Other pages in this series (34)
Routine-entry index — the bare 晴 and the rain gauge
By reign: share of days with '晴' alone, longest and mean run of such days, share of solar-July (monsoon) days with rain recorded; and the share of rain-gauge days (Lee et al. 2024) on which the cell says 雨.
Only 73% of days with measured rain carry 雨 in the cell — and the weather string transcribed by Lee et al. 2024 gives the same 72.7%, so this is a property of the cell. Days whose gauge amount is written as 0 (989; the paper's Table 1 describes a 0 mm Ilgi entry as a trace) carry a rain, snow or hail token on 98%; of the 36,138 days with no gauge entry, 雨 appears on 1. The form leans toward omitting rain.
- Formula
- Gauge recall = rain_any days among days with gauge amount (rrn) > 0 (also shown: precip_any — any rain, snow or hail token — and the share of Lee et al. 2024's weather string carrying 雨). Counted per calendar date: a date with two weather nodes counts as a rain day if either node records rain (the 'any' policy; the 45,632 weather nodes fall on 45,447 dates, and 12 of the 185 dates with two nodes disagree. Of the four dates that appear twice in Lee's file the row with a gauge value is kept; up to 1.2.2 those duplicates added four rows to the join, hence 45,636 and 189). Of Lee's 8,563 measured-rain days, 243 have no weather node in the diary and drop out. Days whose gauge amount is written as 0 (the paper's Table 1 describes a 0 mm Ilgi entry as a trace; flag e marks weather–amount inconsistencies and is not this set) and days with no gauge entry are counted separately. 晴 runs are counted over unique dates and broken by calendar gaps. Wilson 95% intervals by 50-year period.
- How to read it
- Recall well below 1 means the cell omitted precipitation. Almost no 雨 on days without a gauge entry can be read as 'not invented', but the two records (gauge reports and the weather cell) come from the same scribes and are not independent evidence. Long 晴 runs and few July rain days mark reigns whose cell is furthest from observation.
- Caveats
- The 1.2 'false rain 1.7%' pooled zero-amount days (most of which carry a rain or snow token) with no-entry days in one denominator and was withdrawn in 1.2.1; 1.2.2 moved from node rows to calendar dates (72.4%→72.6%). 1.2.3 removed Lee's duplicate dates and corrected the node counts (the rates are unchanged). The meaning of 0 mm follows the wording of Lee et al. 2024 (trace) and was not checked cell by cell. The gauge record has its own gaps (Lee et al. 2024 flag columns) and its measurement day does not coincide exactly with the diary day. The comparison is limited to 1778–1907.
- External check
- Lee, Shin & Hong 2024: Seoul rain-gauge daily series 1778–1907.
- Reliability card
- externalExternal match — agreement with other datasets and observations
Dates matched to the gauge: 45,447 (45,632 weather nodes; 185 dates with two nodes, 12 of them disagreeing on rain — counted as rain if either node is). 雨 on 6,038 of 8,320 measured-rain days = 72.6% (95% 71.6%–73.5%); any rain, snow or hail token on 6,049, 雨 in Lee et al. 2024's string on 6,045 = 72.7%. Zero-amount days 989: rain, snow or hail token on 969; days without a gauge entry 36,138: 雨 on 1.
| Reign | Days | 晴 only | Longest 晴 run | Mean run | July days | July rain |
|---|---|---|---|---|---|---|
| Injo | 8,857 | 71.6% | 62 | 4.9 | 734 | 22.5% |
| Hyojong | 3,523 | 74.7% | 46 | 4.9 | 291 | 13.4% |
| Hyeonjong | 5,412 | 73.9% | 35 | 4.8 | 469 | 24.9% |
| Sukjong | 15,984 | 75.7% | 51 | 5.1 | 1,367 | 12.3% |
| Gyeongjong | 1,513 | 69.9% | 32 | 4.0 | 124 | 44.4% |
| Yeongjo | 19,358 | 68.5% | 37 | 4.0 | 1,589 | 41.4% |
| Jeongjo | 8,718 | 76.0% | 72 | 5.5 | 745 | 42.0% |
| Sunjo | 12,509 | 77.2% | 71 | 6.0 | 1,046 | 36.5% |
| Heonjong | 5,193 | 73.9% | 40 | 5.2 | 447 | 40.3% |
| Cheoljong | 5,108 | 80.1% | 86 | 6.8 | 431 | 42.0% |
| Gojong | 14,510 | 82.5% | 133 | 7.1 | 1,286 | 31.3% |
| Sunjong | 1,108 | 82.0% | 33 | 6.9 | 93 | 33.3% |
| Period | Matched days | Gauge rain days | With 雨 | Recall | 95% lo | 95% hi | Any precip token | 雨 in Lee's string | Gauge 0 | With precip token | No gauge entry | With 雨 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1750–1799 | 7,957 | 1,420 | 1,058 | 74.5% | 72.2% | 76.7% | 74.8% | 74.5% | 296 | 294 | 6,241 | 0 |
| 1800–1849 | 18,001 | 3,393 | 2,436 | 71.8% | 70.3% | 73.3% | 71.9% | 71.8% | 419 | 409 | 14,189 | 1 |
| 1850–1899 | 17,629 | 3,241 | 2,383 | 73.5% | 72.0% | 75.0% | 73.6% | 73.7% | 223 | 215 | 14,165 | 0 |
| 1900–1949 | 1,860 | 266 | 161 | 60.5% | 54.5% | 66.2% | 60.5% | 60.5% | 51 | 51 | 1,543 | 0 |
| 1778–1907 | 45,447 | 8,320 | 6,038 | 72.6% | 71.6% | 73.5% | 72.7% | 72.7% | 989 | 969 | 36,138 | 1 |