01 / Compare two choices
Keep the dates in view.
All targets end 31 December 2024. Reference years identify window endpoints; each member uses the same calendar months as its target. A reference mean averages complete window totals, not monthly rates.
On narrow screens, scroll charts and tables horizontally. Keyboard: Tab to the region, then use Left/Right.
B Second comparison
Inspect reference windows and completeness
The two total charts share a zero origin and the same absolute scale, including the selected reference-member totals in that scale. Longer accumulations cover more time; these bars are totals, not rainfall rates. “Above” and “below” are descriptive directions, not performance scores. Totals display 2 decimals in inches or 3 in mm; means and scaled assumptions display 4; percentages display 6. Rounding and converted digits add no measurement precision.
02 / Inspect the fixed design
Twenty declared comparisons.
Each cell is 100 × (target total ÷ reference mean − 1). Select a cell to load it into comparison B. The double border marks the loaded choice. The unavailable cell remains part of the design.
The reference sets are nested; 24-month windows overlap and some share dates with the target. These dependent contrasts are not independent robustness trials. No trend, significance test, percentile or return period is estimated.
View all comparison totals
03 / See the accumulation
A year with rainfall concentrated in February.
On narrow screens, scroll the chart horizontally or open the values table below.
View 2024 monthly values and trace counts
04 / Synthetic demonstration
What if one month were unknown?
Every date from 1 July 2018 through 31 December 2024 is present with a PRCP value in the frozen CSV. This synthetic demonstration hides one known month at a time. It does not alter the observations or the primary comparisons above. The reference is fixed to the 2019–2023 calendar-year mean.
Strict completeness withholds the full total. Zero fill assumes the missing month had zero rain. Scaling multiplies the observed 11-month sum by 12/11 and assumes exchangeable months despite seasonality. Neither replacement is adopted as an imputation.
Inspect all 12 synthetic masks
05 / Evidence and boundaries
Reproducible arithmetic. Limited inference.
Observation and missingness rules
2,376 unique contiguous daily dates form 78 complete calendar months. There are no missing PRCP values and no nonblank quality flags in this frozen input; all source flags are W. The 58 trace flags are retained: numeric 0.00 inches at the supplied precision does not mean physical zero rainfall.
Provider calendar-date keys are used as supplied, with leap days retained. Observation-time attributes are blank; a UTC interval or daily reporting cutoff is not inferred. Daily amounts are summed at hundredth-inch precision. Inches × 25.4 converts units without adding measurement precision.
Fixed-reference rule
A requested target or reference window must contain every calendar month. If any requested reference member is incomplete, its entire selected reference mean and percentage are unavailable. No endpoint is dropped. A zero reference denominator would make the percentage undefined.
The 24-month reference with endpoints 2019–2023 requires January–June 2018, outside this extract. Its endpoint-2019 member has 18 of 24 months; the other four complete members do not replace it. The existing 2020–2023 option uses four complete reference windows. Selecting it changes the reference set; it does not fill the months outside the extract.
The local coverage check found no date gaps or blank PRCP values within the extract. Reported zeros, including 58 trace values, are observations. A coverage cutoff cannot supply absent rainfall amounts. See the methods and completeness rule and two synthetic daily counterexamples.
Station and scope limits
Legacy NOAA HOMR remarks dated 20 May 2024 describe a move; formal records change 24 May. Neither establishes the exact observational handoff. The daily series retains USW00093134 and the old USC name while the metadata crosswalk remains unresolved. All selected targets include post-change observations.
NOAA’s station-practice biases are not homogenized. Matching numeric values and flags do not establish physical-site continuity. This point series does not represent a county or fire footprint and supports no fuel, ignition, climate-trend or causal/predictive wildfire conclusion.
Inspect and reproduce the comparison
The frozen NOAA extract, calculation source and build instructions make the selected dates, reference members and completeness rules inspectable. The offline build embeds the data, charts and controls in one page.
Data sources and reproducibility
This explorer uses a frozen NOAA-only snapshot accessed 21 October 2026. The app and portable build use local data. GHCN-Daily Version 3 identifies the cited collection; the accompanying hashes identify this snapshot.
- Menne et al. (2012), GHCN-Daily Version 3, NOAA NCDC — PRCP, USW00093134, 1 July 2018–31 December 2024, accessed 21 October 2026.
- Menne et al. (2012), dataset overview, JTECH 29, 897–910.
- Exact NOAA daily-summaries request · API and units · GHCN-Daily flags.
- NOAA station history · GHCN-Daily metadata and caveats · NCEI source policy.
- Inspectable files: frozen app results, frozen NOAA daily CSV, input hashes and citations, and source and rebuild instructions. See the methods and limitations and synthetic daily-mask analysis.
See NCEI’s linked source policy for source-data terms. No additional license for the original code, charts or prose is granted by this export. This work sample is not an official NOAA product or endorsement.
Frozen daily CSV SHA-256: 113a694665a7d1be8e66f02442d4aac6c32fc3f846f9ec903fe192b5c19a7fed