1. Introduction

It was recognized more than 40 years ago that continental clouds have larger “colloidal stability” than maritime clouds (Squires 1958). Laboratory experiments of relevance to coalescence in clouds in cloud condensation nuclei (CCN)-rich and -lean environments (Gunn and Phillips 1957) led to the suggestion that highly continental clouds, which form in a CCN-rich atmosphere, would produce less precipitation. Smoke from burning vegetation as well as urban and industrial air pollution serve as small CCN that form large concentrations of small cloud droplets. This in turn suppresses the drop coalescence and the warm-rain processes, as well as the ice precipitation (Rosenfeld 2000; Borys et al. 2003; Andreae et al. 2004), and thus prolongs the time required to convert the cloud water that exists in small drops into large hydrometeors that can precipitate. On the other hand, precipitation-forming processes can be accelerated with the addition of large (>1 μm) hygroscopic CCN. These CCN can occur naturally in the form of salt from sea spray (Rosenfeld et al. 2002), salty dust from desert playas (Rudich et al. 2002), air pollution emissions from paper mills (Eagen et al. 1974), or even artificial hygroscopic cloud-seeding materials (Silverman 2003).

Precipitation can be shut off completely because of high concentrations of small-aerosol polluting clouds with top temperatures >−10°C (Rosenfeld 1999, 2000; Rosenfeld and Woodley 2003). The suppression of the precipitation-forming processes does not necessarily always lead to a net reduction of precipitation. The delay of early precipitation in deep convective clouds with warm bases, such as those that occur in the Tropics and the southeast United States in the summer, can lead to their invigoration and to overall additional precipitation (Andreae et al. 2004; Khain et al. 2005), mainly because of the dynamic response of the cloud development to the delayed precipitation.

In orographic clouds, the slowing of the conversion rate of cloud water into precipitation would be manifested as a net reduction of surface precipitation, because orographic clouds are often shallow and short lived because of their forced termination downwind of the topographic ridgeline. Borys et al. (2003) showed that the addition of as little as 1 μg m⁻³ of sulfate aerosols to a clean background can reduce the orographic snowfall rate in the Colorado Rocky Mountains by up to 50%, resulting from suppression of the riming of ice crystals with the smaller cloud droplets.

Suppression of orographic precipitation by anthropogenic aerosols is expected to be quite abundant, especially on the west coast of the continents in the subtropics and midlatitudes, where the precipitation over hills is a major source for the scarce water found there. Pristine air that comes from the ocean becomes polluted over the densely populated coastal plains before ascending over the downwind hills, where the pollution can have its detrimental effect on the precipitation. Givati and Rosenfeld (2004, 2005) related observed trends of decreasing orographic precipitation to the possible microphysical effects of pollution aerosols. They defined the suppression of orographic precipitation as a reduction in the orographic enhancement factor Ro, where Ro is defined as the ratio between the precipitation amounts over the hills to the precipitation amounts in the upwind lowland. Givati and Rosenfeld (2004) quantified for the first time the rainfall losses over hills and mountains downwind of major coastal urban areas in California and Israel. The suppression rate was found to be 15%–25% of the annual precipitation in hilly areas in California and Israel (Givati and Rosenfeld 2004, 2005). They showed that the suppression occurred mainly in the relatively shallow orographic clouds within the cold air mass of cyclones, and not in warm air masses. The Ro time series for relatively pristine areas crosswind to the polluted areas did not show any trend with time, and so served as controls to the polluted areas. Furthermore, Givati and Rosenfeld (2005) showed that the clouds that were most susceptible to the detrimental effects of air pollution over the hills in northern Israel were also the clouds for which glaciogenic cloud seeding caused the greatest rain enhancement. Givati and Rosenfeld (2005) suggest that it is reasonable to expect that clouds that are most sensitive to rain suppression because of aerosols that slow the rate of conversion of cloud droplets into precipitation particles would also be the clouds that would react most positively to cloud seeding that accelerates that rate.

Griffith et al. (2005) indicated similar reductions in mountainous precipitation in Utah and Nevada in a work that followed the approach utilized in the study of Givati and Rosenfeld (2004). They examined the ratio of winter precipitation between valleys to mountain precipitation downwind of the Salt Lake City and Provo, Utah, metropolitan complexes. The ratio of precipitation at mountain stations that are located in rural settings in Utah and Nevada remained stable (Griffith et al. 2005), supporting the role of pollution aerosols in decreasing Ro over the mountains to the east of Salt Lake City. While the decreasing trends of Ro in Utah occurred on the western slopes in precipitation that fell during westerly wind flows, similar decreasing trends in Ro of up to 30% were noted on the eastern slopes of the Rocky Mountains during easterly flows, downwind of Denver and Colorado Springs, Colorado (Jirak and Cotton 2006). It is difficult to explain how a change in the atmospheric circulation patterns could explain the decreasing trends of the orographic enhancement factor during both westerly and easterly winds over the respectively facing slopes, preferentially downwind of major urban areas. The temporal changes in aerosols remain the most plausible explanation.

With this background, the objectives of this study are as follows: 1) to examine the suppressive effect of aerosols on precipitation on a larger scale than California or Israel by looking at the ratio between the precipitation amounts over the hills to the precipitation over the upwind lowland areas in the western United States, from the Pacific coast inland to the Rocky Mountains; 2) to account for any variability that might be because of climatic fluctuations of the atmospheric circulation patterns; and 3) to relate the indicated trends in precipitation to trends in aerosol properties.

2. The study areas and method

3. Aerosol properties and trends

The underlying assumption of the previous studies (Givati and Rosenfeld 2004, 2005) was that the trend of Ro encompassed the period of main growth of the population, so that the overall emissions at the end of the period are larger than those at the beginning of the measured period, although not necessarily requiring a monotonic growth throughout the analysis period. Respectively, the overall precipitation suppression was evaluated as ending/starting values of Ro, as calculated using a simple linear regression line of Ro as a function of year. Air pollution indices showed recovery (decreasing trend) in California after the Clean Air Act was legislated in 1977, but Ro continued to decrease, and certainly did not show any recovery since then. Givati and Rosenfeld (2004) suggested that the there has been no decrease in the concentrations of small CCN, which are responsible for the suppression of precipitation, and pointed out the large increase in diesel consumption as a possible source of increasing production of such aerosols.

The current study expands the scope to the whole western United States to both pristine and polluted areas. Analyses of aerosol properties and trends are expanded in order to be correlated with the observed trends of precipitation, based on the IMPROVE monitoring program. Aerosol mass concentrations and composition during the winter months (October–March) were obtained for the areas of interest for the 1988–2003 (data are available online at http://vista.cira.colostate.edu/improve/). The trends were calculated using the tool provided at the IMPROVE Web site (online at http://vista.cira.colostate.edu/views/web/AnnualSummaryDev/Trends.aspx).

This tool provides quarterly averaged values of the parameters. The annual “winter” value is the average of the first and last quarters for that year. The trends of the mass concentration of the particles smaller than 2.5 μm [particulate matter (PM) 2.5] and of the particles with 2.5 < diameter < 10 μm (PM 10–PM 2.5) were calculated and are presented in Fig. 2. In addition, the trends of the following components of the fine particles were plotted: combined ammonium nitrates and sulfates, combined elemental and organic carbon, and soil materials. It is assumed here that small CCN concentrations are correlated positively with the mass of the soluble fraction of the fine particles, and hence act to suppress precipitation. Small CCN concentrations are certainly positively correlated with the fine fraction of sulfates and nitrates; and are likely to be so with the carbonaceous aerosols after having time between emission and reaching the mountain ranges to chemically mature and develop some solubility. Soil particles constitute a small fraction of the fine particles, but probably a larger fraction of the coarse particles. No information is available in the IMPROVE data about the composition of the coarse fraction. However, even insoluble mineral particles can become hygroscopic by interacting with air pollution and becoming coated with soluble materials such as sulfates, and so act to enhance precipitation (Levin et al. 1996).

The analyzed IMPROVE stations were selected based on the availability of the longest possible time series of aerosol measurements that are geographically closest to representing the conditions at the higher-elevation gauge of the rain gauge pairs. The location of the selected IMPROVE stations are shown in Fig. 1, and their details are summarized in Table 2. The average aerosol concentrations probably differ considerably from those actually interacting with the clouds over the topographical barriers. However, we can assume that the levels and trends of the aerosol concentrations that interact with the clouds are correlated with those measured at the IMPROVE stations.

According to Fig. 2 the central Sierra Nevada region(Sequoia and Yosemite, Figs. 2a and 2b) is much more polluted than the northern Sierra region (Lassen Volcano, Fig. 2c), in line with the strong decreasing trends of Ro in the central Sierra region, but not in the north, as reported by Givati and Rosenfeld (2004). Furthermore, while the coarse particles show a decreasing trend, the fine particles are stable at very high levels of 8.5 μm m⁻³ in the Sequoia area, mainly in the form of nitrates, sulfates, and organic carbon. The fine particle mass is smaller in Yosemite, probably because of the higher elevation of the monitoring station, but still shows a strong increasing trend mainly because of organic carbon (see the P values and the regression equations for the IMPROVE aerosol trends in Table 3). The decreasing trend of coarse aerosols that can have an enhancing effect on precipitation works in the same direction as the increasing trend of the fine aerosols that suppress precipitation. Coarse aerosol mass concentrations are decreasing also in the pristine areas in northern California and Oregon (Figs. 2c–e), but their concentrations are much smaller than in the central Sierra region (Figs. 2a and 2b). The concentration of the fine aerosols in these pristine areas is very small, and also is decreasing slightly. These fine aerosols are primarily composed of organic carbon, probably from residential wood stove combustion. Farther north in the densely populated and industrial area of the Puget Sound (Fig. 2f) the air is polluted once more, with an increasing trend of the fine aerosols, mainly because of nitrates.

If these IMPROVE stations represent the conditions during orographic precipitation, it would be consistent with decreasing trends of orographic precipitation in central California, stable amounts from northern California to central Oregon, and then decreasing trends again in Washington to the east of the densely populated area of Seattle, around the Puget Sound.

Figure 2h shows that Lone Peak to the east of Salt Lake City has high concentrations of fine aerosols, with slightly increasing levels resulting from nitrate and sulfate, along with a decreasing trend of coarse aerosols. This is compatible with the reported strong decrease (−24% during 1949–2004) of orographic precipitation to the east of Salt Lake City (Griffith et al. 2005). The much cleaner air to the west and south (Great Basin and Bryce Canyon, Figs. 2i and 2l) is consistent with the reported stability of the orographic precipitation in these regions (Griffith et al. 2005).

4. Ro trend analysis for winter and summer precipitation

Figure 1 summarizes the results of the winter Ro trend analysis between hilly rain gauges versus relatively lower gauges upwind in different locations across the western United States. It can be seen that Ro has decreased significantly over the years (numbers in Fig. 1 are the end/start of the winter Ro as shown in Fig. 3) in all of the areas except for the pristine regions of northern California, Oregon, southern Idaho, and central Utah (regions 8, 9, 10, 11, and 6 in Fig. 1). The full statistics for all the pairs are provided in Table 4. Winter is defined here as being from October to May. The Ro values do not change much (most Ro values increase slightly) when restricting winter to October–March, so that the IMPROVE aerosols and rain gauges pertain exactly to the same months.

Figure 3 shows the winter and the summer Ro trends for gauges located in the Phoenix, Arizona, area (number 2 in Fig. 3), Albuquerque, New Mexico (3), Salt Lake City, Utah (5), Levan, Utah, (6) Steamboat Springs, Colorado (7), and Seattle, Washington (12). The trends for additional polluted areas in California can be found in Givati and Rosenfeld (2004). While the Ro for the winter orographic precipitation (shallow, short-lived clouds) is decreasing, the same pairs of gauges show almost a stable trend for the summer precipitation (deep convective, long-lived clouds). Those findings are in agreement with what we found for many clustered hilly versus plains rain gauges in California and Israel (Givati and Rosenfeld 2004, 2005). Here we can also see that suppression of winter orographic precipitation occurs not only when maritime air is polluted over coastal urban areas, but also hundreds of kilometers farther inland. Table 4 provides the gauge details, correlations between the pairs/clusters of gauges, and summarizes the Ro trend analysis for all gauges from Fig. 1. No seasonal separation was done in areas with little summer precipitation, such as California (marked with an asterisk in Table 4).

5. Testing trends in climate indices as an alternative explanation for decreasing Ro

The most likely alternative explanation for the reduction in Ro is a decreasing trend in the cross-mountain component of the low-tropospheric wind velocity and moisture flux during rain events. Givati and Rosenfeld (2004) applied a radiosonde regression model and found that the relevant meteorological conditions during rainy days did not change systematically over the years, and that the observed trends in Ro are likely caused by nonmeteorological reasons, such as anthropogenic air pollution. Precipitation in the western United States may also be subject to climatic fluctuations like the El Niño–Southern Oscillation index (SOI) (Allan et al. 1991) and its multidecadal counterpart, the Pacific decadal oscillation (PDO) (Zhang et al. 1997; Mantua et al. 1997). The PDO index is defined as the leading principal component of the North Pacific monthly sea surface temperature variability. The PDO is a long-lived El Niño–like pattern of Pacific climate variability. The SOI is defined as the normalized pressure difference between Tahiti and Darwin (Ropelewski and Jones 1987). Figure 4 displays the winter values (October–April) of the PDO (Fig. 4a) (after Mantua et al. 1997) and the SOI (Fig. 4b) from 1900 to 2003. It can be seen that during this period the PDO had three phases: a positive (warm) phase from 1900 to 1944, a negative (cold) phase from 1945 to 1975, and another positive phase since 1976.

Dettinger et al. (2004) characterized the effects of the PDO and the SOI on the ratio of precipitation between mountains and plain areas in the California Sierra Nevada region. During negative PDO and positive SOI, the westerly wind component is stronger so that the mountain–plains orographic factor is higher than in the positive PDO phase (El Niño like). There is less overall precipitation in the negative than in the positive PDO phase (Dettinger et al. 2004). Figure 5 displays the relations between the PDO and SOI to the ratio between Cuyamaca and San Diego, California (Figs. 5a and 5b), and between Miami, and Phoenix (Figs. 5c and 5d). It can be seen that the Ro is weakly negatively correlated with the PDO, and even more weakly so with the SOI. These weak relations could still explain the indicated trends if the PDO and SOI would have large trends with time. This is not the case, however, as shown in the next section.

To better understand the possible effects of those climate indices on the orographic precipitation in the western United States, we analyzed the hill–plain Ro of precipitation with respect to the winter values of the PDO and SOI. Pairs of hill–plain gauges with the longest record were chosen to test the climate indices fluctuations through the whole century.

Figure 6 presents the Ro of annual precipitation between hilly and plain stations for three pairs of stations with the longest available records: Cuyamaca–San Diego, Miami–Phoenix, and Lake Spaulding–Eureka, California. Figures 6a and 6b are for locations downwind of urban areas, while Fig. 6c represents a relatively pristine area. The Ro time series were tested with respect to three PDO categories: (Fig. 6a) PDO < −0.5, (Fig. 6b) a neutral category of −0.5 < PDO < 0.5, and (Fig. 6c) PDO > 0.5. It can be seen in the figure that the Ro between Cuyamaca and San Diego and between Miami and Phoenix decrease for all PDO categories, while the Ro in the pristine area (Lake Spaulding versus Ukiah) remains stable with time in all the PDO categories. In the polluted areas, suppression of orographic precipitation increases with time, regardless of the PDO category.

Figure 7 shows the same methodology but for the Ro of the station pairs with the longest records with respect to three SOI categories, as was done for the PDO. It can be seen that as with the PDO (Fig. 6), the Ro decreases for all SOI categories for the polluted pairs, but not for the relatively pristine one.

Tables 5 and 6 display the results of multiple linear regressions, where the precipitation amounts and the Ro are explained by variability in the PDO, SOI, and year as the three independent regression parameters. The dependent variables in the regressions were the individual precipitation amounts from the hilly and plains area rain gauges (see Table 5) and the Ro of precipitation amounts between those hilly and plains gauges (see Table 6). It can be seen in Table 5 that the PDO and the SOI do affect the precipitation amounts. For example, P values for the PDO effect on precipitation in San Diego and Cuyamaca are 0.004 and 0.005, respectively. Nonetheless, when we test the effect of the PDO and SOI not on the absolute precipitation amounts but on the ratio between them, no significant effect is found. The only variable that has a significant effect on the ratio is the year. The Ro decreases significantly over the years, even when the variability in the PDO and SOI is taken into account by the multiple regression model. The “years” variable represents another change that occurs with time and affects Ro. With the lack of any other alternative explanation of which we are aware, we suggest that the decreasing Ro with the years is likely the result of the increasing small CCN aerosols and possibly the decreasing of giant CCN in the form of coarse aerosols over the years.

6. Discussion and conclusions

In this study, we expanded the analysis of the possible effects of air pollution on orographic precipitation from California to the whole western United States. Analyses of trends of the orographic enhancement factor Ro along the coastal mountain ranges of the whole west coast of the United States showed a pattern of decreasing Ro by as much as −24% from the south border to central California, no decrease from northern California to Oregon, and a renewed decrease (−14%) from Washington to the east of the Puget Sound. Both absolute precipitation amounts and Ro are affected by fluctuations in the atmospheric circulation patterns, such as those associated with the PDO and SOI. However, these climatic fluctuations could not explain the observed trends in Ro. A case in point is the decreasing trend of Ro in both the western (this study) and eastern (Jirak and Cotton 2006) slopes of the Rockies. It is difficult to imagine a trend in the circulation patterns that would result in such a trend.

Aerosol measurements since 1988 from the IMPROVE aerosol monitoring network in the west coast mountain ranges show that the negative trends in Ro were associated with elevated concentrations of fine aerosols (PM 2.5). The PM 2.5 showed stability or some increase in the areas where fine aerosols levels were elevated and decreased trends of Ro were noted. That increase came mainly from increasing nitrates and organic carbon that overcompensated for the decreasing trends in sulfates. In any case, there is certainly no indication of a decrease in the fine aerosols at the more polluted areas during the last two decades, which at least correspond to the lack of recovery of orographic precipitation in response to the recent efforts to reduce the air pollution.

Strong decreasing trends of the coarse aerosols (PM 10–PM 2.5) was noted, especially in the areas with elevated levels of PM 2.5. Coarse aerosols may enhance precipitation if they act as giant CCN. This is likely the case when the coarse particles interact with polluted air and are coated with soluble materials such as sulfates (Levin et al. 1996). If so, a decreasing trend in the concentrations of the coarse aerosols, in conjunction with stable or increasing concentrations of the small aerosols, might explain the continued trend of orographic precipitation losses during the last two decades, despite the indicated decreases in PM 10, based on conventional air quality standards. It remains to be determined why coarse aerosols have been decreasing during the last two decades while fine aerosols remained stable or increased at the same time.

It has been assumed that air masses well inland would not be as pristine as those impacting the coastline during onshore flow, and therefore that the impacts of pollution aerosols on orographic precipitation well inland would be smaller. However, analyses of Ro over the inland mountains of the western United States all of the way to the Rockies shows that large decreases of Ro occur there (from −15% to −25%) in areas that are potentially affected by anthropogenic sources. The IMPROVE aerosol measurements show elevated levels of PM 2.5 in the areas that experienced decreasing trends of Ro, such as those to the east of Salt Lake City, Utah, Phoenix, and Albuquerque, New Mexico. Low aerosol levels were recorded in southwest Utah and Idaho where the Ro trends were stable. The extent of decrease in Ro at the northern Rockies is not certain because of the fact that the winters of 1998–2003 are responsible for most of the decrease (see panel 7A in Fig. 3). The decrease in Ro is contributed by the much-reduced values at Steamboat Springs, Colorado, with no obvious deviation at Hayden, Colorado, at the same time. No similar strong decrease in the recent years was indicated anywhere else in the gauges analyzed for this study. If the Ro that occurred in the northern Colorado Rockies really decreased, it was associated with rather low aerosol levels measured on Mount Zirkel. These low concentrations may mainly reflect the high altitude (3243 m) of the measuring station. Such low concentrations of PM 2.5 (averaging about 2 μm m⁻³) do not exclude the possibility of long-range air pollution transport being responsible for the possible decrease in Ro. Measurements conducted on top of a nearby mountain of the same elevation (Borys et al. 2003) have shown that fine aerosol concentrations as low as 1 μm m⁻³ can cut the snow precipitation rate in half relative to pristine conditions. In any case, we should not expect a perfect match between the IMPROVE and rain gauge trends, because the IMPROVE aerosol data cover only the last two decades, whereas the trends in Mount Zirkel are mainly based on data from the previous decades.

Evidently, decreases in Ro occurred during winter orographic precipitation, but not during convective summer precipitation over the same mountain ranges. This is in line with the expectation that aerosol-induced changes in the rate of the conversion of cloud water to precipitation would result in a respective net change of surface precipitation in shallow and short-lived orographic clouds, but not necessarily in deeper and longer-living thermally driven convective clouds.

These reported findings suggest that anthropogenic air pollution has had a major impact on orographic precipitation well beyond the local scale of the pollution sources. It appears that air pollution suppresses much of the orographic precipitation over the western United States, which is responsible for most of the water resources in this semiarid part of the world. This is an issue with major economical and societal implications, not only to the United States, but also to many other densely populated parts of the world where the livelihood of the inhabitants depend on water from orographic precipitation, which might be compromised by the air pollution produced by the inhabitants. Examples of such densely populated semiarid areas that depend on orographic precipitation that is susceptible to air pollution produced by the local inhabitants include the Mediterranean basin, the west coast of the United States, the Middle East, Southeast Asia and China, and the southeastern part of Australia and South Africa.