Abstract

A better understanding of the effects of precipitation and source area on blowing dust in the Sonoran Desert has been sought through the study of 1190 dust episodes occurring during the 1948–78 time period at Blythe, California, and Yuma, Arizona. Threshold mean hourly wind speeds (MHWSs) increase directly with prior precipitation in proportion to the inhibiting effect of the vegetation canopy, which blooms following periods of increased precipitation. Because of the time required for the vegetation canopy to fully develop and the persistence of the vegetation canopy once developed, correlation between the threshold MHWS and precipitation is highest for 4–6-month windows of total precipitation prior to each dust event at both stations. Many dust events associated with unusually low MHWSs are clustered in time, and these events can be correlated with interstate highway construction and soil preparation for new irrigation projects. Since threshold MHWSs for blowing dust lie well below the recorded MHWSs during most dust events at most times, it is possible to predict that mean annual precipitation could in the future increase to about 8 cm per 6 months without significantly reducing the occurrence of blowing dust episodes. On the other hand, increases in future mean annual precipitation to 10–12 cm per 6 months would raise the threshold MHWS to the point that conditions for blowing dust would be substantially reduced. Many of the infrequently occurring periods of elevated precipitation correlate in time with El Niño–Southern Oscillation events, which typically repeat every 2–8 yr. Average MHWSs, and threshold MHWSs for blowing dust, vary with wind direction at Blythe and Yuma. These variations can be related to variations in the susceptibility of upwind source areas in most instances, but in one or more instances this variation may be related to storm type.

Introduction

Correlation between susceptibility to blowing dust, antecedent precipitation, and other factors is known and documented in the Sonoran Desert and elsewhere although explaining this correlation precisely in terms of source area, threshold wind speeds, vegetation, soil condition, storm type, topography, and other factors has presented problems (Chepil 1956; Gillette and Passi 1988; Brazel 1989; MacKinnon et al. 1990). Soil condition (soil moisture, crusting, soil disturbance) and vegetation are controlling factors for potential dust emission, but the relative contribution of each is not easily understood and quantified. In this paper we have sought a better understanding of the conditions for blowing dust in the Sonoran Desert by examining 1190 dust episodes occurring during the 1948–78 time period at Blythe, California, and Yuma, Arizona.This area is a good natural laboratory for such study because blowing dust episodes occur frequently enough to yield a large statistical sample, and other studies in this area provide a prior basis of understanding to build on. We have examined wind speeds as a function of prior precipitation and wind direction, and derived threshold wind speeds, in order to test the following: 1) threshold wind speed increases with increased antecedent precipitation; 2) growth and decay of vegetation canopies in the source area are the dominant controlling factors for threshold wind speed; 3) response of the vegetation canopy to amount of precipitation will be approximately the same regardless of storm type or duration, wind direction, or time of year; and 4) threshold wind speed varies with natural susceptibility of upwind source areas.

The data

Records of blowing dust are contained in the hourly weather observations from the Blythe and Yuma airport weather stations. For his study of dust climatology in the western United States, Changery (1983) extracted from these hourly weather observations (forms MF-10A) a record of the duration and intensity of blowing dust episodes in the period 1948–78. Changery defined dust events as times of reduced visibility. Specifically, he recorded the beginning and ending times of every incidence of reduced visibility due to blowing dust through four visibility thresholds: 11 300, 4800, 1600, and 1000 m. Starting with Changery’s log sheets of visibility data, we went back to the original MF-10A hourly weather observations for stations Blythe and Yuma and extracted mean hourly wind speed and wind direction during each dust episode as defined by visibility falling through the 11 300-m threshold. Other functions of wind speed, namely, highest hourly wind speed, lowest hourly wind speed, mean extreme wind speed, highest extreme wind speed, and wind direction, were also extracted for each dust episode during 1948–78. Monthly totals of precipitation were extracted for both stations from the Historical Climatology Network database compiled by the Carbon Dioxide Information Analysis Center of the Oak Ridge National Laboratory. Two spreadsheets were created containing all this information for the Blythe and Yuma stations. These spreadsheets contain one entry line for each dust episode, containing the following columnar entries for each event: year; month; day; time of day; begin times and end times when visibility dropped below 11 300, 4800, 1600, and 1000 m, respectively; mean hourly wind speed; wind direction; highest hourly wind speed; lowest hourly wind speed; mean extreme wind speed; highest extreme wind speed; wind direction; and total monthly precipitation for the current month. In all there were 379 dust episodes for Blythe and 811 dust episodes for Yuma. Results of this study were derived from analyses performed on information contained in these spreadsheets.

The sites

Blythe and Yuma lie within the most arid part of the Sonoran Desert. Mean annual precipitation is about 7 cm at Yuma and 9 cm at Blythe, with this precipitation being unevenly distributed through time; the area experiences periods of months with almost no rainfall.

Vegetation consists of ephemerals, which bloom after precipitation events, and larger brush-type plants, which manage to survive by growth, decay, and dormancy cycles in response to times and amounts of precipitation. The larger scrub-type perennial plants are the dominant features of the vegetation canopy and they consist predominantly of creosote (Larrea tridentata), white bursage (Ambrosia dumosa) or delta-leaf bursage (Ambrosia deltoidea), and big galleta (Hilaria rigida) (Crosswhite and Crosswhite 1982).

Features of the surficial geology are summarized in Fig. 1. Geologically, thearea is a mountain-and-basin region characterized by northwest-trending fault-block terrane. The uplifted fault blocks of mostly Mesozoic volcanics and metamorphics are in an advanced stage of erosion, and their erosion products are the predominant source of sediments making up the intervening bolson, or basin-filling, deposits, which at the surface are mostly of Quaternary age. Surficial sediments, which are sources of dust, are predominantly these bolson deposits, together with alluvial deposits along the rivers and washes, and evaporite deposits in dry lake beds. The areas having the greatest susceptibility to blowing dust are several dry lake beds west of Blythe, and noncultivated regions lying west and south of Yuma (East Mesa and Gran Desierto), which are underlain by Colorado River deltaic deposits.

Fig. 1.

Index map of the Blythe, California–Yuma, Arizona area showing the location of physiographic and cultural features and containing a classification of land surface areas according to probable susceptibility to blowing dust.

Fig. 1.

Index map of the Blythe, California–Yuma, Arizona area showing the location of physiographic and cultural features and containing a classification of land surface areas according to probable susceptibility to blowing dust.

Agriculture is possible only along the Colorado and Gila Rivers and in low-lying areas of the Imperial Valley where surface water is available (Fig. 1). Agricultural activities, which rely on year-round irrigation, do not at present contribute significantly to the supply of dust in the area (MacKinnon et al. 1990).

For both Blythe and Yuma temporal distribution throughout the year of dust episodes over the 30-yr period of study is trimodal, with peak frequencies occurring in February–April, July–August, and November (Fig. 2). Dust episodes have occurred on the average at a rate of 13 per year at Blythe and a rate of 27 per year at Yuma. Whereas the events are unevenly distributed with respect to time of year, substantial numbers of events have occurred during all months of the year.

Fig. 2.

Number of low-visibility events (visibility less than 11 300 m) due to blowing dust at (a) Blythe, California, and (b) Yuma, Arizona, for the period of study 1948–78, as a function of month of the year. Months numbered 1 through 12 are January through December, respectively.

Fig. 2.

Number of low-visibility events (visibility less than 11 300 m) due to blowing dust at (a) Blythe, California, and (b) Yuma, Arizona, for the period of study 1948–78, as a function of month of the year. Months numbered 1 through 12 are January through December, respectively.

Rainfall (Fig. 3) occurs sporadically in the late fall to early spring and also in the late summer to early fall at both localities. Rainfall is negligible during the late spring to early summer. Whereas the temporal distribution of dust episodes and rainfall differ significantly from one another, both are associated with frontal events, which move through the area during the winter season, and with monsoonal convective storms, which occur during the late summer.

Fig. 3.

Mean rainfall for the period of study 1948–78 at (a) Blythe, California, and (b) Yuma, Arizona, as a function of month of the year. Months numbered 1 through 12 are January through December, respectively.

Fig. 3.

Mean rainfall for the period of study 1948–78 at (a) Blythe, California, and (b) Yuma, Arizona, as a function of month of the year. Months numbered 1 through 12 are January through December, respectively.

Frequency distribution of wind speeds for all events is shown in Fig. 4. The mean of all mean hourly wind speeds is 929 cm s−1 at Blythe and 850 cm s−1 at Yuma; the median is 923 cm s−1 at Blythe and 894 cm s−1 at Yuma. The highest mean hourly wind speed recorded was 2240 cm s−1 at Blythe and 1967 cm s−1 at Yuma; 95% of all events recorded lie between the extremes of 380 cm s−1 and 1568 cm s−1 at Blythe and 290 cm s−1 and 1296 cm s−1 at Yuma.

Fig. 4.

Histogram of mean hourly wind speeds (MHWSs) recorded during low-visibility events due to blowing dust at (a) Blythe, California, and (b) Yuma, Arizona, during the period of study 1948–78.

Fig. 4.

Histogram of mean hourly wind speeds (MHWSs) recorded during low-visibility events due to blowing dust at (a) Blythe, California, and (b) Yuma, Arizona, during the period of study 1948–78.

Frequencies of dust events as a function of wind direction are illustrated in Fig. 5 for both sites. The principal correlation is between frequency of dust events and frequency of high wind events, but there are also variations in mean wind speed for dust events as a function of wind direction, which are probably related to susceptibility of upwind sources and possibly also to fetch. High winds at Blythe occur with greatest frequency from the northwest to north and from the southwest to west. High winds at Yuma occur with greatest frequency from the north, from the southeast to south, and from the west to west-northwest. To some degree, direction of prevailing winds is influenced by topography, as examination of Fig. 1 reveals.

Fig. 5.

Number of dust events occurring at (a) Blythe and (b) Yuma during the period of study 1948–78, as a function of wind direction.

Fig. 5.

Number of dust events occurring at (a) Blythe and (b) Yuma during the period of study 1948–78, as a function of wind direction.

Mean hourly wind speeds, threshold mean hourly wind speeds, and prior precipitation

Distribution of mean hourly wind speed (MHWS) as a function of 1–12-month periods of prior precipitation, for all 1948–78 dust episodes, is presented in expanded form (Fig. 6, Blythe; Fig. 7, Yuma). In both of these expansions the MHWSs of all dust events are plotted on 12 successive xy plots, but each plot depicts the MHWSs plotted against different time period functions of prior precipitation. In this paper threshold MHWS is defined as the minimal MHWS necessary to entrain dust into the atmosphere in quantities sufficient to reduce visibility below 11 300 m. By definition, all MHWSs associated with dust events exceed threshold MHWS. Most of the MHWSs shown in Figs. 6 and 7 are randomly distributed, having wind speeds well in excess of the threshold wind speed. The lower edge of the envelope for each xy plot is defined by a few dust events, the MHWSs of which barely exceed threshold MHWS. Increase of threshold MHWS with increasing prior precipitation is expected and is shown on most of the xy plots.

Fig. 6.

MHWS (cm s−1) for each dust episode at Blythe, California, occurring during the period 1948–78, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. For each dust event MHWS was paired with prior-month precipitation for 12 time windows ranging from the prior month to the prior 12 months, MHWS = f( pp); pp = Σ pi, i = m − 1, mw and w = 1, 2, 3, . . . , 12, where pp is the prior precipitation, m is the current month during which each particular dust episode occurred, pi is the total monthly precipitation for a given month as designated, and w is the length of time of the prior precipitation window in months.

Fig. 6.

MHWS (cm s−1) for each dust episode at Blythe, California, occurring during the period 1948–78, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. For each dust event MHWS was paired with prior-month precipitation for 12 time windows ranging from the prior month to the prior 12 months, MHWS = f( pp); pp = Σ pi, i = m − 1, mw and w = 1, 2, 3, . . . , 12, where pp is the prior precipitation, m is the current month during which each particular dust episode occurred, pi is the total monthly precipitation for a given month as designated, and w is the length of time of the prior precipitation window in months.

Fig. 7.

MHWS (cm s−1) for each dust episode at Yuma, Arizona, occurring during the period 1948–78, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. See Fig. 6 for further explanation.

Fig. 7.

MHWS (cm s−1) for each dust episode at Yuma, Arizona, occurring during the period 1948–78, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. See Fig. 6 for further explanation.

Threshold MHWSs as a function of 1–12-month periods of prior precipitation were obtained by deriving the lower envelope for each xy distribution depicted in Figs. 6 and 7; these are shown in a complementary series of xy plots (Figs. 8 and 9). The regression curve using exponential best fit is shown on each plot. Correlation (R2) of each fit is plotted as a function of prior precipitation for Blythe and Yuma, respectively, in Fig. 10. Also shown in Fig. 10 are the zero intercepts for each regression analysis. These illustrations highlight several important relationships between threshold MHWS for blowing dust and prior precipitation: 1) the threshold MHWS increases exponentially with increasing prior precipitation; 2) the response of the threshold MHWS to increased precipitation is not immediate, and correlation, which is poor between threshold MHWS and precipitation for the prior 1–2 months, increases to a maximum correlation in the prior 4–8-month period and deteriorates in the prior 9–12-month period; 3) maximum correlation occurs at 5 months at Blythe and 6 months at Yuma, corresponding to R2 values of 0.766 (R = 0.875) and 0.788 (R = 0.888), respectively; and 4) zero intercepts are in the range of 150–180 cm s−1 at both stations in the 4–8 months prior precipitation windows.

Fig. 8.

Lower edge of the envelope of MHWSs (cm s−1) at Blythe, California, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. For all xy distributions, the lower edge of the envelope was obtained by binning 1 cm s−1 intervals of prior precipitation, frequency of MHWS = frequency of pp between pp = x − 1 and pp = x, where x represents the upper limit of precipitation of each successive bin in centimeters. The lowest MHWS within each bin was selected to define the lower edge of the envelope. An exponentialregression curve has been fitted to each set of plotted points.

Fig. 8.

Lower edge of the envelope of MHWSs (cm s−1) at Blythe, California, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. For all xy distributions, the lower edge of the envelope was obtained by binning 1 cm s−1 intervals of prior precipitation, frequency of MHWS = frequency of pp between pp = x − 1 and pp = x, where x represents the upper limit of precipitation of each successive bin in centimeters. The lowest MHWS within each bin was selected to define the lower edge of the envelope. An exponentialregression curve has been fitted to each set of plotted points.

Fig. 9.

Lower edge of the envelope of MHWSs (cm s−1) at Yuma, Arizona, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. See Fig. 8 for further explanation.

Fig. 9.

Lower edge of the envelope of MHWSs (cm s−1) at Yuma, Arizona, plotted as a function of antecedent monthly precipitation totals (pp) in centimeters, for 1- through 12-month windows of antecedent precipitation. See Fig. 8 for further explanation.

Fig. 10.

Results of the exponential regression analyses performed on each dataset shown in Figs. 8 and 9: R2 (a) and zero intercept (b) are plotted for each of the 1- through 12-month windows of antecedent precipitation.

Fig. 10.

Results of the exponential regression analyses performed on each dataset shown in Figs. 8 and 9: R2 (a) and zero intercept (b) are plotted for each of the 1- through 12-month windows of antecedent precipitation.

Corresponding to the high correlation associated with the 5-month window of prior precipitation at Blythe and the 6-month window of prior precipitation at Yuma, the regression equations are as follows.

 
formula

Using the above regression equations, threshold MHWS is plotted against time at Blythe and Yuma for the 1948–78 time period (Figs. 11 and 12). The MHWSs of each dust episode plotted against time also appear on the plots.

Fig. 11.

Threshold MHWS Ut at Blythe derived from the regression analysis of the 5-month window of antecedent precipitation, plotted as a function of time for the 1948–78 period (solid line). Peaks labeled “E” are those associated with ENSO events. MHWS u for each dust episode at Blythe is also plotted as a function of time.

Fig. 11.

Threshold MHWS Ut at Blythe derived from the regression analysis of the 5-month window of antecedent precipitation, plotted as a function of time for the 1948–78 period (solid line). Peaks labeled “E” are those associated with ENSO events. MHWS u for each dust episode at Blythe is also plotted as a function of time.

Fig. 12.

Threshold MHWS Ut at Yuma, Arizona, derived from the regression analysis of the 6-month window of antecedent precipitation, plotted as a function of time for the 1948–78 period (solid line). Peaks labeled “E” are those associated with ENSO events. MHWS u for each dust episode at Blythe is also plotted as a function of time.

Fig. 12.

Threshold MHWS Ut at Yuma, Arizona, derived from the regression analysis of the 6-month window of antecedent precipitation, plotted as a function of time for the 1948–78 period (solid line). Peaks labeled “E” are those associated with ENSO events. MHWS u for each dust episode at Blythe is also plotted as a function of time.

The few events defining the lower limits of the envelope have MHWSs that are well below the main body of MHWSs recorded in most events (Figs. 11 and 12). To more accurately characterize lower limits of the MHWSs of the main body of dust events, a second set of regression analyses for both stations were done as previously described except that the fifth percentile instead of lowest value from each 1-cm bin was used to define the lower envelope. Correlation, which is poor between threshold wind speeds and precipitation for the prior 1–2 months, increases to a maximum correlation in theprior 3–6-month period and deteriorates in the prior 7–12-month period; the maximum correlation, which occurs at 4 months at Blythe and 6 months at Yuma, corresponds to R2 values of 0.834 (R = 0.913) and 0.781 (R = 0.883), respectively. Zero intercepts in the prior 3–6-month period are in the range of 400 cm s−1.

These regression equations can be used to compute a threshold MHWS at any time, and they have the attribute of allowing such computation from monthly precipitation alone. However, many of the dust events having the lowest MHWSs apparently result from unusually high susceptibility such as would occur following man-induced soil disturbances. Caution is therefore advised in applying the above regression equations to long precipitation records and precipitation proxies in order to hindcast naturally occurring threshold wind speeds; use of a regression derived from the lower envelope, as defined by the fifth percentiles of MHWS, is probably a better simulation of naturally occurring conditions.

Those dust events having the unusually low MHWSs are clustered in time. Clusters occurred during 1950–52, 1957–58, 1960, 1965–66, 1969, 1971, and 1974 at Yuma; and during 1959–62, 1969, and 1971–72 at Blythe. It is possible that one or two of these events may have been mistakenly associated with fog or smoke rather than blowing dust. Many of these events may be associated with intense summer convective storms, which are characterized by high winds in the source areas but only moderate winds at the weather station. Many of these events are probably also associated with local man-made disturbances of the land surface, which for a period of time result in conditions of extraordinarily high susceptibility to dust entrainment. Note that during the late 1950s and early 1960s, there was a large project to bring new land into irrigation in the Palo Verde Valley south of Blythe (G. Davis, Manager, Palo Verde Irrigation District, 1992, personal communication), which would have created a major disturbance of a dry land surface. During 1969 through 1972, there was heavy construction on Interstate Highway 10 through the Blythe area. In the Yuma area irrigation development occurred progressively in the Wellton–Mohawk area east of Yuma during 1948–58; irrigation development in the Yuma–Mesa area in the immediate vicinity of the Yuma Marine Corps Air Station occurred during 1957–58 [B. Stevenson, district conservationist, U.S. Department of Agriculture (USDA) Soil Conservation Service 1992, personal communication]. The All-American Canal was constructed in the 1950s. Construction of Interstate Highway 8 was carried out in the Yuma area in the time frame 1965–78 (Yuma Daily Sun 1978). It is reported that some dust occurrences in the 1970s and 1980s were associated with simulated dust conditions (“dust courses”) and other military-related activities at the U.S. Army Yuma Proving Ground northeast of Yuma and the Luke Air Force Range southeast of Yuma (B. Stevenson 1992, personal communication).

Influence of the vegetation canopy on the correlation between threshold mean hourly wind speeds and prior precipitation

Low correlation between the threshold MHWSs for blowing dust and prior 1–2-month precipitation totals (1–3 months at Yuma; Fig. 10) is very likely because the response of the vegetation canopy to increased or decreased precipitation is not immediate; time is required to develop a new vegetation canopy following precipitation, and once a canopy of vegetation is in place, it can survive for several months without precipitation before it dies, decays, and its effect on threshold MHWS diminishes. Diminished correlation in the 9–12-month window of prior precipitation can also be attributed to the fact (see Figs. 6 and 7) that there are few intervals as long as 9–12 months without significant precipitation; that is, wet–dry cycles are beginning to repeat and cancel each other out. Highest correlation between threshold MHWS and prior precipitation in the 4–6-month window and deteriorating correlation in the 7–8-month window is therefore, we suspect, a function of the response time to precipitation necessary to grow a vegetation canopy and the response time to drought that is necessary for a vegetation canopy to die and waste away. The shorter response time at Blythe as compared to Yuma may be a function of a relatively greater contribution to the vegetation canopy of grasses and ephemerals at Blythe. At Yuma, the effect of slower-growing and more drought-resistant shrubform vegetation may be the dominant factor affecting the response time of the vegetation canopy.

The threshold MHWS for entrainment of blowing dust lies well below the MHWS of a majority of recorded dust episodes at both Blythe and Yuma (Figs. 11 and 12), even if the higher-threshold MHWS of 400 cm s−1 (ignoring unusually low speed events) is assumed. Only at infrequent intervals is precipitation sufficient so that the resulting vegetation canopy significantly impedes dust entrainment. In effect, drought is the norm and significant precipitation the exception, at both stations. Therefore, the significantly larger number of events occurring at Yuma compared to Blythe must be a result of a higher number of dust-producing wind events per unit time at Yuma. Statistics of the distribution of MHWSs for all events show that both stations are similar in terms of means and medians for event MHWSs, although those for Blythe are slightly higher (Fig. 4). Brazel (1989) shows that for 17 stations in the arid southwest there is considerable scatter in the frequency of dust events for the most arid sites.

We can now speculate regarding vegetation-related changes in susceptibility to blowing dust, which would likely accompany future precipitation changes in the area. It seems evident that increased drought will only result in marginal increases in the overall frequency of dust episodes, since threshold MHWSs are already well below the MHWSs encountered in most dust episodes. Modest increases in precipitation would for the same reason not be expected to significantly alter the incidence of blowing dust episodes per unit time; precipitation could increase to 8 cm over a 6-month period before the vegetation canopy would begin to significantly impede entrainment of blowing dust. On the other hand, an increase in precipitation to 10–12 cm over a 6-month period would be expected to result in a vegetation canopy sufficient to significantly reduce the number of dust events, and an increase to 16–18 cm over a 6-month period would be expected to largely eliminate the occurrence of blowing dust. Comparing frequency of dust episodes versus precipitation at 17 stations in Arizona, California, and Nevada, Brazel (1989) showed that dust episodes are rare at sites where mean annual precipitation exceeds 30 cm; this result and the overall inverse relationship between dust storm frequency and mean annual precipitation as illustrated by Brazel (1989) agree generally with our predictions.

Correlation between periods of high precipitation and El Niño–Southern Oscillation

For the period 1948 through 1978, threshold wind speed as shown in Figs 11 and 12 is a proxy for a moving 5-month (Blythe) or 6-month (Yuma) window of total precipitation. This time series of precipitation shows that precipitation atboth stations is characterized by long intervals of relative drought, which is the norm, interspersed with infrequent intervals of anomalously high precipitation. Many of these high-precipitation events correlate with historic El Niño–Southern Oscillation (ENSO) events (see Quinn et al. 1987; Cole et al. 1992). The correlation is particularly impressive at Blythe, where seven of nine high-precipitation events, and all the highest precipitation events, correlate with ENSO events. At Yuma the correlation is much lower, only 5 of 15 precipitation events, and none of the highest precipitation events, correlate with ENSO events. This significant lack of synchroneity between two stations only 80 km apart is surprising and suggests local rain-shadow effects resulting from Yuma’s proximity to the mountain ranges of Baja California, which lie to the southwest. Some of the Yuma high-precipitation events occur in the warmer season and may be related to local monsoon conditions deriving from Yuma’s proximity to the Gulf of California.

Effect of source area on threshold mean hourly wind speeds

Source areas for blowing dust have been characterized for the Blythe–Yuma region in Fig. 1. As previously stated, mountains, wetlands, irrigated agricultural regions, water-covered areas, and urban areas are not significant source areas for dust. Mountain basins underlain by pediments and bolson deposits are significant sources for dust. As we have seen, susceptibility of the bolson areas varies with the condition of the vegetation canopy as influenced by prior precipitation. Susceptibility also varies locally. Near the mountains, desert pavement is more ubiquitous and it provides some protection against deflation of the underlying dust. Farther from the mountains, desert pavement is less continuous and it affords less protection for the underlying soil. Susceptibility is greatly increased throughout the bolsons by any disturbance of the desert pavement, for example a passing vehicle. A major increase in susceptibility occurs as a result of any land surface modification such as road construction, canal digging, or bench leveling. The areas most susceptible to blowing dust are dry lake beds and nonirrigated portions of the Colorado River delta. These areas have the least vegetation (or no vegetation at all) and much of the underlying sediment is distal, containing a large fraction of finer grain sizes (silt and clay) and lacking sufficient pebble- and cobble-sized grains to form desert pavement. Susceptibility in these areas is also greatly influenced by the condition of the vegetation canopy, except where there is no vegetation. Delta and dry lake bed areas are also sites of greatly increased susceptibility if the soil surface is disturbed. In bolson and delta areas the sporadically spaced desert shrubs inhibit deflation under their canopy, resulting in the soil being typically banked up around each plant, such that each occupies its own little hillock. This effect was observed to be most pronounced in the delta areas (East Mesa and Yuma Desert) where hillocks of 2-m relief are not uncommon. It is reported (G. Davis 1992, personal communication) that susceptibility of dry lake beds west of Blythe has been increased following high-rainfall events, which bring into the lake beds a renewed supply of fine-grained, unconsolidated sediment; once dried out, these vegetation-free areas are particularly susceptible to dust entrainment. West of Yuma the Algodones Dunes extend northwestward along the eastern edge of the Imperial Valley. While these dunes are certainly a local dust source, they constitute a lag deposit of mostly sand-sized sediment, thought to be less prolific a source of blowing dust than the East Mesa area lying immediately to the west. The East Mesa area is underlain by sediments containing a large percentage of silt- and clay-sized material. Areas of theColorado River delta, including the Imperial Valley, which are now subject to intensive year-round irrigated agriculture, were probably very significant sources of blowing dust before they were irrigated and placed into production. In this regard the original U.S. Army Signal Corps meteorological log books from Fort Yuma, now in the possession of the Arizona Historical Society (courtesy of C. Brooks, curator), contain a daily record of dust storms for some years in the late 1800s. From 4 October, 1875 through 4 October 1876, a total of 55 dust storms occurred at Fort Yuma, twice the average annual occurrence in 1948–78 and more than the total number of events recorded at Yuma during any year between 1948 and 1982.

Variation of mean hourly wind speeds and threshold mean hourly wind speeds with wind direction

Mean hourly wind speeds vary with wind direction. This variation no doubt contributes to the noise level associated with derived threshold MHWSs, and it is logical to relate these variations in MHWS versus direction to the dust-yielding potential of upwind source areas. At Blythe, the most prolific source areas with significant fetch lie to the south of the airport along the Paloverde Mesa, and to the west in Chuckwalla Valley (Ford Lake and Palin Lake) (Fig. 1). Dust events having the lowest MHWSs at Blythe are those associated with south and west winds; dust events having the highest MHWSs at Blythe are those associated with north-northwest to northeast winds (Fig. 13). At Yuma, the obvious potential source areas for dust having a long fetch lie west to northwest of Yuma (East Mesa) and south-southeast (SSE) to south-southwest of Yuma (Yuma Desert and Gran Desierto) (Fig. 1). Dust events having the lowest MHWS at Yuma are those associated with west-southwest to west-northwest winds and northeast (NE) to east winds; dust events having the highest MHWSs at Yuma are associated with north-northwest to north-northeast winds and with southeast to south-southeast winds; the high MHWS associated with east-southeast winds is probably not significant because of the small number of east-southeast events (Fig. 14). The same variation observed in MHWS versus wind direction is also apparent in the variation in threshold MHWS versus wind direction at Yuma (Summary of Meteorological Observations—Surface, 1991; courtesy of Chief Warrant Officer Davis, weather officer, Yuma Marine Corps Air Station). For percentage of time when MHWSs in excess of 566 cm s−1 are blowing, percent frequencies are highest for the north and south-southeast directions (Fig. 15); for percentage of time when dust is blowing, percent frequencies are highest for the west and west-northwest directions (Fig. 16). Percent frequency of time that dust is blowing exceeds percent frequency of time that MHWSs in excess of 566 cm s−1 are blowing when winds are from the west, west-northwest, northeast, east-northeast, and east; obviously, the threshold MHWS for blowing dust is less than 566 cm s−1 for these directions (Figs. 15 and 16). Percent frequency of time that dust is blowing is less than percent frequency of time that MHWSs in excess of 566 cm s−1 are blowing when winds are from the northwest, north-northwest, north, north-northeast, southeast, south-southeast, and south-southwest; the threshold MHWS for blowing dust is therefore greater than 566 cm s−1 for these directions (Figs. 15 and 16). The Yuma pattern suggests higher-susceptibility source areas for dust west of Yuma and east to northeast of Yuma, andlower-susceptibility source areas for dust north to northwest and south to southeast of Yuma. This pattern correlates with the occurrence of high-susceptibility sources west and east of Yuma and the absence of a high-susceptibility source north of Yuma, but it does not correlate with the occurrence of a high-susceptibility source south to southeast of Yuma (Fig. 1). There is no obvious explanation for this seeming contradiction. Possibly the local, convective, short-duration summer storms moving up from the Gulf of California, which are associated with southeasterly winds, are not as efficient at transporting dust for given wind speeds as the winter frontal systems, which account for most of the westerly and northerly winds.

Fig. 13.

MHWS for each dust event occurring at Blythe, plotted as a function of wind direction. Wind directions 1–16 correspond to the points of the compass, proceeding clockwise with 1 being north and 16 being north-northwest. Filled triangles show means of hourly wind speeds for each wind direction. Horizontal line shows mean wind speed for all dust events.

Fig. 13.

MHWS for each dust event occurring at Blythe, plotted as a function of wind direction. Wind directions 1–16 correspond to the points of the compass, proceeding clockwise with 1 being north and 16 being north-northwest. Filled triangles show means of hourly wind speeds for each wind direction. Horizontal line shows mean wind speed for all dust events.

Fig. 14.

MHWS for each dust event occurring at Yuma, plotted as a function of wind direction. Wind directions 1–16 correspond to the points of the compass, proceeding clockwise with 1 being north and 16 being north-northwest. Filled triangles show means of hourly wind speeds for each wind direction. Horizontal line shows mean wind speed for all dust events.

Fig. 14.

MHWS for each dust event occurring at Yuma, plotted as a function of wind direction. Wind directions 1–16 correspond to the points of the compass, proceeding clockwise with 1 being north and 16 being north-northwest. Filled triangles show means of hourly wind speeds for each wind direction. Horizontal line shows mean wind speed for all dust events.

Fig. 15.

Mean annual percentage of time that winds in excess of 566 cm s−1 are blowing from the directions shown, at Yuma Marine Corps Air Station.

Fig. 15.

Mean annual percentage of time that winds in excess of 566 cm s−1 are blowing from the directions shown, at Yuma Marine Corps Air Station.

Fig. 16.

Mean annual percentage of time that winds blowing from the directions shown are causing reduced visibility (visibility less than 11 300 m) due to blowing dust, at Yuma Marine Corps Air Station.

Fig. 16.

Mean annual percentage of time that winds blowing from the directions shown are causing reduced visibility (visibility less than 11 300 m) due to blowing dust, at Yuma Marine Corps Air Station.

We have seen that MHWSs vary with wind direction (Figs. 13 and 14) and that these variations derive from susceptibilities to blowing dust in the upwind source areas. It follows that if one adjusts or normalizes MHWSs associated with each wind direction to a common mean, scatter resulting from differences in source areas would be reduced, and the resulting “normalized” dataset would contain a clearer signal of the effect of the vegetation canopy on susceptibility to blowing dust. Accordingly, a third regression analysis was done as described previously, using the entire dataset for each station, with each event adjusted or normalized by an amount equal to the difference between the MHWS for each direction and the MHWS for all events from each station. The difference between the mean for each wind direction and the overall mean is shown in Figs. 13 and 14. As predicted, correlation is improved between threshold MHWS as a function of prior precipitation summed over the various 1–12-month precipitation intervals. Results are shown in Figs. 17 and 18. In all but the 7-month window at Yuma, correlation remained about the same or increased; for all windows, correlation increased at Blythe. For the 5-month window at Blythe, R2 increased from 0.766 to 0.850 (R = 0.922). Zero intercept for windows having a high correlation increased to a little over 200 cm s−1 at Yuma and about 250 cm s−1 at Blythe.

Fig. 17.

For Blythe, comparison of R2 and zero intercept between the raw dataset (as in Fig. 8) and the dataset in which all dust events are adjusted for differences in mean wind speed associated with each wind direction.

Fig. 17.

For Blythe, comparison of R2 and zero intercept between the raw dataset (as in Fig. 8) and the dataset in which all dust events are adjusted for differences in mean wind speed associated with each wind direction.

Fig. 18.

For Yuma, comparison of R2 and zero intercept between the raw dataset (as in Fig. 9) and the dataset in which all dust events are adjusted for differences in mean wind speed associated with each wind direction.

Fig. 18.

For Yuma, comparison of R2 and zero intercept between the raw dataset (as in Fig. 9) and the dataset in which all dust events are adjusted for differences in mean wind speed associated with each wind direction.

Threshold friction velocities estimated from threshold mean hourly wind speeds

Potential for blowing dust at a specific location at a specific time is proportional to wind energy available for transport of that dust. Available wind energy may be expressed in terms of friction velocity, a term that describes the wind energy profile and can be derived from measured wind speed, instrument height, and an index of surface roughness. Theoretical aspects of friction velocity and definitions are provided in Stockton and Gillette (1990) and MacKinnon et al. (1990). Conditions for dust entrainment in the atmosphere are met when friction velocity exceeds threshold friction velocity at a specific site at a specific time. For the purposes of this study friction velocity can be regarded as proportional to measured MHWSs. Threshold friction velocity can be regarded as proportional to a threshold MHWS expressed in the same units as MHWSs measured at the weather stations.

Threshold MHWSs at times of zero prior precipitation (zero intercepts) lie consistently in the range of 150–180 cm s−1 (Fig. 10) for the 4–7-month windows of prior precipitation associated with the lower envelope enclosing all dust events, and in the range of 400 cm s−1 for the 3–6-month windows of prior precipitation associated with the lower envelope enclosing the main bodyof dust events. This value is taken to be representative of the threshold MHWS required to blow dust from an unvegetated, desert surface during times of drought and optimum conditions for dust entrainment in the source area. Conditions associated with the lower of the zero intercepts probably include natural or anthropogenic soil disturbance. At Yuma, using the known anemometer height z of 1005 cm, a surface roughness factor z0 of 1 cm, and a von Kármán’s constant k of 0.4, a threshold MHWSs Ut of 180 cm s−1 and 400 cm s−1 equate to threshold friction velocities (ut of 11 cm s−1 and 23 cm s−1, respectively, using the equation:

 
formula

Wind tunnel tests have been conducted to determine threshold friction velocities for various types of desert surficial sediments (Gillette et al. 1980). Results show that threshold friction velocities in the range of 10–15 cm s−1 are substantially below the expected ranges of threshold friction velocities as derived from wind tunnel experiments; threshold friction velocities of 30–50 cm s−1 are required to erode loose, dry sand with minimal crusting, such as would occur on the surface of sand dunes or a surface covered with a layer of aeolian sand. Inasmuch as we would expect threshold velocities to be exceeded for an event of sufficient scale to drop visibility below 11 300 m in a whole area, it suggests to us that in the Yuma area the actual wind speeds controlling dust entrainment are from higher-speed gusts well in excess of mean hourly wind speeds recorded at the weather station.

For about half the dust events used in this study, “extreme wind speeds” were observed and recorded hourly on the MF-10A log sheets. A comparison between the MHWS and the mean extreme wind speed for each dust episode for which extreme wind speeds were recorded is shown in Fig. 19. From threshold MHWSs of 180 cm s−1 and 400 cm s−1, threshold mean extreme wind speeds of 450 cm s−1 and 800 cm s−1, respectively, are obtained, using the ratio between mean extreme wind speed and MHWS of 2.5 and 2.0, respectively, from Fig. 19. Threshold mean extreme wind speeds of 450 cm s−1 and 800 cm s−1 yield threshold friction velocities (ut) in the range of 25 cm s−1 and 46 cm s−1, respectively. These threshold friction velocities compare favorably with the lowest ranges of those obtained from wind tunnel threshold friction velocities required to erode loose dry sand [30–50 cm s−1; Gillette et al. (1980)]. We postulate that for those dust events for which MHWSs lie at or near threshold MHWS the extreme wind speeds for these same dust events, which are recorded on hourly weather observations, are of the right magnitude to compute valid threshold friction velocities.

Fig. 19.

(a) Ratio of mean extreme hourly wind speed to MHWS u, plotted as a function of MHWS, for all 1948–78 dust episodes occurring at Blythe, California, for which mean extreme hourly wind speed was recorded. (b) Ratio of mean extreme hourly wind speed to MHWS u, plotted as a function of MHWS, for all 1948–78 dust episodes occurring at Yuma, Arizona, for which mean extreme hourly wind speed was recorded.

Fig. 19.

(a) Ratio of mean extreme hourly wind speed to MHWS u, plotted as a function of MHWS, for all 1948–78 dust episodes occurring at Blythe, California, for which mean extreme hourly wind speed was recorded. (b) Ratio of mean extreme hourly wind speed to MHWS u, plotted as a function of MHWS, for all 1948–78 dust episodes occurring at Yuma, Arizona, for which mean extreme hourly wind speed was recorded.

Conclusions

Mean hourly wind speeds (MHWSs) recorded during blowing dust events (as defined by visibility falling below 11 300 m) at Yuma, Arizona, and Blythe, California during the period 1948–78 lie principally between 300 and 1300 cm s−1 at Yuma and between 400 and 1600 cm s−1 at Blythe, with the mean of MHWSs being 850 cm s−1 and 929 cm s−1 at each station, respectively.

Significant numbers of blowing dust episodes have occurred during every month of the year at both stations, but the highest number of events during the period of study haveoccurred during February–April, July–August, and November. Mean frequency of dust events was 27 per year at Yuma and 13 per year at Blythe.

Threshold MHWSs for blowing dust increase with increases in prior precipitation in proportion to the inhibiting effect of the vegetation canopy, which blooms following periods of increased precipitation. Because of the time required for the vegetation canopy to fully develop, and the persistence of the vegetation canopy once developed, correlation between threshold wind speed and prior precipitation is highest for 4–6-month windows of prior precipitation at both stations.

Many of the dust events associated with unusually low MHWSs recorded at Blythe and Yuma during the 1948–78 time frame are probably also associated with human-induced disturbances of the desert floor. These events can be correlated in time with periods of heavy construction on interstate highways and with preparation of additional land surfaces for irrigation.

We have used the 5-month (Blythe) and 6-month (Yuma) windows of prior precipitation to observe or compute threshold MHWSs. Threshold MHWSs are very similar at both stations during times of drought (zero intercept at 150–180 cm s−1 for all events including low wind speed events; zero intercept at 400 cm s−1 for the main body of events). For comparable levels of prior precipitation, however, higher-threshold MHWSs are derived at Blythe as compared to Yuma, probably reflecting a better-developed stand of vegetation during most times because of the higher long-term mean of precipitation at Blythe.

Since threshold MHWSs for blowing dust lie well below the recorded wind speeds during most dust events at both stations, we conclude that windstorms of sufficient strength to blow dust are more frequent at Yuma than at Blythe, but that windstorms occurring at Blythe have, on the average, higher wind speeds. We also conclude that the overwhelming majority of strong-wind events occurring at both stations will be accompanied by blowing dust.

Because threshold MHWSs lie significantly below MHWSs recorded in most dust events at Yuma, we conclude that mean annual precipitation could increase substantially at this station without significantly reducing the occurrence of blowing dust episodes. At Blythe, on the other hand, substantial increases in future mean annual precipitation would raise the threshold MHWS to the point that conditions for blowing dust would be substantially reduced.

Many of the infrequently occurring periods of elevated precipitation correlate in time with El Niño–Southern Oscillation events, which typically repeat every 2–8 yr. This correlation is especially pronounced at Blythe where seven of nine periods of anomalously high precipitation correlate with El Niño–Southern Oscillation events.

Susceptibility to blowing dust varies areally in the Sonoran Desert region. Mountains, irrigated agricultural areas, wetlands, rivers and lakes, and urban areas are characterized by low susceptibility; basin areas underlain by bolson deposits and pediments have susceptibility ranging from low to moderate; and dry lake beds and nonirrigated portions of the Colorado River delta are highly susceptible to blowing dust. Disturbance of the desert soil surface greatly increases susceptibility to blowing dust, particularly in areas of desert pavement underlain by fine-grained sediments.

Average MHWSs, and threshold MHWSs for blowing dust, vary with wind direction at Blythe and Yuma. This variation introduces a noise level into the effects of vegetation and prior precipitation on threshold wind speeds for blowing dust. It is obviously related to variations in the susceptibility of upwind source areas in most instances, but in one or more instances it may possibly be related to storm type. At Yuma threshold MHWSs are less than 566 cm s−1 for westerly and easterly winds and greater than 566 cm s−1 fornortherly and southeasterly winds. Removal of this “noise” due to variation in susceptibility of source areas results in a higher correlation between threshold MHWS and prior precipitation.

Threshold friction velocities computed from MHWSs recorded during dust events are significantly lower than those obtained from wind tunnel experiments over loose desert soils. Threshold friction velocities computed from mean extreme wind speeds, on the other hand, are comparable to those obtained from wind tunnel experiments. We conclude that the mean extreme wind speeds recorded during blowing dust episodes yield friction speeds in line with those required for entrainment of dust in the atmosphere in significant amounts.

Acknowledgments

Log sheets from Changery’s (1983) study were obtained from Anthony J. Brazel, Laboratory of Climatology, Arizona State University, who generously provided copies of copies of these documents, the originals of which have been lost. Valuable knowledge of local soil conditions was provided by Bobbi Stevenson, USDA Soil Conservation Service for Yuma, and by Gerald Davis, Palo Verde Irrigation District for Blythe. A historical summary of weather at Yuma was provided by Chief Warrant Officer Davis, weather officer at the Yuma Marine Corps Air Station. An examination of early weather logbooks kept by the Meteorological Sergeant at Fort Yuma was kindly arranged by Carol Brooks, Arizona Historical Society.

REFERENCES

REFERENCES
Brazel, A. J., 1989: Dust and climate in the American southwest. Paleoclimatology and Paleometeorology: Modern and Past Patterns of Global Atmospheric Transport, M. Leinen and M. Sarnthein, Eds., NASA ASI Series C, Vol. 282, Kluwer Academic Publishers, 65–96.
Changery, M. J., 1983: A dust climatology of the western United States. U.S. Nuclear Regulatory Commission Rep. NUREG/CR-3211, 25 pp. [Available from National Climatic Data Center, 151 Patton Ave., Room 120, Asheville, NC 28801-5001.]
.
Chepil, W. S., 1956: Influence of moisture on erodibility of soil by wind. Proc. Soil Sci. Soc., 20, 288–292.
Cole, J. E., G. T. Shen, R. G. Fairbanks, and M. Moore, 1992: Coral monitors of El Niño/Southern Oscillation dynamics across the equatorial Pacific. El Niño: Historical and Paleoclimatic Aspects of the Southern Oscillation, H. Diaz and V. Markgraf, Eds., Cambridge University Press, 349–375.
Crosswhite, F. S., and C. D. Crosswhite, 1982: The Sonoran Desert. Reference Handbook on the Deserts of North America, E. L. Bender, Ed., Greenwood Press, 163–295.
Gillette, D. A., and R. Passi, 1988: Modeling dust emission caused by wind erosion. J. Geophys. Res., 93, 14 233–14 242.
——, J. Adams, A. Endo, D. Smith, and R. Kihl, 1980: Threshold velocities for input of soil particles into the air by desert soils. J. Geophys. Res., 85, 5621–5630.
MacKinnon, D. J., D. F. Elder, P. J. Helm, M. F. Tuesink, and C. A. Nist, 1990: A method of evaluating effects of antecedent precipitation on duststorms and its application to Yuma, Arizona, 1981–1988. Climate Change, 17, 331–360.
Quinn, W. H., V. T. Neal, and S. E. Antunez de Mayo, 1987: El Niño occurrences over the past four and a half centuries. J. Geophys. Res., 92, 14 449–14 461.
Stockton, P. H., and D. A. Gillette, 1990: Field measurement of the sheltering effects of vegetation on erodible land surfaces. Land Degradation and Rehabilitation, Vol. 2, John Wiley and Sons, 77–85.
YumaDaily Sun, 1978: 2 August.

Footnotes

Corresponding author address: Dr. Troy Leon Holcombe, NOAA/National Geophysical Data Center, 325 Broadway, Boulder, CO 80303-3328.