export
Work on a copy and retain station identity, sensor configuration, maintenance, timezone and accumulation settings. A neat CSV is not proof that the instrument measured continuously. Record which columns came from the logger and which were calculated.
2. Identify what the rain column means
A counter is an accumulated depth; an interval depth belongs to a stated time window; intensity is a depth rate. A daily total is not a one-hour increment. Do not paste tip counts into a depth field or feed intensity readings to an event separator without a justified interval conversion.
3. Reconcile clocks and interval ends
The rainfall tools use minute-precision end labels in one fixed-offset clock. Translate changing daylight-saving labels first and keep that transformation documented. The depth ending at01:00 for a 60-minute step represents00:00–01:00. A missing 01:00 record does not mean that hour was dry. Duplicate, descending, impossible calendar dates and off-grid times stop the calculation.
4. Distinguish missing from observed dry
Write a measured zero as0. Use a blank or NA for an unavailable measurement. Other quality codes must be reconciled externally, not left to numeric coercion. A period with no observations is not an observed dry period. Keep a separate explanation for instrument outages, blocked collectors and suspect readings; these tools do not decide whether a physically present number is trustworthy.
5. Reduce accumulated counters conservatively
Inspect deltas at the intended step. A reset or roll-over may have an instrument-specific explanation, but the new count alone does not reveal how much rain fell before it. A positive difference across a gap is not an evenly distributed sequence. The counter tool retains these intervals as unknown and exports NA rather than a fabricated depth.
A reset changes what can be summed
At hourly counter readings 0,1,3,0,2 the known ordinary increments are1,2 and 2 mm. Their sum is 5 mm; rainfall during the decrease is unknown. Calling5 mm a complete four-hour total would overstate the evidence.
6. Choose an event definition for the project
Enter the dry gap explicitly and compare sensible alternatives under the governing study method. An observed dry run at least as long as the chosen gap separates groups. Shorter internal dry runs remain within event duration. A gap with no observations cannot establish separation and creates unknown-boundary fragments instead.
First/last wet groups need sufficient known dry context too. Retain left/right flags alongside depths, durations and interval intensities; do not export only the apparently clean rows and then fit a frequency model as if they were complete independent events.
7. Compare station consistency without automatic corrections
Align target and reference periods and retain the reason for a chosen split. Double-mass accumulation compares usable pairs only. Segment ratios can change because of many measurement, exposure, network or climatic circumstances. The descriptive checker neither proves a cause nor computes a defensible correction for the original observations.
8. Keep spatial weights separate from data availability
Obtain weights from the actual basin/station geometry or the approved method. Enter every station even when some observations are missing. The network tool withholds the full-basin mean for those periods rather than reallocating missing weight. A complete0,0,0 period is genuinely different from NA,0,NA.
9. Retain a qualified, reproducible packet
- Original CSV and the exact clock/depth/interval basis.
- Counter reset/gap evidence or interval data provenance.
- Chosen dry gap and event boundary flags.
- Reference construction, original-row split and paired exclusions.
- External station-weight map and period-level network coverage.
- Full audit CSV; distinguish it from the bounded screen/print preview.
What this workflow deliberately does not do
No observation is automatically corrected, gap-filled, declared calibrated or converted into a design storm, runoff, tank yield, forecast or legal conclusion. Existing rainwater-harvesting tools use planning assumptions for a different user decision. Keep that calculation separate from measurement evidence.
Sources
- Colorado CDSS TSTool — Delta, missing endpoints and counter resets
- IETD method documentation — interval rainfall and chosen dry gaps
- Colorado CDSS TSTool — Cumulate and double-mass analysis
- NOAA NWS — Basic Hydrologic Concepts: mean areal precipitation