Anthony J Brazel Andrew W Ellis

Introduction

The Central Arizona and Phoenix LTER (CAP LTER) is one of two urban LTERs in the world network (Grimm et al. 2000; see http://caplter.asu.edu). Many LTER sites display a detectable climatic signal related to the El Nino-Southern Oscillation (ENSO) phenomenon (Greenland 1999). The purpose of this chapter is twofold: (1) to provide some insight into the role of the tropical Pacific Ocean as a driver of several climatic (and thus, ecologically related) variables in the CAP LTER location of central Arizona, and (2) to suggest the linkages of ENSO events to selected ecosystem processes near and within the geographical region of CAP LTER (figure 7.1a).

From past studies, it is clear that the seasonal and annual climate regimes of the southwestern United States, particularly water-related parameters, are linked to the periodicities and anomalies of what is known as the Multivariate ENSO Index (MEI) and Southern Oscillation Index (SOI) (e.g., Wolter 1987; Molles and Dahm 1990; Redmond and Koch 1991; Woolhiser and Keefer 1993; Wolter and Timlin 1993; Cayan and Redmond 1994; Redmond and Cayan 1994; Cayan et al. 1999; Redmond and Cayan 1999; Simpson and Colodner 1999; Redmond 2000; and Mason and Goddard 2001). In Arizona, and especially in the CAP LTER region, precipitation is bimodal during the year with peaks in winter (mostly midlatitude-derived frontal storms) and in mid-to-late summer, mostly in the form of convective thunderstorms during the North American monsoon season. Recent studies show a strong connection between ENSO and winter moisture in Arizona, such that it is even possible to forecast impending conditions in advance (Pagano et al. 1999). These studies have established relationships between the climate of the southwest-

Figure 7.1 Geographical region of the Central Arizona and Phoenix Long-Term Ecological Research site (CAP LTER). (a) CAP LTER boundaries within Arizona. (b) CAP LTER study area.

ern United States and ENSO by demonstrating monthly and daily timescale effects on inputs of moisture and resultant streamflow in Arizona (e.g., Molles and Dahm 1990; Cayan et al. 1999; and Simpson and Colodner 1999). The synoptic- and large-scale circulation patterns associated with anomalies of MEI/SOI in the southwestern United States provide additional insight into regional forces that drive the CAP-LTER climate (e.g., Redmond and Koch 1991). Generally, when the warm phase of the tropical Pacific Ocean occurs (El Niño, thus negative SOI, positive MEI), across the Southwest precipitation is generally anomalously high. Conversely, when the cool phase occurs (La Niña, thus positive SOI, negative MEI), an input of moisture that is less than normal appears to be the case in the Southwest (e.g., Cayan et al. 1999). Generally, when neither El Niño nor La Niña occurs, it is unclear what the precipitation regime across the Southwest will be in relation to ENSO.

Increased daily, monthly, seasonal, and annual-to-decadal moisture or extended dry periods have important implications for the Southwest and the CAP LTER ecosystem. We suggest several potential linkages, and review three specific examples in this chapter: (1) studies related to the hanta virus and ENSO that have been conducted for the nearby Four Corners Area and New Mexico by researchers of the Sevilleta LTER site (e.g., Parmenter et al. 1999) and others; (2) our interpretation of ENSO phases in relation to past and ongoing stream ecological research on Sycamore Creek at the CAP LTER (analyzing data from Grimm 1993); and (3) possible impacts of ENSO on concentrations of river constituents routinely observed by the U.S. Geological Survey for the Phoenix region. (CAP LTER researchers have recently constructed a detailed nitrogen budget for this region; see Baker et al. 2001). The first example suggests linkages in a trophic cascade from inputs of moisture, to increased vegetative cover and insects, to abundance of fleas and mammals, to human plague incidences perhaps spanning over more than a year for the cascade. The second suggests linkages from inputs of moisture to a typical individual desert stream and its ecological conditions spanning short periods to months. The third example suggests linkages from seasonal inputs of moisture to larger river systems, and observed winter variations of many stream constituents over a quarter of a century upstream and downstream of the built-up urbanized and agricultural sector of the CAP LTER area.

ENSO effects are most obvious on the winter climate, but also anomalous conditions in the tropical Pacific Ocean in winter may influence the summer monsoon season in Arizona. Breaks and bursts in the monsoon and overall monsoon seasonal intensity are strongly related to flash-flood risks, local storm damage, dust storm frequencies, the urban heat island, human comfort and energy demand, and vegetation green-up and biomass. Thus, it is likely that climate impacts on natural and human components of CAP LTER are substantial at the timescale of MEI/SOI variations. Scientists in CAP LTER are just beginning to study ecosystem response and feedback to a host of natural and human-induced processes. The human dimension acts as a driver to ecosystem change and, in turn, is affected by these changes (Brazel et al. 2000; Collins et al. 2000; and Grimm et al. 2000).

This chapter certainly does not purport to explain all of the cascading effects on CAP LTER at a quasi-quintennial timescale. However, we explicitly relate indices of the warm and cool phase of the tropical Pacific Ocean to several climatic vari ables in CAP LTER. The work is intended to outline those relationships that deserve further study to expand our knowledge of cascading climate effects on the urban ecosystem. In this analysis, data are expressed at the monthly timescale for the period 1951-1999 using thermal and moisture climatic variables. The variables include (1) maximum, minimum, mean, and range of temperature in rural and urban locales, (2) regional temperature and precipitation, (3) evaporation at reservoirs in or near CAP LTER, (4) snowfall at the upper end of a major watershed important to CAP LTER, (5) simulated soil moisture surpluses and deficits in CAP LTER, and (6) streamflow within a representative natural stream in CAP LTER (Sycamore Creek). This analysis hopefully will assist researchers in (1) the development of hypotheses for retrospective analyses of urban ecosystem dynamics, (2) the recognition of the climate context of field experiments conducted at CAP LTER, (3) the climate context of a repetitive 3- to 5-year snapshot ecosystem survey of 200 points at CAP LTER in rural-urban locales, and (4) the illustration of external drivers on the local ecosystem, thus making links from CAP LTER to large scale (global and regional) change more explicit. Our analysis also provides composite views of regional atmospheric circulation features that are expected with anomalies in the SOI and MEI. Thus, regional explanations of local area effects are more easily facilitated and understood.

Method of Analysis Teleconnection Indexes

In examining the linkages between Pacific teleconnections and CAP LTER, two teleconnection indexes were correlated with climate characteristics (thermal and moisture) across central Arizona. Monthly values of the Southern Oscillation Index (SOI) and the Multivariate El Niño-Southern Oscillation Index (MEI) were collected for the 49-year period 1951-1999 (e.g., March in figure 7.2). The beginning date of the study period is confined by the SOI and the MEI records, whereas records of climatic data confined the ending date. SOI values, representing differences in monthly sea level pressure values across the southern Pacific Ocean, were obtained from the U.S. Climate Prediction Center (CPC). MEI values were obtained directly from K. Wolter of the Climate Diagnostics Center (CDC) of the National Oceanographic and Atmospheric Administration (NOAA).

The MEI is explicitly used in our analysis. The MEI is built from six observed variables across the tropical Pacific. The variables are (1) sea-level pressure (P), (2) zonal (U) and (3) meridional (V) components of the surface wind, (4) sea surface temperature (S), (5) surface air temperature (A), and (6) total fraction of the sky covered by cloud (C). The MEI is calculated separately for each of twelve moving bimonthly seasons. After spatially filtering the individual fields into clusters, the MEI is calculated as the first unrotated Principal Component (PC) of all six observed fields combined. In doing this, the total variance of each field is first normalized prior to the extraction of the first PC on the covariance matrix of the com-

Figure 7.2 Southern Oscillation Index (SOI) and Multivariate El Nino-Southern Oscillation Index (MEI) values during March for the period 1951-1999.

bined fields. Finally, the computed MEI values are standardized with respect to the 1950-1993 reference period. Negative values of the MEI represent the cold ENSO phase, or La Niña, whereas positive values represent the warm ENSO phase, or El Niño. The sea level pressure (P) loadings characterize the Southern Oscillation. For example, negative MEI values (La Niña) are derived from negative pressure anomalies in the west and positive pressure anomalies in the east. The latitudinal (U) component of the surface wind corresponds to east-west wind direction anomalies along the equator near the international dateline. The meridional (V) component of the surface wind corresponds to north-south wind direction anomalies north of the equator across the Pacific Ocean, largely reflecting oscillation of the Inter-Tropical Convergence Zone (ITCZ). Sea (S) and air (A) surface temperatures indicate the typical ENSO pattern of temperature anomalies from the western South American coastline to the date line. Finally, total cloudiness (C) across the central equatorial Pacific versus over the Philippines and north of Australia indicate the migration of convective activity.

Thermal Data

To examine the covariance of MEI with near-surface air temperatures across the area, U.S. climate division data for central Arizona (division 6; figure 7.1a) were obtained from the National Climatic Data Center (NCDC; NOAA 1983a, b). Monthly values represent mean monthly temperatures as calculated from all regional stations at which daily maximum and minimum near-surface air tempera-

tures are recorded (over 30 stations in the division 6 region). As such, the data represent the temporal variation within the general monthly lower atmospheric temperature record across the region as a whole.

To more closely examine MEI associations with temperature within the Phoenix urban area, daily maximum and minimum temperature values for a central Phoenix station (Sky Harbor Airport, Phoenix AP in figure 7.1b) and rural Wickenburg (figure 7.1b) were obtained for the period of study. Data were taken as a subset of the Summary of the Day database of the NCDC. Using daily maximum and minimum temperature data, daily temperature range values were calculated (maximum minus minimum), as were daily rural-urban differences (urban minus rural) in maximum and minimum temperature. All daily temperature values were translated into monthly means.

Moisture Data

With a burgeoning population in the desert setting of CAP LTER, water resources are of constant concern (Carter et al. 2000). Given the general convective nature of regional precipitation, and therefore large spatial inhomogeneity, precipitation data from the climate divisional records were obtained from NCDC to represent the variation in monthly mean precipitation across the CAP LTER region as a whole, again using over 30 sites. Since the water resources of the area are also dependent on spring snowmelt across the higher terrain to the north, daily snowfall values for Flagstaff in northern Arizona (figure 7.1a) were extracted from the Summary of the Day database of NCDC. Daily values were summed to monthly totals through the period of study to correlate with MEI values.

To translate monthly thermal and moisture variables into aspects of the climatic water conditions for the region, monthly divisional temperature and precipitation means were used to calculate mean monthly soil moisture values. The Thornth-waite-Mather climatic water budget technique (Thornthwaite and Mather 1955; Mather 1978) was used as a first approximation to produce monthly soil moisture surplus and deficit values, of which only deficit values were considered because of the infrequency of soil moisture surpluses in the CAP LTER area. To further represent the temporal variability in the water resources, daily streamflow values for Sycamore Creek (important to CAP LTER objectives; Grimm 1993; figure 7.1b) were totaled to monthly values for the period of study. Streamflow data were obtained from the U.S. Geological Survey stream gauge database (www.usgs.gov). Later we illustrate the links of SOI/MEI and streamflow oscillations to processes of stream ecology (Grimm 1993; Grimm et al. 1997).

To represent the temporal variability of evaporation from an open water surface (e.g., reservoir)—a parameter of extreme interest to water managers—data for daily pan evaporation at two reservoirs close to the CAP LTER area were obtained from the Summary of the Day database of the NCDC. Daily pan evaporation totals at Roosevelt Dam northeast of Phoenix and San Carlos Dam east of Phoenix (figure 7.1a) were totaled to monthly values for the period of the study.

Quality Assurance of Data

The SOI and MEI data are quality controlled and complete for the period 1951 through 1999 to form a comprehensive data set for the study period (figure 7.2). Likewise, U.S. climate division data are complete for the full period of study. Daily temperature data for Phoenix and Wickenburg, snowfall data for Flagstaff, and evaporation data at the Roosevelt and San Carlos Dams extend through the period of study, but are not entirely complete. Daily streamflow data are available for the period 1961-1997. In processing the incomplete records of daily data, for each month of the period of study a threshold value of 90% coverage of daily data was required. Otherwise, the monthly data value was labeled as missing. Subsequently, 90% coverage of monthly data was required for inclusion in covariance calculations with MEI values, a threshold value we accept as representative based on the work of Stooksbury et al. (1999).

As MEI values are representations of monthly deviations in standard values, all climatic variables were standardized to monthly Z-scores. In calculating Z-scores, the mean of each distribution was subtracted from each observation and then subsequently divided by the standard deviation of the distribution. The products were distributions of monthly climatic variables through the period of study (19511999) where each distribution has a zero mean and number units of standard deviations.

Analysis of Covariance

To assess the covariance between the SOI and MEI, simple correlation coefficients (r) were calculated to determine the extent of the covariance of monthly values of each index for the period of study. Likewise, to assess the extent of the covariance between each of the teleconnection indices and variables representing the climate of the CAP LTER area, correlation coefficients were calculated. Concurrent relationships were tested as well as lagged relationships; a monthly MEI value was correlated with each climate variable for each month as well as for each month of the subsequent 11-month period (in a manner similar to Greenland 1999). Finally, for each calculated correlation coefficient, a t-test for the significance of correlation was determined to highlight relationships of significance.

Physical Forcings

Lastly, to gain a physical understanding of any significant statistical relationships between MEI values and CAP LTER climate variables, the characteristics of the larger background synoptic atmosphere were diagnosed. For those intra-annual periods exhibiting strong statistical relationships, synoptic atmospheric composites and anomalies were constructed using only data from years with extreme MEI values. An extreme year was identified as possessing an MEI value in either the 10th or 90th percentile, which is to say those years possessing one of the five highest and five lowest MEI values for the 49-year period of study.

Figure 7.3 Correlation between monthly values of the SOI and MEI indices. Each correlation is significant at the 99% level.

Using data from the National Center for Environmental Prediction (NCEP) re-analysis data set (Kalnay et al. 1996), simple composites of the synoptic atmosphere on a 2.5°-latitude by 2.5°-longitude spatial resolution were constructed. Composites of 500-mb geopotential height (large-scale atmospheric flow), 850-mb air temperature (regional thermal conditions), and 850-mb specific humidity (regional moisture conditions) were created. The purpose of these specific composites is to illustrate the anomalies in the synoptic atmosphere affecting the CAP LTER climate, and, in turn, the ecosystem, and driven by the remote atmospheric anomalies represented by the MEI.

Results

SOI-MEI Covariance

Correlation coefficients measuring the significance of the covariance between SOI and MEI values (figure 7.3) indicate a rather significant inverse relationship throughout the annual period. The covariance is highly significant during late summer through early spring. However, the correlation between the two indexes decreases dramatically in May and June just before the typical onset of the monsoon in the southwestern United States. There is a good agreement between the two indexes. However, the MEI is correlated slightly better with the climate parameters of central Arizona than is the SOI, particularly during May and June. This seems reasonable, since more descriptive parameters of activity in the tropical Pacific Ocean are included in the MEI. For this reason, discussion from this point forward will be confined to the relationship between the MEI and the variability of climate in the CAP LTER area.

Climate of Central Arizona and Phoenix Long-Term Ecological Research Site 125 MEI-Temperature Associations

It is clear that there is very little association between MEI values and mean monthly temperature across CAP LTER (table 7.1). Inverse relationships are common in fall through early spring (October-March), whereas positive relationships exist from spring through late summer (April-September). Inverse relationships are indicative of decreased (increased) temperatures during El Niño (EN) [La Niña (LN)] events of eastern tropical Pacific Ocean warming (cooling). From spring through late summer, positive relationships indicate increased (decreased) temperatures during EN (LN) conditions. Still, there are no significant associations between the MEI and mean monthly CAP LTER temperatures.

We examined MEI associations with daily maximum and minimum temperatures for an urban (Phoenix) and a rural (Wickenburg) location. This provides greater insight into associations between the MEI and CAP LTER temperatures. It is apparent that MEI mean temperature relationships are weakened by the fact that the relationship between the MEI and maximum daily temperature tends to be opposite to that between MEI and minimum daily temperature. This is evidenced by the significance of the associations between MEI values and monthly means of daily temperature range (table 7.1). High (EN) [low (LN)] MEI values are associated with decreased (increased) maximum temperatures during the period October through March in Phoenix, and for every month of the year, but during August at rural Wickenburg. February and March relationships are significant at each location, as is the November relationship at Wickenburg. At Phoenix, a positive relationship exists from April through September and is significant during July, the typical month of the commencement of the monsoon circulation. However, from spring through summer, EN (LN) conditions are associated with increased (decreased) maximum temperatures.

Throughout the year, and most obvious in spring (March-June) and fall (October-November), a significant positive relationship exists between the MEI and minimum temperatures at Phoenix, whereby EN (LN) conditions are associated with higher (lower) minimum temperatures. The same positive relationship between MEI values and minimum temperatures exists at Wickenburg from the middle of the monsoon season (August) through early spring (March) and is significant during the monsoon season (August-September). However, the relationship weakens in spring and early summer, and it reverses significantly just prior to the monsoon (June-July). During this period, minimum temperatures at Wickenburg are inversely associated with the MEI, where EN (LN) conditions are associated with lower (higher) minimum temperatures.

Within the MEI-temperature range correlation, the products of the associations between the MEI and maximum and minimum daily temperatures can be seen. An inverse relationship between MEI values and temperature range in Phoenix exists throughout the year and is most significant (October-June) outside the monsoon season. An inverse relationship suggests that EN (LN) conditions are associated with decreased (increased) daily temperature range. The inverse associations exhibit a similar intra-annual pattern at Wickenburg (August-May), but are not quite as strong, with significant inverse relationships occurring only during the periods

Table 7.1 Correlation between monthly MEI values and monthly mean regional temperature and monthly means of daily maximum and minimum temperature, daily temperature range, and temperature difference for and between Phoenix and Wickenburg

Month

Regional Mean

Phoenix

Wickenburg

Phoenix-Wickenburg

Max

Min

Range

Max

Min

Range

Max

Min

January

-0.04

-0.12

0.19

-0.28*

-0.10

0.07

-0.14

0.01

0.22

February

-0.11

-0.27*

0.20

-0.49**

-0.31*

0.21

-0.51**

0.14

0.06

March

-0.13

-0.26*

0.27*

-0.61**

-0.29*

0.16

-0.45**

0.18

0.20

April

0.10

0.07

0.36**

-0.49**

-0.05

-0.07

-0.01

0.16

0.42**

May

0.12

0.03

0.37**

-0.42**

-0.08

0.04

-0.15

0.35**

0.39**

June

0.01

0.01

0.31*

-0.44**

-0.16

-0.27*

0.15

0.22

0.48**

July

0.09

0.25*

0.22

-0.02

-0.04

-0.29*

0.25*

0.27*

0.39**

August

0.21

0.10

0.20

-0.15

0.05

0.23*

-0.17

0.10

-0.01

September

0.13

0.09

0.21

-0.19

-0.08

0.24*

-0.29*

0.20

-0.02

October

-0.03

-0.05

0.25*

-0.29*

-0.17

0.11

-0.24*

0.19

0.22

November

-0.10

-0.22

0.26*

-0.52**

-0.30*

0.13

-0.40**

0.21

0.18

December

-0.02

-0.14

0.21

-0.33**

-0.16

0.19

-0.26*

0.09

0.09

*Significance level of 95%.

**Significance level of 99%.

*Significance level of 95%.

**Significance level of 99%.

September through December and February through March. However, a positive relationship between the MEI and temperature range exists at Wickenburg during July, when EN (LN) conditions are associated with larger (smaller) temperature ranges.

In examining the correlation between MEI values and urban-rural differences in daily temperature (table 7.1), a typical method used in urban heat island studies and important in the energy service sector (Brazel et al. 1993), the greatest association of the MEI is with urban-rural minimum temperature differences (time of day when heat islands are more pronounced). High (low) MEI values associated with an EN (LN) situation are correlated with large (small) differences between Phoenix and Wickenburg minimum temperatures year round, but most significantly during the spring and summer period of April through July. Because urban minimum temperatures in Phoenix are nearly always milder than those at surrounding rural locations, the positive relationship suggests that minimum temperatures are greater than usual during EN and lesser during LN at Phoenix than at Wickenburg. This is supported by the stronger relationship between the MEI and minimum temperature at Phoenix than at Wickenburg, where the relationship actually reverses in June and July (table 7.1).

For much of the year a positive relationship between MEI values and urban-rural maximum temperature differences exists, most significantly in May and July (table 7.1). Positive correlation between MEI and urban-rural maximum temperature differences suggests that high (low) MEI values associated with EN (LN) conditions are associated with large (small) differences in temperature. Although not always the case, Phoenix maximum temperatures are typically warmer than those at surrounding moist rural locations. As such, the relationship suggests that during EN (LN) situations, especially in May and July, Phoenix maximum temperatures are generally larger than those at Wickenburg by an amount that is greater than (less than) usual. The significance to the urban ecosystem of the MEI/SOI forcers has not previously been demonstrated for CAP-LTER. Currently, these urban-rural climate differences and their impacts on a host of processes (e.g., human stress, heat stress on plants, energy consumption, arthropod abundance, cotton and dairy production) are the focus of a "feedbacks" subgroup of CAP-LTER researchers (L. A. Baker et al., unpubl. data, 2002).

MEI-Precipitation Associations

Correlations between MEI values and mean CAP LTER (climate division 6) monthly precipitation indicate a positive relationship during the fall through spring, most significantly during the months November-December, February-March, and May (table 7.2). During these periods, high (low) MEI values corresponding to EN (LN) conditions are associated with greater (small) amounts of precipitation across the CAP LTER area. The same is true of the relationship between the MEI and snowfall in Flagstaff in late winter (February-March; table 7.2). A significant inverse relationship between MEI values and mean CAP LTER precipitation exists in July. This indicates that during the month in which the monsoon season typically begins, EN (LN) conditions are associated with a(n) decrease (increase) in precipitation.

Table 7.2 Correlation between monthly MEI values and monthly CAP LTER area precipitation, Flagstaff snowfall, and CAP LTER area soil moisture deficit, pan evaporation, and streamflow

Sycamore Creek

Regional Flagstaff Soil Moisture Roosevelt San Carlos Stream-Month Precipitation Snowfall Deficit Evaporation Evaporation flow

Sycamore Creek

Regional Flagstaff Soil Moisture Roosevelt San Carlos Stream-Month Precipitation Snowfall Deficit Evaporation Evaporation flow

January

0.09

0.05

-0.17

0.07

-0.23

0.18

February

0.60**

0.26*

-0.39**

-0.12

-0.52**

0.36**

March

0.53**

0.27*

-0.41**

-0.38**

-0.54**

0.40**

April

0.06

-0.05

-0.29*

-0.41**

-0.29*

0.45**

May

0.36**

-0.32*

-0.50**

-0.56**

0.32*

June

-0.10

-0.14

-0.44**

-0.26*

0.23*

July

-0.37**

0.33**

0.26*

0.12

0.15

August

0.09

-0.03

0.13

-0.09

0.07

September

-0.07

0.11

0.02

-0.16

-0.17

October

0.22

-0.13

-0.12

0.24*

November

0.30*

-0.16

-0.31*

-0.03

-0.37**

0.43**

December

0.32*

-0.05

-0.26*

0.03

-0.12

0.25

*Significance level of 95%. **Significance level of 99%.

*Significance level of 95%. **Significance level of 99%.

MEI-Climatic Water Associations

In translating MEI associations with temperature and precipitation within the CAP LTER area into associations with climatic water variability (table 7.2), it is evident that the MEI is significantly associated with climatic water parameters during the period late fall through early summer. Soil moisture across the CAP LTER area exhibits a significant inverse relationship with MEI values during the period November through May, excluding January (table 7.2). The inverse relationship indicates that when MEI values are high (low), indicating EN (LN) conditions, soil moisture deficit values are low (high). In other words, under EN (LN) conditions, when precipitation tends to be increased (decreased) and temperatures tend to be decreased (increased), the soil moisture deficit typical of the region is decreased (increased). The relationship between MEI and the soil moisture deficit becomes significantly positive in July (table 7.2), indicating that EN (LN) conditions at the inception of the monsoon season are associated with increased (decreased) soil moisture conditions.

As in the case of the soil moisture deficit, an inverse relationship exists between the MEI and pan evaporation at the Roosevelt (significant March-June) and San Carlos (significant November, February-June) reservoirs (table 7.2). High (low) MEI values associated with EN (LN) conditions are linked to decreased (increased) evaporative loss from an open water surface. As with soil moisture deficit, MEI-

evaporation relationships become positive in July, significantly so at the Roosevelt reservoir (table 7.2). As such, EN (LN) conditions during July are associated with increased (decreased) evaporative rates.

Taken together, MEI relationships with temperature, precipitation, soil moisture deficit, and evaporative loss lead to an association with streamflow. For nearly the entire year, a positive relationship exists between MEI values and streamflow at Sycamore Creek, significantly so during the period October through June, excluding January (table 7.2). High (low) monthly MEI values, indicating EN (LN) conditions, are associated with higher (lower) monthly streamflow. During the monsoon season, the relationship weakens (July-August) and actually reverses direction (September). The implications for these relationships are discussed in the section entitled Sycamore Creek Stream Ecology.

Summary

The correlation between MEI and various CAP LTER climate variables examined within this study indicate that the strongest relationships occur in late winter and spring, most significantly in March (table 7.3). In general during this period, high MEI values indicative of EN conditions are associated with (1) decreased maximum temperatures, (2) increased minimum temperatures, (3) decreased temperature ranges, (4) increased urban-rural temperature differences, (5) increased precipitation (including snowfall over higher terrain to the north), (6) increased soil moisture, (7) decreased evaporative losses, and (8) increased streamflow. Low MEI values (LN conditions) are associated with opposite responses. MEI-climate associations are high in midautumn as well, with generally the same strength of the relationships as in spring. Of additional interest is the reversal in the nature of the MEI relationships with precipitation and many of the climatic water variables in July, the month during which the annual monsoon typically begins. In this case, high MEI values indicative of EN conditions are associated with decreased precipitation and increases in soil moisture deficits and pan evaporation rates (table 7.3).

Atmospheric Dynamics

To better understand the physical forcing that drives the variation in CAP LTER climate with EN and LN conditions, synoptic atmospheric composites of March and July were constructed for the five strongest EN years (highest MEI values) and the five strongest LN years (lowest MEI values). For March EN, these are the years 1958, 1983, 1987, 1992, 1998 (July: 1972, 1982, 1983, 1987, 1997); for March LN, 1951, 1956, 1971, 1974, 1976 (July: 1954, 1955, 1956, 1964, 1971). In March, it is apparent that the mean 500-mb height pattern is shifted more to the south during EN years than during LN years (figures 7.4a-c). The ridge/trough pattern (figures 7.4a,b) is very similar, however the magnitudes are considerably different, such that during LN (EN) years the Pacific ridge is strengthened (weakened). LN (EN) years seem to be associated with higher (lower) 500-mb heights across the southwestern United States. The strengthened (weakened) Pacific ridge during LN (EN) years is

Table 7.3 Direction of monthly correlation between MEI values and CAP LTER climatic parameters. Positive (P) relationships indicate an increase (decrease) in the variable under EN (LN) conditions, whereas inverse relationships (N) indicate a decrease (increase) under EN (LN) conditions.

Temperature

Mean regional N

Phoenix max N

Wickenburg max N

Phoenix min P

Wickenburg min P

Phoenix range N

Wickenburg range N

Phoenix-Wickenburg max P

Phoenix-Wickenburg min P

Precipitation

Mean regional P

Flagstaff snowfall P

Climatic Water Variables

Soil moisture deficit N

Pan evaporation-Roosevelt P

Pan evaporation-San Carlos N

Streamflow-Sycamore Creek P

Significant (95%) correlation is bold and underlined.

likely to be associated with a more northerly (southerly) storm track, and is associated with a relatively warmer (cooler) and drier (moister) lower atmosphere in March (figure 7.5), accounting for the associations between the MEI and CAP LTER-area climate variables.

LN (EN) conditions during July are associated with a strengthened (weakened) 500-mb ridge across the western United States and 500-mb trough across the eastern Pacific Ocean (figure 7.6). July marks the beginning month of the monsoon season in the southwestern United States, and it is initiated by northward displacement of the subtropical ridge (Adams and Comrie 1997). The stronger (weaker) western U.S. 500-mb ridge during LN (EN) years is associated with a warmer (cooler), but moister (drier) lower atmosphere (figure 7.7). This is opposite to the drier (moister) atmosphere associated with LN (EN) conditions during March (figure 7.5).

Discussion of Results

Our results and others (e.g., Simpson and Colodner 1999) point to significant climate connections between the southwestern United States and periodicities of ENSO as represented by the MEI and SOI. In fact, climate responses may be

N

N

P

P

P

P

P

P

N

N

N

N

N

P

P

P

P

P

P

N

N

N

N

N

N

N

N

N

P

N

N

N

N

P

P

_P

_P

P

P

P

P

P

_P

P

P

P

N

P

N

N

P

P

P

P

P

N

N

_N

N

N

N

N

N

N

N

N

N

N

N

N

P

P

N

N

N

N

N

P

P

P

P

P

P

P

P

P

P

P

P

P

P

_P

P

P

N

N

P

P

P

P

P

P

P

N

N

P

N

P

P

P

P

P

N

N

N

N

N

_N

N

N

P

N

P

N

N

N

N

N

N

N

N

P

P

P

N

N

P

N

N

N

N

N

P

N

N

N

N

N

P

P

P

P

P

P

P

N

P

P

P

Figure 7.4 Mean March 500-mb height during the five strongest years of (a) El Niño and (b) La Niña. (c) their differences taken as El Niño (EN) minus La Niña (LN); negative differences are indicated by a dashed line.

greatly predictable at seasonal timescales (Pagano et al. 1999). The cascading-like effects through the climate system—from the Pacific tropical ocean temperatures, to southwestern U.S. circulation dynamics, to central Arizona seasonal thermal and hydrological regimes—are quite pronounced for the fall/winter/spring time frame. Weak inverse connections to EN are even evident for the North American monsoon regime of summer (EN yields dry summer; LN yields active, wet summer). At the upper level of what might be viewed as a local CAP LTER climate cascade, therefore, there exist variable processes of moisture (precipitation, evaporation, and soil moisture), local storms, and clear/cloudy day frequencies, for example. These components are likely to have strong connections to the main driver variables analyzed in this chapter (e.g., MEI yields distinct variations in thermal/moisture inputs).

Month-to-month lag effects (not shown) are also pronounced for initial MEI

Figure 7.5 Mean March 850-mb temperature (a) and moisture (b) differences between the five strongest El Niño (EN) and La Niña (LN) years. Differences are taken as EN minus LN. Negative differences are indicated by a dashed line.

anomalies and persistence of local climate responses. Typically, longer lags are most evident for moisture variables from early winter through late spring. Thus, for example, a large positive MEI anomaly is felt from early winter through late spring as increased snow packs in the high country of Arizona, rising stream flows, increasing soil moisture, and reductions in evaporative losses. However, no association with mean temperature variations exists. Daily mean temperature shows little relation to large MEI anomaly years. As stated previously, this appears to be explained by offsetting responses of regional maximum and minimum temperatures. Thus, there is a marked variance in the temperature range: large for negative MEI, and small for positive MEI anomalies. This is likely important to ecosystem components that are sensitive to threshold values of temperature, not just to mean temperature (e.g., growing degree-day accumulation for plants; cooling degree-day accumulation for energy consumption).

A measure of the urban heat island effect (using Phoenix airport minus Wickenburg, Arizona—an urban minus rural site used previously; Balling and Cerveny 1987) shows a surprisingly significant relation to MEI anomalies. When positive anomalies occur (i.e., EN), larger urban-rural differences (bigger heat islands) are evident for the late spring and early summer months. Because there is an inverse relationship between positive MEI (EN events) and summer moisture (drier), it most likely means more clear nights and lower humidity values result. These are

Figure 7.6 Mean July 500-mb heights for the five strongest years of (a) El Niño and (b) La Niña (c) their differences, taken as EN minus LN; negative differences are indicated by a dashed line.

the sorts of local conditions that promote chances for intense heat island development in CAP LTER urban locales, especially because the heat island is predominantly a nighttime phenomenon (Brazel et al. 2000).

Ecosystem Examples

These climate responses potentially result in more complex cascades in the CAP LTER ecosystem. For example, dust storms (Brazel 1987), wildfires (Swetnam and Betancourt 1990), vegetation change (Li and Kafatos 2000), and water quantity and quality (Carter et al. 2000) are all driven by surface processes that are a combination of natural and human-impacted environmental conditions. Detailed linkages

Figure 7.7 Mean July 850-mb (a) temperature and (b) moisture differences between the five strongest El Niño (EN) and La Niña (LN) years. Differences are taken as EN minus LN. Negative differences are indicated by a dashed line.

have yet to be established to disentangle natural from human controls for many themes such as health risks, human comfort levels, energy demand variations, transportation impacts, air quality variations, local urban flooding, and the variability of water uses at local scales. On transportation issues, for example, personal correspondence with Arizona Department of Transportation Office officials and independent analysis of precipitation intensity and traffic data in the local CAP LTER area by A. Ellis (unpubl. data, 2001) have revealed that precipitation events are strongly related to urban-area traffic accident frequencies in a nonlinear fashion; that is, light rainfall initially stimulates higher accidents rates, moderate rainfall, lessening rates, and very high rainfall rates, high accident rates. Part of this pattern relates to driver behavior and levels of risk perception. Subtle differences in rainfall intensity rates that relate to accident variability may or may not be significantly related to phases of ENSO at daily-to-seasonal timescales. More research is needed on this issue. Transportation is also disrupted by blowing dust in central Arizona (Brazel 1991). An analysis of the period 1948-1984 revealed a strong link of incidences of dust storms to lack of antecedent fall/winter precipitation, little surface vegetative armoring, human disturbance of dust source areas, enhanced entrain-ment due to exceedance of threshold wind speeds, and subsequent incidences of blowing-dust-related accidents on the major interstates and other roads in central Arizona (Brazel et al. 1986; Brazel and Nickling 1987; Brazel 1991). Those authors did not relate this pattern to ENSO per se, but in retrospect it is clear that the arid antecedent years were associated with LN event years. Three more specific examples of significant ENSO impacts are provided here.

Hantavirus and ENSO

One of the startling findings recently in the health risk area in the southwestern United States (primarily Four Corners Area and New Mexico) is that of the link of the hantavirus to environmental moisture parameters, and thus possibly to ENSO phases and climate change (e.g., Hjelle and Glass 2000; and Parmenter et al. 1999; and Sprigg and Hinkley 2000). Much of this work has been conducted under the auspices of the Sevilleta LTER site in New Mexico. Increased precipitation apparently creates a trophic cascade wherein small mammal abundance (related to increased plants and insects) leads to an increase in plague hosts, which in turn results in higher hantavirus incidences. The recent 1990s EN events of 1991-1992, 1993-1994, and 1997-1998 have been linked to subsequent accelerated virus incidences (Hjelle and Glass 2000). Parmenter et al. (1999) of the Sevilleta LTER explain the cascade in a three-stage scale analysis of moisture (ENSO, regional, local), emphasizing the strong relationship to, and need to understand, local precipitation processes. They illustrated an insignificant, yet suggestive correlation of plague case rates to the previous winter moisture conditions using the SOI index. We reanalyzed Hjelle and Glass's (2000; figure 1) 1990s data, and found a statistically significant relation to a previous winter's MEI (r2 = 0.42). The cascade illustrates large moisture lag effects in this case, larger than an annual period from MEI variability to hantavirus events.

Sycamore Creek Stream Ecology

In an analysis of aquatic ecosystems related to climate change, Grimm et al. (1997) illustrate the sensitivity of a range of western U.S. streams to a number of environmental variables, among them precipitation/runoff and net basin supply, in addition to anthropogenic variables (e.g., diversions, withdrawal, and consumptive use). As indicated previously in this chapter, streamflow is correlated to MEI variability for Sycamore Creek. Specific to Sycamore Creek is Grimm's analysis (Grimm 1993) of hydrological characteristics of extreme wet and dry years, annual runoff, number of floods and other stream-specific conditions (wetness and dryness durations relative to an 11-m3/s peak discharge threshold). We reanalyzed Grimm's (1993) data set of 5 wet and 5 dry years in terms of the MEI index to shed light on the more regional and hemispheric climate connections to this local stream system. We found that, on average, the set of wet years was associated with a December-January winter mean value of MEI = 1.31 (on the EN side of the teleconnection), whereas the dry years on average yielded a value of -0.22 for MEI (toward LN conditions). Two more specific and important hydrological characteristics for Sycamore Creek are the "days in succession" (number of days in the water year < 30 days since a spate of > 11-m3/s peak discharge occurred) and "days in drying" (number of days in the water year > 200 days since a spate of > 11-m3/s peak discharge occurred). These two hydrologic parameters specific to Sycamore Creek turn out to be strongly correlated to the MEI Index (r2=0.61 for MEI vs. days in succession—a direct relation showing that higher MEI values indicative of EN relate to higher incidences of lessening time spans between spates; r2 = 0.48 for MEI vs. days in dryness—an indirect relation in which lower MEI, non-EN periods, are associated with increases in periods between spates). Grimm (1993) also analyzed a biotic control factor for the stream, a time during which biotic interactions predominate. She found that the percentage of time during which neither postflood succession nor drying were occurring was relatively constant (mean = 177 days or 48% of the time during the year) even across differing hydrologically wet and dry years. Thus, links of ENSO to local biotic controls as opposed to disturbance controls in the stream would be relatively weak, and the cascade from ENSO to other related biotic factors would be limited. The ENSO cascade here is seasonal and appears to be restricted ecologically to the more disturbance-related aspects of the stream rather than to biotic aspects.

CAP LTER Area River Constituents

A third cascade example specifically related to the strong moisture signal of ENSO at CAP LTER is the cascade of inputs of moisture, stream runoff, and resultant dissolved and mineral river constituents upvalley and downvalley of the metropolitan Phoenix area. Several CAP LTER scientists have focused considerable efforts on creating a composite nitrogen budget for the CAP LTER region based on data for 1988-1996 (Baker et al. 2001; Lauver and Baker 2000). They suggest that there is a hydrologic control, especially for unusually high flow years such as 1993 when an ENSO event occurred at the midpoint of their analysis period. Overall, however, nitrogen fluxes in streams are relatively small in the budget (e.g., riverine export is low, about 3% overall of total input to the ecosystem of CAP LTER; Baker et al. 2001). They also indicate, however, that the N concentration was about 20 times higher in the outflow than in the inflow, reflecting N gained from agricultural drainage, urban runoff, and wastewater, and that N export from the ecosystem via the Gila River was twice as high as the surface-water input. Based on this, we accessed the U.S. Geological Survey's database on stream constituents (web site: www.usgs.gov) and used data for the past 25-year period to develop direct correlations of MEI index values with selected stream constituents for the winter months when MEI correlates highest with inputs of moisture (February through May). Table 7.4 presents the results for an upstream site of the metropolitan region below Bartlett Dam on the Salt River (upvalley of urban; USGS 09502000) and for a site at Gillispie Dam on the Gila River downstream of the metropolitan region (downvalley of urban; USGS 09518000). Many significant correlations are evident in table 7.4 (assuming a standard significance level of 0.05, for example, shown in parentheses). For the Salt River, oxygen, pH, solids, dissolved solids, calcium, sodium, magnesium, chloride, and sulfate all show a significant relation to MEI. For the Gila River, pH, solids, dissolved solids, calcium, sodium, magnesium, chloride, sulfate, total nitrogen, nitrogen nitrite and nitrite total, and nitrogen nitrite and nitrite dissolved are all significantly related to variations in the MEI index. R values are shown to illustrate the directionality of the relationships, positive or negative, versus the MEI index. Generally, the greater the MEI (EN phase), the less the concentration of constituents becomes per volume of water. Two differences between the Salt River site (upvalley of urban) and the Gila River site (downvalley of urban) emerge. (1) Dissolved solids are higher per volume with more runoff at Gila, whereas they

Table 7.4 Correlations (with significance levels in parentheses) of MEI versus February-May monthly concentrations over the 25-year period 1971-1995

Parameter

Gila Rivera

Salt Riverb

Water temperature

0.08

(0.38)

0.02 (0.889)

Turbidity

0.16

(0.19)

-0.05 (0.756)

Oxygen

0.10

(0.36)

0.38 (0.016)

pH

0.23

(0.06)

0.23 (0.002)

Arsenic

-0.03

(0.81)

0.20 (0.254)

Solids

-0.35

(0.004)

-0.47 (0.001)

Dissolved solids

0.23

(0.030)

-0.54 (0.000)

Dissolved calcium

-0.50

(0.000)

-0.44 (0.001)

Dissolved sodium

-0.54

(0.000)

-0.38 (0.003)

Dissolved magnesium

-0.51

(0.000)

-0.36 (0.007)

Dissolved chloride

-0.54

(0.000)

-0.44 (0.002)

Dissolved sulfate

-0.52

(0.000)

-0.37 (0.010)

Total nitrogen

-0.18

(0.250)

Little Data

Nitrogen nitrite + nitrate total

-0.40

(0.000)

-0.07 (0.970)

Nitrogen nitrite + nitrate dissolved

-0.49

(0.000)

0.10 (0.950)

" Gila River site is USGS station 09518000 (diversions at Gillespie Dam at 33"13'45" lat., 112" 46'00" long). b Salt River site is USGS station 09502000 (below Stewart Mountain Dam at 33" 33'10" lat., 111" 34'33" long).

" Gila River site is USGS station 09518000 (diversions at Gillespie Dam at 33"13'45" lat., 112" 46'00" long). b Salt River site is USGS station 09502000 (below Stewart Mountain Dam at 33" 33'10" lat., 111" 34'33" long).

are lower at the Salt River site. (2) Nitrogen-related parameters show no correlation at the Salt River site where there is little input, whereas there is a significant MEI climate signal at the Gila River site, where flushes of nitrogen elements occur more readily. With flood releases and significant variations in river discharges that reach downvalley of the urbanized region, the input of nitrogen-related constituents emanating from the urban/agricultural lands and released sediments varies significantly in relation to MEI at a seasonal timescale. Immediately downstream of Bartlett dam, only elemental constituents and not N-related constituents appear to be significantly related to the MEI variability. Hence, it appears that the human-modified urban/agricultural ecosystem in CAP LTER tends to create a positive feedback, or amplification, to the climate signal-stream constituent relationship.

The Future

Spigg and Hinkley (2000) have suggested that global warming may increase the frequency of EN events in the future. It has been hypothesized that a major impact of continued global warming might be an increased frequency of EN events in the Southwest desert area. Should this occur, increases in moisture inputs may result (presumably doubling of moisture in some areas). This could have many positive and negative benefits for the southwestern United States and CAP LTER. For example, more water may be available for the rapidly growing central Arizona area from increased snowpacks in the mountains and runoff into critical reservoirs of central Arizona. But possible negative impacts may occur in the form of reservoir releases, flood risks, ecosystem disturbance, and damage to urban areas. These sce narios are critical for the populace of this region (e.g., Carter et al. 2000). As illustrated in the three ecosystem case examples above, enhanced frequencies of EN may also cause intensification of disease risks, disturbance in streams, and severe variability in river constituents and sediment transfers. However, there remains considerable uncertainty in the global warming-enhanced EN scenarios, and researchers in the LTER network certainly share common goals in unraveling the science and ecology of possible shifts in the climate system that will cascade into important local site effects.

Acknowledgments We acknowledge David Greenland for encouraging us to make a contribution for CAP LTER; anonymous reviewers; and Nancy Grimm and Charles Redman, principal investigators of CAP LTER for encouraging us to pursue the climate aspects of CAP LTER with support from NSF and grant number DEB 9714833. We would like to acknowledge the influence of K. Wolter in making us aware of the MEI index and for sharing its database. We also thank Barbara Trapido-Lurie for some of the cartographic work.

References

Adams, D.K., and A.C. Comrie, 1997. The North American monsoon. Bulletin of the American Meteorological Society, Vol. 78, 2197-2213.

Baker, L. A., D. Hope, Y. Xu, J. Edmonds, and L. Lauver, 2001. Nitrogen balance for the Central Arizona-Phoenix (CAP) ecosystem. Ecosystems, Vol. 4, 582-602.

Balling, R. C., Jr., and R.S. Cerveny, 1987. Long-term associations between wind speeds and the urban heat island of Phoenix, Arizona. Journal of Climate and Applied Meteorology, Vol. 26, 712-716.

Brazel, A. J. 1987. Dust and climate in the American Southwest. In Paleoclimatology and Paleometeorology: Modern and Past Patterns of Global Atmospheric Transport. Edited by M. Leinen and M. Sarinthein. NATO ASI Series, Kluwer Academic Pubs., 65-96.

Brazel, A. J., 1991. Blowing dust and highways: The Case of Arizona, U.S.A. In Highway Meteorology. Edited by A. Parry and L. Symons. Taylor and Francis Books Ltd., London, 131-161.

Brazel, A. J., and W. G. Nickling, 1987. Dust storms and their relation to moisture in the Sonoran-Mojave desert region of the Southwestern United States. Journal of Environmental Management, Vol. 24, 279-291.

Brazel, A. J., W. G. Nickling, and J. Lee, 1986. Effect of antecedent moisture conditions on dust storm generation in Arizona. In Aeolian Geomorphology. Boston: Allen and Unwin. 261-271.

Brazel, A. J., N. Selover, R. Vose, and G. Heisler, 2000. The tale of two climates—Baltimore and Phoenix urban LTER sites. Climate Research, Vol. 15, 123-135.

Brazel, A. J., H. J. Verville, and R. Lougeay, 1993. Spatial-temporal controls on cooling degree hours: An energy demand parameter. Theoretical and Applied Climatology, Vol. 47, 81-92.

Carter, R. H., P. Tscharert, and B. J. Morehouse, 2000. Assessing the sensitivity of the South-west's urban water sector to climatic variability: Case studies in Arizona. Climate Assessment for the Southwest (CLIMAS), Report Series CL1-00, University of Arizona.

Cayan, D. R., and K. T. Redmond, 1994. ENSO influences o

Was this article helpful?

0 0

Post a comment