QUANTIFYING THE -SHADOW EFFECT Results from the Peak District, British Isles

Alexander J. Stockham, David M. Schultz, Jonathan G. Fairman Jr., and Adam P. Draude

How can the definition of be applied with quantitative data?

ountains affect the distribution of and meteorology textbooks (e.g., Marshak 2008, p. 733; . For example, Houze (2012, his Ahrens 2009, p. 156; Ackerman and Knox 2015, 136– M Fig. 3) showed a variety of physical mechanisms 137; Petersen et al. 2017, p. 162), as well as the Glossary for orographically enhanced precipitation. Many of Meteorology (Huschke 1959; Glickman 2000), as ranges commonly show large gradients being caused by ascending air and precipitation on in precipitation across them: the , , the windward side of a followed by Cascade of State, Sierra descending air and a reduction in precipitation on of , and the as the leeward side (i.e., the shadow). a whole. The most common explanation for these Yet, the rain-shadow effect does not appear widely gradients is rain (or precipitation) shadowing. Rain in the scientific literature. A search through the shadows are described in geography, climatology, and American Meteorological Society’s (AMS) Journals Online website turns up only four articles with “rain shadow(s)” or “precipitation shadow(s)” in the title: AFFILIATIONS: Stockham*—University of , Sheffield, Brady and Waldstreicher (2001), Ralph et al. (2003), Siler United Kingdom; Schultz and Fairman—Centre for Atmospheric et al. (2013), and Siler and Durran (2016). Expanding to Science, School of Earth and Environmental Sciences, University other publishers through the Web of Science, only 27 of , Manchester, United Kingdom; Draude—School of more articles have “rain shadow” in the title, mostly just Physics and Astronomy, University of Manchester, Manchester, using the term, pertaining to paleoclimates, stable iso- United Kingdom * Current affiliation: IN-PART Publishing Ltd., London, topes in precipitation, geomorphology, or . Only United Kingdom three pertain to the dynamics and microphysics CORRESPONDING AUTHOR: Prof. David M. Schultz, of the rain shadow. Expanding the search to the more [email protected] general topic “orographic precipitation” shows that The abstract for this article can be found in this issue, following the table most articles focus on the enhancement of precipita- of contents. tion on the windward side, rather than the reduction of DOI:10.1175/BAMS-D-17-0256.1 precipitation on the leeward side. Even the most recent

In final form 17 September 2017 summary of mountain weather and forecasting only ©2018 American Meteorological Society has five brief citations to precipitation shadows within its 750 pages (Chow et al. 2013). This article is licensed under a Creative Commons Attribution 4.0 license. At first glance, an explanation for the rain shadow is easily demonstrated and intellectually satisfying:

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 777 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC air goes up on the windward side producing clouds received 4 mm of rain and a leeward site received and precipitation, and air goes down on the leeward 3 mm of rain, would that qualify as a rain shadow? side inhibiting clouds and precipitation. Upon deeper 4) What if there were less precipitation on the lee reflection, however, such an explanation may not be side for reasons other than the classic up–down so obvious, especially when attempting to quantify mechanism? In other words, the physical mecha- the magnitude of the rain-shadow effect for a given nism for the rain shadow is not explicitly stated location on different time scales. in the definition. For example, Smith et al. (2009) found that the water lost as a result of precipita- WHAT IS A RAIN SHADOW AND HOW tion during orographic ascent over the island of CAN IT BE QUANTIFIED? The Glossary of Dominica was insufficient to explain the lesser Meteorology website defines a rain shadow as “A climatological precipitation in the lee. Instead, region of sharply reduced precipitation on the lee the shallower trade- inversion and reduced side of an orographic barrier, as compared with instability on the lee side were responsible for the regions upwind of the barrier” (http://glossary reduced precipitation. Is this still a rain shadow? .ametsoc.org/wiki/Rain_shadow). How to quantify the rain-shadow effect from this definition poses These questions raise the specter that the seemingly several problems. obvious definition is challenging to apply when faced with real data, a point we explore next. 1) No time scale is mentioned in this definition. Is this definition applied instantaneously within THE PEAK DISTRICT AND ITS RAIN a storm, applied to storm-total precipitation, or SHADOW. To further this point about the defini- applied to climatological data? Or, is the time scale tion lacking quantification, consider one location irrelevant for the application of this definition? that has been recognized as a classic rain-shadow Certainly the published literature refers to rain locality: the Peak District of the Pennines range, shadows at all time scales, from instantaneous United Kingdom (e.g., Garnett 1956; Milner 1968; radar imagery (e.g., Brady and Waldstreicher Chaun and Lockwood 1974; Wheeler 1990, 2013). 2001) to storm-total precipitation (e.g., Ralph et al. Wheeler (2013) writes, “One term alone dominates 2003; Sindosi et al. 2015) to average seasonal and the precipitation patterns of [northeast England]: annual precipitation amounts (e.g., Nieto Ferreira ‘rain shadow’.” A large east–west gradient in annual- et al. 2013; Kenworthy 2014; Lenaerts et al. 2014; average precipitation extends across the United Narkhedkar et al. 2015) to geological time scales Kingdom and Ireland (Fig. 1), consistent with the (e.g., Galewsky 2009). Such diversity of usage dominant westerly flow associated with classic wet would imply that there is no agreement on the British weather. Manchester, which is west of the necessity of a time scale to classify a reduction in Peak District, receives about 1,200 mm of rain each precipitation across a mountain range as a rain year. Fifty-six kilometers to the east at Sheffield, about shadow. 700 mm of rain falls each year. Although Manchester 2) What about the spatial scale of the reduction of and Sheffield city centers are both about 30 m above precipitation? How far away from the mountain sea level, the majority of the intervening high terrain crest does the rain-shadow effect extend? It would of the Peak District lies 300 m above sea level, with seem that a rain shadow would have to persist the highest point, Kinder Scout, at 636 m above sea farther downstream than just the immediate level. Although some rain shadows, such as those of descent in the lee of the mountains, although this the Andes and Himalayas, may feature an order of is not explicitly stated. An extensive minimum magnitude decrease in precipitation across them (e.g., in precipitation downstream of the mountain Barros et al. 2006; Viale and Nuñez 2011; Lenaerts would require mountain-wave breaking and the et al. 2014), the Peak District only experiences about entrainment of dry air aloft, which would be a a 40% reduction. Although the Peak District has function of the height of the mountains, the static modest elevations in comparison to mountain ranges stability profile, and the moisture profile. As such, that host more extreme rain shadows, a persistent applying any distance criteria might be a func- westerly flow and large gradient in precipitation tion of the specific mountain range or synoptic across the Peak District would suggest a strong rain- pattern. shadow effect. 3) What does “sharply reduced” mean when trying For example, a classic rain-shadow event occurred to quantify the rain shadow? If a windward site on the morning of 5 December 2015 (Fig. 2). On

778 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC this day, west-southwesterly flow brought rain to these weak rain shadows was the reduced leeside northwestern England and Scotland for many hours mountain waves caused by the low-level stable air (e.g., Fig. 2a), but little precipitation fell on the east that precedes the warm front. side of the Pennines, with the strongest gradient Second, precipitating systems sometimes weaken in 10-h accumulated precipitation lying north of as they move eastward as a result of the weakening of Manchester and Sheffield (Fig. 2b). The repeated forcing for ascent on the synoptic scale or mesoscale, occurrence of this pattern over many precipitation rather than the systems weakening due to orographic events could easily explain the gradient in precipi- forcing. The United Kingdom being near the clima- tation across the United Kingdom. However, even tological jet exit region makes such an explanation a cursory look at weather events passing over the possible in some cases, albeit at a larger horizontal United Kingdom readily shows that not all precipita- scale than the cross-mountain scale. Thus, even if tion events exhibit a rain shadow. less precipitation occurred in the lee, the rain shadow First, not all weather events may be associated may not necessarily explain it. with reduced precipitation in the lee. Sometimes, Third, the airflow over the mountains may not be precipitating systems show little change in intensity captured as a simple flow up and over the mountains. as they pass over the Peak District (e.g., Jaroszweski Stable low-level flow on the upwind side may be et al. 2015). In other cases, precipitation may blocked and not traverse the mountains, as reviewed increase toward the east. For example, westerly flow during the warm season often exhibits growing con- vective storms from west to east, producing more pre- cipitation toward the east (e.g., Thielen and Gadian 1996; Bennett et al. 2006). Even for a single mountain range under westerly flow, there may be substantial along-mountain variability of the rain-shadow effect (e.g., Viale and Nuñez 2011; Shi and Durran 2015), and the magnitude of the rain- shadow effect may depend on the synoptic pattern (e.g., Browning et al. 1975; Sweeney and O’Hare 1992; Siler et al. 2013; Mass et al. 2015; Siler and Durran 2015, 2016). For example, Siler et al. (2013) and Mass et al. (2015) showed that storms with a strong rain- shadow effect were associ- ated with precipitation in the warm sector, whereas storms with a weak rain- shadow effect were associ- ated with the passage of warm and occluded fronts.

Siler and Durran (2016) Fig. 1. Annual-average radar-derived precipitation (mm) over the British Isles found that the reason for between 2006 and 2013 [modified from Fairman et al. (2015)].

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 779 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC by Houze (2012, section 3.5) and Colle et al. (2013). those on the west side of the mountains. In another The downstream impact of the leeside descent may example, Sato (2005) showed that regions down- be enhanced by wave breaking over the terrain and stream of the TianShan Mountains were present, even entrainment of dry air, enhancing the downstream without the upstream mountains, suggesting no rain- length of the rain shadow. More importantly, different shadow effect there. Instead, Sato (2005) found that air masses may bring different weather to the west what made the strong gradient remarkable was not the and east sides of the Peak District. Specifically, the leeside reduction of precipitation, but the windward eastern United Kingdom may be more susceptible enhancement of precipitation. Should this situation to dry, but snowy, air masses originating from the correctly be called a rain shadow? European continent in winter (e.g., Ogden 1997; Pike All of these scenarios for how weather systems 1999) and more unstable air masses from Europe evolve over the United Kingdom differ from the during summer (e.g., Lewis and Gray 2010). In con- classic rain shadow with a uniform steady westerly trast, the western United Kingdom is more susceptible flow impinging on the Peak District and wringing to moister air masses originating from the North moisture out during orographic ascent, leaving Atlantic . So, any precipitation gradients across drying in its lee. Thus, these different scenarios the mountains may simply be because of access to raise the question of how important the rain-shadow different air masses. As another example, consider effect is to explaining the climatological distribution the Rocky Mountains. Although a rain shadow is of precipitation across the United Kingdom (Fig. 1). the traditional explana- tion for the dryness in the western Great Plains (e.g., Rosenberg 1987), this may not explain all of the re- duction in precipitation over central North America (e.g., Harrington 2008). Flow in the lee over cen- tral North America may experience either moist air because of return flow from the Gulf of (e.g., Crisp and Lewis 1992; Weiss 1992) or dry air because of cold-air incur- sions from the Arctic (e.g., Schultz et al. 1997, 1998) compared to the moister air masses originating from the North Pacific Ocean and making landfall on the west coast of North America. Again, the lee side is often under the influence of different air masses than

Fig. 2. Radar mosaics showing (a) radar rainfall rates (mm h −1) at 0815 UTC 5 Dec 2015 and 850-hPa geopotential height at 1200 UTC and (b) radar- derived 10-h accumulated pre- cipitation during 0800–1800 UTC 5 Dec 2015.

780 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC Does the rain shadow occur during all precipitation weather stations within 20-km radii of the Sheffield events? If not, how important is it, and when does it and Manchester city centers, recorded over the 30-yr occur? How many days of the year is the rain-shadow period 1 January 1981–31 December 2010 (Fig. 3). The effect actually operating to explain the climatological daily 24-h period is from 0000 to 2359 UTC. These distribution of rain across England? This study repre- data are obtained from the Met Office’s Integrated sents an attempt using precipitation-gauge statistics Data Archive System (MIDAS). To establish the geo- to quantify the rain-shadow effect and to provide an graphic distribution of precipitation across the Peak answer to the public about how important the rain District, each of the 54 weather stations is identified shadow actually is. Thus, the purpose of this study is as being located in one of four regions: Manchester (25 to critically examine the definition of a rain shadow, stations), Sheffield (11 stations), the Manchester side of highlighting the ambiguity in its definition and the the Peak District (7 stations), or the Sheffield side of the ambiguity in providing quantitative information to Peak District (11 stations) (Fig. 3). The amount of pre- evaluate the rain-shadow effect using the example cipitation from each station in each region is averaged of the Peak District. This study does not address the together to produce a regional average for each day. physical mechanisms for the rain shadow. These 54 stations were determined because they had over 5000 observations in 30 yr (45.7% of the total DATA AND METHODS. The dataset consists number of days). We chose 5000 observations because of the daily precipitation amount (mm) from 54 some stations were not operating or reporting during the whole 30-yr period. If we limit the dataset to sta- tions where 90% or more of possible observations existed over the 30 yr, then we would have only 11 sta- tions in Manchester and 4 in Sheffield, too few to produce a reliable measure of whether the lee side had experienced a region of widespread rain or not. Importantly, when a station was operating, the dataset indicated nearly perfect re- porting from that station. By lowering the threshold of station longevity, we were able to include more stations into our measure of how widespread the precipitation was or was not. Such an indication ensures that our results are representative and not merely a result of incom- plete sampling due to a small number of stations. The averaging of the pre- cipitation amount into four regions further ensures that the data are robust, despite some missing station data. The prevailing wind direction is determined

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 781 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC from the 850-hPa wind direction from the European westerly on British weather. In contrast, only 8% Centre for Medium-Range Weather Forecasts of the days (28 days yr−1) have easterly flow. (ECMWF) interim reanalysis (ERA-Interim; Dee To test the possible occurrence of a rain shadow, et al. 2011) at 1200 UTC for each day. The 850-hPa level we perform three tests. equates to about 1.4–1.5 km, which is above the crest height of the Peak District. Westerly days are classified Test 1: Does it rain more on the upwind side of the Peak by a wind direction of 225°–315° at the closest grid point District? The first test is to compare the daily average to Manchester (53.25°N, 2.25°W), and easterly days are precipitation amount for each region to establish the classified by a wind direction of 45°–135° at the closest distribution of precipitation across the Peak District grid point to Sheffield (53.25°N, 1.5°W). Westerly winds as a function of westerly or easterly flow. A distinct prevail over the Peak District on about 57% of the days pattern in the distribution of precipitation occurs un- (207 out of 365 days yr−1), indicating the dominance of der both westerly and easterly flow (Table 1). In each case, the lowest amount of precipitation is recorded in the city region on the lee side of the range, the high- est amount of precipitation is recorded in the region of the Peak District first ex- posed to the wind flow, and the second highest amount of rainfall is recorded in the city region that first encounters the prevailing wind direction (Table 1). The difference between re- gions of highest and lowest daily precipitation is great- est under westerly flow. When the flow is westerly, 1.5 mm (50%) more average daily precipitation occurs on the Manchester side of the Peak District compared to the Sheffield side. In comparison, under easterly flow, 0.8 mm (40%) more average daily precipitation occurs on the Sheffield side of the Peak District Fig. 3. A map of the Peak District region with weather stations included in the dataset, colored according to the four regions (red represents Manchester, compared to the Manches- blue represents Peak District Manchester side, green represents Peak District ter side. Overall, westerly Sheffield side, and black represents Sheffield). flows produce more daily

Table 1. Overall daily average precipitation amounts (mm) recorded under westerly and easterly flows over the 30-yr period (Jan 1981–Dec 2010).

Location Peak District Peak District Manchester Sheffield (Manchester) (Sheffield) Westerly flow 3.9 4.5 3.0 1.7 1.5 2.0 2.8 2.4 Easterly flow

782 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC mean precipitation across the domain. In isolation, Manchester under a prevailing westerly flow and an the results from this first test are consistent with the enhancement in the number of days of precipitation expectations of a rain-shadow effect for both westerly in the Peak District and Sheffield under a prevailing and easterly flow. easterly flow (Tables 2 and 3). Out of the annual average of 207 westerly flow days, precipitation Test 2: Does it rain more often on the upwind side of was recorded in Manchester on 154 occasions and the Peak District? Second, the number of days when in Sheffield on 121 occasions (Table 2), equating to 0.1 mm or more of the average daily precipitation was an annual difference of 33 days each year in which recorded across each region is examined to determine Manchester receives precipitation and Sheffield does whether there is an enhancement in the number not (16% of all westerly flow days). In comparison, of days of precipitation in the Peak District and under the 28 easterly flow days each year, there are

CHOICE OF PRECIPITATION DURATION

ecause the definition of a rain shadow rain on the lee side for 3 of those 6 h, in Manchester and dry in Sheffield, only B does not specify a time scale, using then the rain shadow would only have a dynamic analysis could ascertain the rain gauge data requires selecting a existed for 3 out of the 6 h, but would reason why. There is the added compli- time scale. So, what should be chosen? not be classified as a rain shadow using cation that as the data become hourly, One-hour precipitation amounts? our approach with 24-h data. So, our at least for the Pennines, the number of Twenty-four-hour precipitation amounts? calculation could be performed by stations decreases. The analysis could be Because precipitation occurrence is hour-long intervals by saying that x% performed, but it would be difficult to fractal (e.g., Olsson et al. 1993; Harris of hours with westerly flow qualify as a compare with the analysis from the 24-h et al. 1996; Kiely and Ivanova 1999), no rain shadow, where x ≤ 24 h. The 24-h data. Not that a rain shadow could not natural time scale exists upon which to values can be thought of as a conser- be defined for hourly data, but a convinc- perform this analysis. The actual usage vative limit of the percentage of time ing argument is more easily made that of the expression “rain shadow” in the with a rain shadow. Furthermore, our the rain shadow is more likely operating literature provides no insight into this statement that 17% of days have a rain with multihour accumulated rainfall. dilemma. Examples of rain shadows shadow for the 24-h period is correct Finally, the 24-h data were just meant being referred to on short time scales within the context of the question that to be representative of the type of calcu- exist (e.g., instantaneous radar imagery we have asked with the data we have lation that one could do to demonstrate as in our Fig. 2), as do rain shadows of used, which was about the number this point. A different threshold would storm-total precipitation (comparable to of days during which a possible rain produce different results to a different our 24-h data). Rain shadows can also be shadow was occurring. question with a different interpretation. inferred from precipitation climatologies To redo the analysis with 1-h data, Using a different interval would be a of rainfall from radar, satellite, or gauge the number of hours of rain shadows different study than what we intended. data (as in our Fig. 1). Thus, our choice of over the course of the year (from 1-h Although using a finer-resolved rainfall a 24-h rainfall dataset is not inconsistent data) would be more or less consistent data interval would produce a quantita- with the definition of rain shadow, and with the number of days of rain shadows tively different result, whether the result our choice is not inconsistent with how it per year (from 24-h data). There would would be substantially different is unclear. is used practically in the literature. be a small difference because many rain- We encourage readers to explore these Reviewers raised the concern shadow days with 24-h data would not issues with their own rainfall datasets whether our results would be differ- have rain shadows for all 24 h, whereas with different temporal resolutions for ent if more frequent rainfall data were other non-rain-shadow days would have other regions of the world. Thus, we used. Consider the following. If it some hours of hourly rain shadows. Thus, argue that the time resolution of the continuously for 24 h (0000–2359 we believe that the results may change rainfall is an open point for debate and UTC) on the upwind side and no rain quantitatively with hourly data, but likely any suitably defensible choice is entirely falls on the leeside, then all of the would not be qualitatively different. reasonable. twenty-four 1-h periods would qualify Does talking about the rain shadow Thus, we argue that performing this as rain shadows, too. Thus, using 24- hour by hour make any sense? The finer calculation with 1-h rain gauge data or 1-h rainfall data would yield no the increment of time, the more chal- would not yield qualitatively different difference. Of course, it may only rain lenging determining the existence of a results. What is optimal? Three hours? for 6 h within that 24-h period such rain shadow becomes from point rainfall Six hours? There is no single best that identifying a rain shadow with measurements, especially when scattered answer. We hope that discussing these 1-h data would determine that 6 h of showers are occurring. Does a rain issues out in the open will be beneficial the day were associated with a rain shadow even have any meaning under to the overall discussion of how to shadow. Alternatively, if it were to this environment? Even if it were raining identify and quantify a rain shadow.

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 783 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC on average 2 days each year in which precipitation is day were used, reducing the size of the dataset recorded in Sheffield but not in Manchester (7% of slightly (i.e., 207 days yr−1 of westerly flow down to all easterly flow days) (Table 3). Thus, as measured 197 days yr−1, and 28 days yr−1 of easterly flow down by the frequency of precipitation, the results of the to 26 days yr−1). second test seem to support the idea that the rain For this analysis, we define a rain shadow as a day shadow (as defined in this study) is not particularly where it rained on the windward-side stations, but not common in westerly flow and is even rarer in easterly on the leeward-side stations. This definition is con- flow (cf. 33 vs 2 days yr−1). sistent with the simple conceptual models of the rain Tests 1 and 2 are the types of bulk precipitation shadow found in textbooks that show precipitation on amount and frequency statistics that are often em- the windward side and none on the leeward side (e.g., ployed to explain rain shadows. For example, these Marshak 2008, p. 733; Ahrens 2009, p. 156; Ackerman bulk climatological data are often used to explain the and Knox 2015, p. 137; Petersen et al. 2017, p. 162). rain-shadow effect from west to east across the United Plus, the absence of precipitation on the leeward-side Kingdom (e.g., Hill 1983; Sweeney and O’Hare 1992; stations would be characterized as a sharp reduction Hand 2005). of precipitation across the topography, consistent with the AMS . Of course, the Test 3: When it rains on the upwind side, is it also rain-shadow effect could still be operating with less raining on the downwind side? In a third—and more precipitation occurring downwind, perhaps caused rigorous—test of the rain-shadow effect, the number by so-called spillover precipitation [i.e., hydrometeor of individual days where a rain-shadow effect might drift over the crest onto the leeside; e.g., Sinclair et al. have been occurring was calculated. For this analysis, (1997); Colle (2004)]. We acknowledge that ours is a only daily rainfall data that were available for both restrictive definition of a rain shadow and that other the Manchester and Sheffield regions on a particular reasonable choices could have been made, but we adopt

Table 2. Precipitation occurrence in each of the four regions under westerly flow.

Avg rainy days per year under westerly flow (with precipitation day ≥ 0.1 mm across the region) Peak District Peak District Manchester Sheffield (Manchester) (Sheffield) No. of days when precipitation was recorded across the region 4,616 4,357 4,153 3,643 (1981–2010) Percentage of precipitation days 74% 70% 67% 58% under westerly flow each year Avg No. of days of precipitation 154 145 138 121 per year under westerly flow

Table 3. Precipitation occurrence under easterly flow across the Peak District.

Avg rainy days per year under easterly flow (with precipitation day ≥ 0.1 mm across the region) Peak District Peak District Manchester Sheffield (Manchester) (Sheffield) No. of days when precipitation was recorded across the region 340 349 433 384 (1981–2010) Percentage of precipitation days 41% 42% 51% 45% under easterly flow each year Avg No. of days of precipitation 11 12 14 13 per year under easterly flow

784 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC Table 4. Average number of days per year with or without rain in Manchester and Sheffield under westerly flow. Percentages out of the total number of 197 days appear in parentheses. Percentages do not add up to 100% because of round-off errors.

Manchester avg No. of days per year Manchester avg No. of days per year with zero precipitation amount with nonzero precipitation amount Sheffield avg No. of days per year 45 (22.8%) 34 (17.3%) with zero precipitation amount Sheffield avg No. of days per year 4.6 (2.3%) 113 (57.4%) with nonzero precipitation amount

this conservative approach as a starting point. Specifically, for a west- erly rain-shadow effect, we count the number of days with rain in Manchester but no rain in Sheffield. For an easterly rain-shadow effect, we count the number of days with rain in Sheffield but no rain in Manchester. The result is a count of the number of days where the rain-shadow effect could conceivably be occurring. Moreover, just because precipitation falls on the upstream location but none falls in the downstream location does not neces- sarily mean that the rain- shadow effect is responsible (as discussed earlier). For the 197 days yr−1 with westerly flow, 57% of all westerly flow days were associated with rain at both Manchester and Sheffield (Table 4). These days were most common during the cool season with a mini- mum in the spring (Fig. 4a). Only 17% of westerly flow days occurred where the rain-shadow effect could be operating (Manchester with rain and Sheffield Fig. 4. Monthly distribution of days per year under (a) westerly and (b) easterly flows. RShf represents rain in Sheffield, NRShf represents no rain with no rain). These days in Sheffield, NRMan represents no rain in Manchester, and RMan represents were concentrated in the rain in Manchester. In (a), the black bars represent days per year when the warm season (Fig. 4a). rain-shadow effect was possible. In (b), the gray bars represent days per year In contrast, only 2.3% of when the rain-shadow effect was possible.

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 785 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC westerly flow days showed no rain in Manchester under easterly flow were, given their tendency to and rain in Sheffield, a result opposite to the rain- appear at a higher rate than for reverse-rain-shadow shadow effect—what Roe and Baker (2006) call the days under westerly flow. reverse rain shadow. These results suggest that, for all Only 34 days a year possessed both westerly flow westerly flow days, the non-rain-shadow days were and the requisite precipitation pattern (Manchester nearly 5 times more likely than rain-shadow days with rain and Sheffield with no rain) to be consistent using our definition. with the rain-shadow effect. Although a rain-shadow For comparison, 36% of the 26 days yr−1 with effect can be seen on 34 days each year in which easterly flow were associated with rain at both Manchester receives rainfall and Sheffield does not, Manchester and Sheffield (Table 5). These days were this observation goes only part of the way toward maxima in the winter and spring (Fig. 4b). Only accounting for the regional differences in precipita- 10% of easterly flow days occurred where the rain- tion across the Peak District each year. shadow effect could be operating (Sheffield with rain and Manchester without rain). Interestingly, DOCUMENTARY. The findings of this analysis however, 4.2% of easterly flow days occurred with are incorporated into a documentary entitled Chasing rain in Manchester and no rain in Sheffield, a result Sheffield’s Rain Shadow, which is available on YouTube opposite to the rain-shadow effect. Thus, for all (www.youtube.com/watch?v=3eaVn7JQpOQ). This easterly flow days, the non-rain-shadow days were documentary was created in partial fulfilment of the 9 times more likely than rain-shadow days. A look lead author’s master of science dissertation in science through some of these easterly flow events indicates communication at the University of Sheffield. The that the synoptic pattern is typically characterized documentary explores the origins of Sheffield’s rain by a strong equivalent-barotropic low pressure sys- shadow and features interviews with meteorology tem to the south or west of the United Kingdom. experts from both sides of the Peak District, with Deep easterly flow is prevalent, and precipitation is the goal of engaging with and educating the public associated with bands wrapped around the cyclone. on the Sheffield rain shadow. Some of the quanti- In these situations, the precipitation is affected little tative information in the documentary has been by the . updated and double-checked for this publication In this third test of the rain-shadow effect, our since recording, but the results and storytelling are results were consistent with those from the second robust, nonetheless. test (Table 6). Specifically, westerly winds were more likely to produce precipitation than easterly winds TOWARD A DEEPER UNDERSTANDING (75% vs 46%), westerly winds were more likely to OF THE RAIN SHADOW. This research explores exhibit a rain shadow than easterly winds (17% vs how to quantify the impact that the rain shadow has 10%), and westerly winds were less likely to exhibit on individual precipitation events across the United a reverse rain shadow than easterly winds (2.3% vs Kingdom. Other analysis approaches could be rea- 9.1%). Although the ratio of rain-shadow days to sonably taken with different criteria chosen for the reverse-rain-shadow days was larger for westerly flow sharp reduction in precipitation on the leeside, and than easterly flow (7.4 vs 2.4), the ratio of days with the analysis could be extended to other regions both rain at both Manchester and Sheffield to rain-shadow north and south of the Peak District. Would they days in westerly and easterly flows was relatively also show similar relationships? Under what syn- similar (3.3 to 3.5). This result seems to suggest how optic conditions is the rain-shadow effect most and relatively common the reverse-rain-shadow days least prominent? For example, one might imagine

Table 5. Average number of days per year with or without rain in Manchester and Sheffield under easterly flow. Percentages out of the total number of 26 days appear in parentheses. Manchester avg No. of days per year Manchester avg No. of days per year with zero precipitation amount with nonzero precipitation amount Sheffield avg No. of days per year 13 (50%) 1.1 (4.2%) with zero precipitation amount Sheffield avg No. of days per year 2.6 (10%) 9.3 (35.8%) with nonzero precipitation amount

786 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC situations such as the floods of Table 6. Comparison of westerly and easterly flows Christmas 2015 (Barker et al. 2016) where widespread (annual-average values). rain on the synoptic scale showed little favoritism for Westerly Easterly one side of the Peak District over another. Indeed, Siler and Durran (2015) have argued that warm fronts Days 207 28 Days with rainfall data at are less likely to produce a strong rain shadow. 197 26 What would high-resolution mesoscale modeling Manchester and Sheffield Rainy days at Manchester show for rain-shadow events? Are there systematic 113 9.3 biases in numerical model output related to the rain- and Sheffield shadow effect (e.g., Colle et al. 2000)? Anecdotal Rainy days at upwind location 147 12 Percentage of days with rainy evidence from our real-time model simulations at 75% 46% ManUniCast.com (Schultz et al. 2015) suggests the upwind location rain-shadow effect is underrepresented in the model Days with rain shadow 34 2.6 output in some cases. Do leeside mountain waves pro- Days with reverse rain shadow 4.6 1.1 duce strong descent immediately to the lee of the Peak Ratio of rain-shadow days to 7.4 2.4 District that enhances the gradient in precipitation, reverse-rain-shadow days or does leeside stability affect the distribution of pre- Ratio of days with rain at cipitation (e.g., Brady and Waldstreicher 2001; Smith both Manchester and Sheffield 3.3 3.5 et al. 2009)? What roles do precipitation spillover and to rain-shadow days multiple ridges play in the rain-shadow effect (e.g., Percentage of rain-shadow days 17% 10% Sinclair et al. 1997; Colle 2004, 2008)? What is the Percentage of reverse- 2.3% 9.1% importance of upstream blocking (airmass transition rain-shadow days across mountains), gravity wave breaking, convec- tion, and vertical wind shear? Finally, what do other mountain ranges around the their comments on earlier versions of this manuscript. We world show? Other regions around the world show thank Chairman of the Editorial Board Jeff Waldstreicher more dramatic rain shadows than the Peak District, for his comments on rain shadows. and an intercomparison between the different ranges to quantify the different factors affecting the rain shadows in different geographical contexts would be REFERENCES an interesting application of this work, a point raised Ackerman, S. A., and J. A. Knox, 2015: Meteorology: by Barros and Lettenmaier (1994). Understanding the Atmosphere. 4th ed. Jones and These and other questions could form the basis for Bartlett Learning, 575 pp. future research on rain shadows. Such work would go Ahrens, C. D., 2009: Meteorology Today: An Introduc- a long way to exploring beyond the simple textbook tion to Weather, , and the Environment. 9th explanation for the weather patterns across the Peak ed. Brooks/Cole, 549 pp. District, as well as other mountain ranges. Barker, L., J. Hannaford, K. Muchan, S. Turner, and S. Parry, 2016: The winter 2015/2016 floods in the ACKNOWLEDGMENTS. This research formed part UK: A hydrological appraisal. Weather, 71, 324–333, of AS’s master of science dissertation in science com- https://doi.org/10.1002/wea.2822. munication at the University of Sheffield. We thank the Barros, A. P., and D. P. Lettenmaier, 1994: Dynamic Met Office and British Atmospheric Data Centre for the modeling of orographically induced precipi- MIDAS precipitation gauge data and the European Centre tation. Rev. Geophys., 32, 265–284, https://doi for Medium-Range Weather Forecasts for the ERA-Interim .org/0.1029/94RG00625. reanalysis. Partial funding for JGF and DMS comes from —, S. Chiao, T. J. Lang, D. Burbank, and J. Putkonen, the U.K. Natural Environment Research Council to the 2006: From weather to climate—Seasonal and University of Manchester for the Precipitation Structures interannual variability of storms and implications over Orography project (PRESTO; Grant NE/I026545/1). for erosion processes in the Himalaya. Tectonics, Funding for APD was provided by the Natural Environment Climate, and Evolution, S. D. Willet et al., Research Council through the Manchester–Liverpool Eds., Geological Society of America Special Papers, Doctoral Training Programme, Grant NE/L002469/1. We Vol. 398, Geological Society of America, 17–38. thank the three anonymous reviewers and Editor Ed Zipser, Bennett, L. J., K. A. Browning, A. M. Blyth, D. J. Parker, as well as James Steenburgh and Daniel Kirshbaum, for and P. A. Clark, 2006: A review of the initiation of

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 787 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC precipitating convection in the United Kingdom. Galewsky, J., 2009: Rain shadow development during Quart. J. Roy. Meteor. Soc., 132, 1001–1020, https:// the growth of mountain ranges: An atmospheric doi.org/10.1256/qj.05.54. dynamics perspective. J. Geophys. Res., 114, F01018, Brady, R., and J. Waldstreicher, 2001: Observations https://doi.org/10.1029/2008JF001085. of mountain wave–induced precipitation shadows Garnett, A., 1956: Climate. Sheffield and Its Region: A over northeast Pennsylvania. Wea. Forecasting, Scientific and Historical Survey, D. L. Linton, Ed., 16, 281–300, https://doi.org/10.1175/1520-0434 British Association for the Advancement of Science, (2001)016<0281:OOMWIP>2.0.CO;2. 44–89. Browning, K. A., C. W. Pardoe, and F. F. Hill, 1975: The Glickman, T. S., Ed., 2000: Glossary of Meteorology. 2d nature of orographic rain at wintertime cold fronts. ed. Amer. Meteor. Soc., 855 pp. Quart. J. Roy. Meteor. Soc., 101, 333–352, https://doi. Hand, W. H., 2005: Climatology of shower frequency org/10.1002/qj.49710142815. in the British Isles at 5 km resolution. Weather, 60, Chaun, G. K., and J. G. Lockwood, 1974: An assess- 153–158, https://doi.org/10.1256/wea.129.04. ment of topographical controls in the distribution Harrington, J., Jr., 2008: Misconceptions: Barriers to of rainfall in the central Pennines. Meteor. Mag., improved climate literacy. Phys. Geogr., 29, 575–584, 103, 275–287. https://doi.org/10.2747/0272-3646.29.6.575. Chow, F. K., S. F. J. De Wekker, and B. J. Snyder, Eds., Harris, D., M. Menabde, A. Seed, and G. Austin, 2013: Mountain Weather Research and Forecasting: 1996: Multifractal characterization of rain fields Recent Progress and Current Challenges. Springer, with a strong orographic influence. J. Geophys. 750 pp. Res., 101, 26 405–26 414, https://doi.org/10.1029 Colle, B., 2004: Sensitivity of orographic precipita- /96JD01656. tion to changing ambient conditions and terrain Hill, F. F., 1983: The use of average annual rainfall geometries: An idealized modeling perspective. J. to derive estimates of orographic enhancement of Atmos. Sci., 61, 588–606, https://doi.org/10.1175/1520 frontal rain over England and for different -0469(2004)061<0588:SOOPTC>2.0.CO;2. wind directions. J. Climatol., 3, 113–129, https://doi —, 2008: Two-dimensional idealized simulations .org/10.1002/joc.3370030202. of the impact of multiple windward ridges on oro- Houze, R. A., Jr., 2012: Orographic effects on precipi- graphic precipitation. J. Atmos. Sci., 65, 509–523, tating clouds. Rev. Geophys., 50, RG1001, https://doi https://doi.org/10.1175/2007JAS2305.1. .org/10.1029/2011RG000365. —, C. Mass, and K. Westrick, 2000: MM5 pre- Huschke, R. E., Ed., 1959: Glossary of Meteorology. cipitation verification over the Amer. Meteor. Soc., 638 pp. during the 1997–99 cool seasons. Wea. Forecasting, Jaroszweski, D., E. Hooper, C. Baker, L. Chapman, and 15, 730–744, https://doi.org/10.1175/1520-0434 A. Quinn, 2015: The impacts of the 28 June 2012 (2000)015<0730:MPVOTP>2.0.CO;2. storms on UK road and rail transport. Meteor. Appl., —, R. B. Smith, and D. A. Wesley, 2013: Theory, 22, 470–476, https://doi.org/10.1002/met.1477. observations, and predictions of orographic precipi- Kenworthy, J. M., 2014: Regional weather and tation. Mountain Weather Research and Forecasting: of the British Isles—Part 7: North west England Recent Progress and Current Challenges, F. K. Chow, and the Isle of Man. Weather, 69, 87–93, https://doi S. F. J. De Wekker, and B. J. Snyder, Eds., Springer, .org/10.1002/wea.2256. 291–344. Kiely, G., and K. Ivanova, 1999: Multifractal analysis of Crisp, C. A., and J. M. Lewis, 1992: Return flow in the hourly precipitation. Phys. Chem. Earth, 24B, 781– Gulf of Mexico. Part I: A classificatory approach 786, https://doi.org/10.1016/S1464-1909(99)00080-5. with a global historical perspective. J. Appl. Meteor., Lenaerts, J. T. M., M. R. Van den Broeke, J. M. Van 31, 868–881, https://doi.org/10.1175/1520-0450 Wessem, W. J. Van de Berg, E. Van Meijgaard, L. H. (1992)031<0868:RFITGO>2.0.CO;2. Van Ulft, and M. Schaefer, 2014: Extreme precipita- Dee, D. P., and Coauthors, 2011: The ERA-Interim tion and climate gradients in revealed reanalysis: Configuration and performance of the by high-resolution regional atmospheric climate data assimilation system. Quart. J. Roy. Meteor. Soc., modeling. J. Climate, 27, 4607–4621, https://doi.org 137, 553–597, https://doi.org/10.1002/qj.828. /10.1175/JCLI-D-13-00579.1. Fairman, J. G., Jr., D. M. Schultz, D. J. Kirshbaum, S. L. Lewis, M. W., and S. L. Gray, 2010: Categorisation of Gray, and A. I. Barrett, 2015: A radar-based rainfall synoptic environments associated with mesoscale climatology of Great Britain and Ireland. Weather, convective systems over the UK. Atmos. Res., 97, 194– 70, 153–158, https://doi.org/10.1002/wea.2486. 213, https://doi.org/10.1016/j.atmosres.2010.04.001.

788 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC Marshak, S., 2008: Earth: Portrait of a Planet. 3rd ed. The 1993 Superstorm cold surge: Frontal structure, Norton, 832 pp. gap flow, and tropical impact. Mon. Wea. Rev., 125, Mass, C., N. Johnson, M. Warner, and R. Vargas, 2015: 5–39; Corrigendum, 125, 662. Synoptic control of cross-barrier precipitation ratios —, —, and —, 1998: Planetary- and synoptic- for the Cascade Mountains. J. Hydrometeor., 16, scale signatures associated with Central American 1014–1028, https://doi.org/10.1175/JHM-D-14-0149.1. cold surges. Mon. Wea. Rev., 126, 5–27, https:// doi Milner, J. S., 1968: Fluctuations in annual and sea- .org/10.1175/1520-0493(1998)126<0005:PASSSA sonal rainfall over the leeside of the south Pennines, >2.0.CO;2. 1761–1965. Weather, 23, 435–441, https://doi.org —, S. Anderson, J. G. Fairman Jr., D. Lowe, G. /10.1002/j.1477-8696.1968.tb03019.x. McFiggans, E. Lee, and R. Seo-Zindy, 2015: Narkhedkar, S. G., S. B. Morwal, B. Padmakumari, C. G. ManUniCast: A real-time weather and air-quality Deshpande, D. R. Kothawale, R. S. Maheskumar, forecasting portal and app for teaching. Weather, 70, and J. R. Kulkarni, 2015: Rainfall mechanism over 180–186, https://doi.org/10.1002/wea.2468. the rain-shadow region of north peninsular . Shi, X., and D. R. Durran, 2015: Estimating the response Climate Dyn., 45, 1493–1512, https://doi.org/10.1007 of extreme precipitation over midlatitude mountains /s00382-014-2403-2. due to global warming. J. Climate, 28, 4246–4262, Nieto Ferreira, R., L. Hall, and T. M. Rickenbach, https://doi.org/10.1175/JCLI-D-14-00750.1. 2013: A climatology of the structure, evolution, and Siler, N., and D. Durran, 2015: Assessing the impact of propagation of midlatitude cyclones in the southeast the tropopause on mountain waves and orographic . J. Climate, 26, 8406–8421, https://doi precipitation using linear theory and numerical .org/10.1175/JCLI-D-12-00657.1. simulations. J. Atmos. Sci., 72, 803–820, https://doi Ogden, R. J., 1997: When the wind is in the east. Weather, .org/10.1175/JAS-D-14-0200.1. 52, 322–324, https://doi.org/10.1002/j.1477-8696.1997 —, and —, 2016: What causes weak orographic .tb05529.x. rain shadows? Insights from case studies in the Olsson, J., J. Niemczynowicz, and R. Berndtsson, 1993: Cascades and idealized simulations. J. Atmos. Sci., Fractal analysis of high-resolution rainfall time 73, 4077–4099, https://doi.org/10.1175/JAS-D-15 series. J. Geophys. Res., 98, 23 265–23 274, https://doi -0371.1. .org/10.1029/93JD02658. —, G. Roe, and D. Durran, 2013: On the dynamical Petersen, J. F., D. Sack, and R. E. Gabler, 2017: Physical causes of variability in the rain-shadow effect: Geography. 11th ed. Cengage Learning, 645 pp. A case study of the Washington Cascades. J. Pike, W. S., 1999: Not fit for man nor beast. Weather, Hydrometeor., 14, 122–139, https://doi.org/10.1175 54, 394–401, https://doi.org/10.1002/j.1477-8696.1999 /JHM-D-12-045.1. .tb04001.x. Sinclair, M. R., D. S. Wratt, R. D. Henderson, and Ralph, F., P. Neiman, D. Kingsmill, P. Persson, A. W. R. Gray, 1997: Factors affecting the distribu- White, E. Strem, E. Andrews, and R. Antweiler, tion and spillover of precipitation in the Southern 2003: The impact of a prominent rain shadow on of —A case study. J. Appl. flooding in California’s : A Meteor., 36, 428–442, https://doi.org/10.1175/1520 CALJET case study and sensitivity to the ENSO -0450(1997)036<0428:FATDAS>2.0.CO;2. cycle. J. Hydrometeor., 4, 1243–1264, https://doi Sindosi, O. A., A. Bartzokas, V. Kotroni, and K. Lagou- .org/10.1175/1525-7541(2003)004<1243:TIOAPR vardos, 2015: Influence of orography on precipitation >2.0.CO;2. amount and distribution in NW ; a case study. Roe, G. H., and M. B. Baker, 2006: Microphysical and Atmos. Res., 152, 105–122, https://doi.org/10.1016/j geometrical controls on the pattern of orographic .atmosres.2014.06.013. precipitation. J. Atmos. Sci., 63, 861–880, https://doi Smith, R. B., P. Schafer, D. J. Kirshbaum, and E. Regina, .org/10.1175/JAS3619.1. 2009: Orographic precipitation in the : Rosenberg, N. J., 1987: Climate of the Great Plains region Experiments in Dominica. J. Atmos. Sci., 66, 1698– of the United States. Great Plains Quart., 7, 22–32. 1716, https://doi.org/10.1175/2008JAS2920.1. Sato, T., 2005: The TianShan rain-shadow influence on Sweeney, J. C., and G. P. O’Hare, 1992: Geographical the arid climate formation in northwestern China. variations in precipitation yields and circulation SOLA, 1, 13–16, https://doi.org/10.2151/sola.2005 types in Britain and Ireland. Trans. Inst. Br. Geogr., -004. 17, 448–463, https://doi.org/10.2307/622710. Schultz, D. M., W. E. Bracken, L. F. Bosart, G. J. Hakim, Thielen, J., and A. Gadian, 1996: Influence of different M. A. Bedrick, M. J. Dickinson, and K. R. Tyle, 1997: wind directions in relation to topography on the

AMERICAN METEOROLOGICAL SOCIETY APRIL 2018 | 789 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC outbreak of convection in northern England. Ann. episodes. J. Appl. Meteor., 31, 964–982, https://doi Geophys., 14, 1078–1087, https://doi.org/10.1007 .org/10.1175/1520-0450(1992)031<0964:SAOFST /s005850050369. >2.0.CO;2. Viale, M., and M. N. Nuñez, 2011: Climatology of winter Wheeler, D. A., 1990: Modelling long-term rainfall orographic precipitation over the subtropical central patterns in north-east England. Meteor. Mag., 119, Andes and associated synoptic and regional char- 68–74. acteristics. J. Hydrometeor., 12, 481–507, https://doi —, 2013: Regional weather and climates of the British .org/10.1175/2010JHM1284.1. Isles—Part 4: North east England and Yorkshire. Weiss, S. J., 1992: Some aspects of forecasting severe Weather, 68, 184–190, https://doi.org/10.1002/wea thunderstorms during cool-season return-flow .2081.

New from Am S BookS !

“ A thoughtful analysis of actions that we need to take to reduce the impacts of extreme weather…a must-read for everyone with an interest in the weather and climate.” — FRAnkLin W. nuTTeR, President, Reinsurance Association of America Living on the Real World: How Thinking and Acting Like Meteorologists Will Help Save the Planet WiLLiAM H. Hooke Meteorologists sift through a deluge of information to make predictions every day. Instead of being overwhelmed by the data and possibilities, they focus on small bits of information while using frequent collaboration to make decisions. With climate change a reality, William H. Hooke suggests we look to the way meteorologists operate as a model for how we can solve the twenty-first century’s most urgent environmental problems. © 2014, PAPeRbAck 978-1-935704-56-0 LiST $30 MeMbeR $22 www.ametsoc.org/amsbookstore

790 | APRIL 2018 Unauthenticated | Downloaded 10/07/21 05:36 AM UTC half-page horizontal -- 6.5” x 4.5625”