Rain Days, Storm Strength and the Open Air: Scheduled Weather Programmes and Residents' Reported Outdoor Time and Mood on Six Orbitals, 1844–1849 CE

1. Introduction

An Orbital is a ring a few million kilometres across, spun for gravity and run by a Hub Mind, with a day and a night produced by its spin and its orbit. Within that frame the Hub has wide discretion over what happens in the air above the residents. Some Hubs keep long dry runs and send rain in short, heavy falls. Others schedule frequent light rain. A few permit storms of real violence, and many do not. How a Hub shapes the seasons is decided by the Hub, often after asking its residents, and no authority above the Orbital sets a common pattern (Garrowby, 1812 CE).

The designers of such programmes have long held views about what they do to the people beneath them. Hollisward (1788 CE) argued that frequent rain keeps residents indoors without anyone intending it, while Vesselrode (1841 CE) argued that violent storms are felt as an intrusion even by residents who say they enjoy them. Neither argument has been tested against records from more than one Orbital. Residents take different amounts of outdoor time for many reasons, since nobody on an Orbital needs to be outside for any purpose but their own (Tarnimont, 1796 CE), and mood has its own seasonal drivers, including the length and brightness of the day (Ormbeck, 1830 CE).

This paper is written in 1850 CE and uses records that end with the last season of 1849 CE. It asks three questions of six Orbitals. Do Orbital-seasons with more scheduled rain days show less reported outdoor time? Do seasons with stronger programmed storms show lower reported mood? And do seasons whose timing shifted more from the previous year differ in either respect?

Two of us design weather programmes on Ossiline Orbital, and the third studies the ecology of Chiark Orbital. We say so here because both Orbitals are among the six, and because the programme descriptions we used were supplied by the Hubs and by the designers who work with them.

2. Methods

Six Orbitals took part, each by invitation of its Hub: Ossiline (Orbital A), Chiark (Orbital B) and four others, which we call C to F because their Hubs asked not to be named. We do not claim that the six stand for Orbitals in general. For each we compiled the Hub's weather programme for every 90-day season from the first season of 1844 CE to the last of 1849 CE, 24 seasons per Orbital and 144 Orbital-seasons altogether. Seasons of 90 days are our unit of analysis, not a division the Orbitals impose: every Hub sets its own seasonal calendar, and we cut each programme into blocks of 90 consecutive days. The compiled programmes and diaries are deposited with Chiark Orbital Hub Records as series OW-1 (1844–1849 CE).

We describe each programme by three measures. All three are our own constructions for this study, not standard quantities in Orbital engineering. Rain days is the number of days in the season on which the programme scheduled any rain. Storm intensity is an index from 0 to 10. Two of us, Irvandel and Tchaltane, scored every programmed storm independently from the peak wind and rain rate in the programme, and the season value is the mean of the scores of its storms (every season had at least one). Seasonal drift is the number of days by which the programmed onset of the season, as the Hub defines it, differed from the onset of the same season in the Orbital's previous year, taken without sign. Since we could not compute drift for the first year, we used each Hub's earlier schedule for 1843 CE.

Both scorers also assigned each season to one of three storm bands, low, moderate or high, so that we could check their agreement. They assigned the same band in 127 of 144 seasons (88%; Cohen's κ = 0.82), and the remaining 17 were settled by discussion before the index values were fixed.

Residents' reports came from opt-in diary panels. In each season, each Hub invited residents by general notice to keep a seven-day diary during the middle of the season, and those who accepted recorded hours spent outdoors each day and rated their mood three times a day on three seven-point items, which we averaged into a single score. A diary of this kind relies on recall and on the participant's own sense of what counts as being outdoors. Daffrenne (1835 CE) compared such reports with observed presence on Chiark and found that the two agreed closely on average although individuals varied. In a validation sample of 300 respondents drawn from several seasons, the three mood items had a Cronbach's α of 0.74, adequate for group means but not for judgements about individuals. The panels held 38 to 130 people per Orbital-season (mean 85) and 12,219 participant-seasons in all. The Hubs do not link diaries across seasons, in keeping with ordinary practice on Orbitals, so we cannot say how many different people contributed and treat each Orbital-season mean as one observation.

The outcome variables are the season mean of daily outdoor hours and the season mean of the mood score. Both are continuous, so we used linear regression with a separate intercept for each Orbital, which removes every stable difference between Orbitals, including their climates and their panels' habits. The three weather measures entered together, scaled to ten rain days, one index point and ten days of drift. Confidence intervals and p-values use the usual t distribution with 135 residual degrees of freedom. As checks, we repeated both models weighted by panel size and added a squared term for rain days to the outdoor model. The 24 seasons of one Orbital are not independent in the way the model assumes, and we return to this under Limitations.

3. Results

Programmes differed widely among the six Orbitals (Table 1). Rain days per season averaged 21.8 (SD 7.3, range 5 to 42) across all 144 Orbital-seasons, from 12.8 on Orbital A to 29.9 on Orbital E. Storm intensity averaged 3.77 (SD 1.37, range 0.5 to 7.0) and drift 6.5 days (SD 4.7, range 0.1 to 21.4). Reported outdoor time averaged 3.15 hours a day (SD 0.43) and mood 4.91 on the seven-point scale (SD 0.36). Orbitals with more rain and stronger storms had lower averages for both outcomes, but those comparisons among six Orbitals cannot separate the weather from every other way in which the Orbitals differ, which is why the models below use only variation within Orbitals.

Within Orbitals, rain days were the clearest correlate of outdoor time (Table 2). Ten more rain days in a season went with 0.30 fewer hours outdoors per day (95% CI 0.22 to 0.39 fewer, p < .001), which is about 18 minutes, with a plausible range of 13 to 23 minutes. A season with one standard deviation more rain days than usual, 7.3 days, was thus associated with about 13 minutes less outdoor time. Adding a squared term for rain days gave an estimate of −0.04 hours (p = .25), so we found no sign of curvature across the observed range.

Storm intensity was the clearest correlate of mood. Each additional point on the index went with 0.153 lower mood on the seven-point scale (95% CI 0.117 to 0.189 lower, p < .001). A standard deviation of the index, 1.37 points, therefore corresponded to about 0.21 scale points. Storm intensity was also weakly associated with outdoor time, at 0.04 fewer hours per point (95% CI 0.01 to 0.08 fewer, p = .02), and rain days were weakly associated with mood, at 0.084 lower per ten days (95% CI 0.002 to 0.166 lower, p = .04).

Seasonal drift was not clearly associated with either outcome. For outdoor time the estimate was 0.07 fewer hours per ten days of drift, with an interval that included no change (95% CI from 0.16 fewer to 0.02 more, p = .14). For mood it was 0.001 lower per ten days (95% CI from 0.088 lower to 0.087 higher, p = .99).

Weighting by panel size left the two main findings in place and moved the weaker ones. The rain-days estimate for outdoor time was 0.32 fewer hours per ten days (95% CI 0.23 to 0.41 fewer), and the storm estimate for mood was 0.152 lower per point (95% CI 0.115 to 0.189 lower). The storm estimate for outdoor time fell to 0.04 fewer hours per point with an interval that included no change (95% CI from 0.07 fewer to 0.003 more, p = .07). The rain-days estimate for mood remained small (0.098 lower per ten days, 95% CI 0.014 to 0.182 lower, p = .02). The drift estimate for outdoor time grew to 0.11 fewer hours per ten days, with an interval that excluded no change (95% CI 0.02 to 0.20 fewer, p = .02), a result absent from the unweighted model.

4. Discussion

Two findings are stable enough to build on. In these six Orbitals, seasons with more scheduled rain days had lower reported outdoor time, and seasons with stronger programmed storms had lower reported mood, in both cases within the same Orbital and under either weighting. The sizes are modest: a gain or loss of about a quarter of an hour of outdoor time for a typical swing in rain days, and about a fifth of a scale point of mood for a typical swing in storm strength. The within-Orbital estimates are modest in size.

The pattern fits the designers' views. Hollisward (1788 CE) expected frequent rain to cost outdoor time, and the rain-days finding supports that expectation for the Orbitals studied here. Vesselrode (1841 CE) expected strong storms to be felt as an intrusion, and a lower mood in stronger-storm seasons agrees with this, though our diaries do not say whether residents experienced the storms themselves or the days around them. We found no sign that the effect of rain on outdoor time bends over the range observed, so the figure of 18 minutes per ten rain days is a reasonable summary only within the range of 5 to 42 days that these Orbitals scheduled.

Weaker findings should not be read as effects. The link of storm intensity with outdoor time was small and changed with weighting, and that of rain days with mood was small and moved a little with weighting. The drift finding is more telling for what it fails to show: in the unweighted model, shifting a season's onset by ten days went with no clear change in either outcome, and the weighted outdoor-time result runs against that and needs repeating before it is believed. We would not advise a Hub to hold its seasons fixed on this evidence, and we would not advise it to move them.

None of this shows that a programme changes residents' behaviour. Hubs often choose weather after asking residents what they want (Lissenhal, 1846 CE), and residents who prefer to stay indoors may ask for rain. The designers on Ossiline sometimes lengthen a rainy run in a season when residents have asked for quiet, and we cannot separate that practice from the associations we found. What the records support is a description: where weather programmes ask for more rain or stronger storms, the people beneath them report less time outdoors and a lower mood. A Hub that wishes to know whether the programme is the cause could vary the weather in one district while holding it in another, with the residents' knowledge and consent.

5. Limitations

Six Orbitals joined by invitation, and four of them asked not to be named. Hubs that chose to share their schedules may differ from those that did not, and the panels were opt-in, so people who like to write about their days, or who spend more time at home, may be over-represented. Diary participation by choice also means we cannot tell how far a panel's mean stands for the Orbital's residents.

The three weather measures are our own and have not been used before. Rain days counts any scheduled rain regardless of its strength, and the storm index depends on two scorers' judgement of a programme's peak figures. Two of us, who also design weather programmes, scored every programme and were not blind to the Orbital or its Hub, so the index may carry our own expectations. The scorers agreed closely on bands, but that agreement is no guarantee that residents would rank storms as they did. The mood score rests on three self-rated items whose reliability suits group means only.

Each Orbital contributes 24 consecutive seasons that are likely to resemble one another, so the observations within an Orbital are not independent and the stated intervals are probably somewhat too narrow. With only six Orbitals, we could not estimate how the associations vary between them, and we cannot rule out that one Orbital drives a given result. The estimates for storms and outdoor time, rain and mood, and drift and outdoor time moved with weighting and should be treated as unsettled. The models contain no term for the position of a season within the year or for a time trend, so a seasonal rhythm shared by weather and mood would be absorbed into the weather coefficients.

Finally, the diaries were kept in the middle of each season, and the rain days and storms of that season were not necessarily those the participants lived through that week. The measures of weather and of behaviour are therefore matched in time only loosely, and a design that records weather and diaries day by day would answer more precisely than ours.