23
© IWA Publishing 2003
Journal of Water and Health
|
01.1
|
2003
Storm effects on regional beach water quality along the
southern California shoreline
Rachel T. Noble, Stephen B. Weisberg, Molly K. Leecaster,
Charles D. McGee, John H. Dorsey, Patricia Vainik and
Victoria Orozco-Borbón
ABSTRACT
Two regional studies conducted during dry weather demonstrated that the Southern California Bight
(SCB) shoreline has good water quality, except near areas that drain land-based runoff. Here, we
repeat those regional studies 36 h after a rainstorm to assess the influence of runoff under high flow
conditions. Two hundred and fifty-four shoreline sites between Santa Barbara, California and
Ensenada, Mexico were sampled using a stratified-random sampling design with four strata: sandy
beaches, rocky shoreline, shoreline adjacent to urban runoff outlets that flow intermittently, and
shoreline adjacent to outlets that flow year-round. Each site was sampled for total coliforms, fecal
coliforms (or E. coli), and enterococci. Sixty percent of the shoreline failed water quality standards
after the storm compared to only 6% during dry weather. Failure of water quality standards
increased to more than 90% for shoreline areas adjacent to urban runoff outlets. During dry weather,
most water quality failures occurred for only one of the three bacterial indicators and concentrations
were barely above State of California standards; following the storm, most failures were for multiple
indicators and exceeded State of California standards by a large margin. The condition of the
shoreline in Mexico and the United States was similar following rainfall, which was not the case
during dry weather.
Key words
| coliform, enterococcus, indicator bacteria, runoff, stormwater
Rachel T. Noble (corresponding author)
UNC at Chapel Hill,
Institute of Marine Sciences,
3431 Arendell St., Morehead City,
NC 28557, USA
Tel.: +1 252 7266841 x150
Fax: +1 252 7262426
E-mail: rtnoble@email.unc.edu
Stephen B. Weisberg
Southern California
Coastal Water Research Project,
7171 Fenwick Lane, Westminster,
CA 92683, USA
Molly K. Leecaster
Idaho National Environment and
Engineering Laboratory, Idaho Falls,
ID, 83415, USA
Charles D. McGee
Orange County Sanitation District,
10844 Ellis Avenue, Fountain Valley,
CA 92708, USA
John H. Dorsey
Department of Natural Sciences,
Loyola Marymount University,
One LMU Drive, Los Angeles,
California 90045, USA
Patricia Vainik
City of San Diego
Metropolitan Wastewater Department,
4918 North Harbor Drive, San Diego,
CA 92106, USA
Victoria Orozco-Borbón
Instituto de Investigaciones Oceanológicas,
Universidad Autónoma de Baja California,
Km. 103 Carretera Tijuana-Ensenada,
Ensenada, México
INTRODUCTION
Land-based runoff is increasingly being recognized as a
Harbor, FL occur at sites near where urban streams enter
source of fecal bacteria and a public health concern at
the estuary. Mallin et al. (2000) found that fecal coliform
swimming beaches. Noble et al. (2000) found that 60% of
concentrations in South Carolina were directly correlated
the Southern California Bight (SCB) shoreline areas
with the percent of impervious surface in the watershed.
receiving urban runoff fail State of California (CA) water
Human viruses are consistently found in southern
quality standards. Lipp et al. (2001a) demonstrated that
California’s urban runoff (Jiang et al. 2001, Noble and
the highest indicator bacteria concentrations in Charlotte
Fuhrman 2001) and Haile et al. (1999) demonstrated that
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
24
Rachel T. Noble et al.
|
Journal of Water and Health
Regional microbiological studies
|
01.1
|
2003
illness rates more than double when swimming at beaches
sponding to four shoreline types: sandy beach, rocky
near urban runoff outlets.
shoreline, perennial urban runoff outlets, and ephemeral
The effect of urban runoff on beach water quality is
urban runoff outlets. Although the basic sample allocation
even more severe following rain events. More than half of
scheme was stratified random, a systematic component
the beach water quality failures in Santa Monica Bay,
was added to minimize clustering of sample sites along the
California, are associated with rain events, even though it
shore. This was accomplished using an extension of the
typically rains less than 15 days per year (Schiff et al., in
National Stream Survey sampling design of Messer et al.
press). Several researchers have found significant correla-
(1986) and Overton (1987). The term ‘urban runoff outlets’
tions between beach bacterial concentration and river
is used to describe storm drains, creeks, and rivers that
discharge (Solo-Gabrielle et al. 2000; Dwight et al. 2002).
contribute freshwater/stormwater inputs to the coastal
Rainfall effects are also apparent on an interannual basis
Pacific Ocean. A total of 81 urban runoff outlets that
as both Lipp et al. (2001b) and Boehm et al. (2002) have
convey 99% of the total freshwater input to the SCB were
demonstrated higher beach bacterial concentrations
identified and differentiated as perennial or ephemeral
during El Niño years.
based upon whether water flowed year-round or season-
While these studies have demonstrated increases
ally, respectively. Sample sites within the perennial and
in bacterial concentration associated with wet weather
ephemeral water outlet strata were selected using two
runoff, they are mostly based on integrating existing public
methods. First, sites were selected at a random distance
health monitoring data, which are focused on high use
within 100 m of the mouth of the outlet (random sites).
beaches and not designed to assess the spatial extent of
Second, a site was placed on the beach at a location as
stormwater influence. The question of spatial extent is
close to the mouth of the outlet as possible (referred to as
particularly important in southern California, where the
the point zero site). At the perennial urban runoff outlets,
rainfall influence on beach water quality is perceived as
random sites were placed around 39 of the 40 outlets, and
severe enough that the health departments routinely issue
point zero sites were placed at 30 of the 40 perennial
warnings to avoid recreational water contact for at least
outlets. At the ephemeral outlets, 36 random sites and
3 days following a storm. Here, we present a survey in
29 point zero sites were sampled from the 41 possible
which 1000 km of the SCB shoreline was synoptically
systems.
sampled the day after a storm to assess the spatial
influence of rainfall on regional water quality.
Samples were collected in sterile sample bottles or
Whirl-Pak bags from ankle-deep water on an incoming
wave just prior to receding, with the sampler positioned
downcurrent from the bottle and the mouth of the bottle
facing into the current. After the sample was taken, the
bottle was tipped to decant enough sample to ensure 2 to
METHODS
5 cm of airspace in the sample bottle. The bottle was then
tightly capped, stored on ice in the dark, and returned to
Samples were collected along the shoreline of the SCB at
the laboratory in time to begin analysis within 6 h of
254 sites between Point Conception, California, and Punta
sample collection. All samples were tested for total
Banda, Mexico (Figure 1). All sites were sampled between
coliforms (TC), fecal coliforms or E. coli (FC), and
06.00 and 10.00 h on 22 February 2000, approximately
enterococci (EC). Collection and processing of samples in
36 h after a storm that deposited ca. 3–7 cm of precipita-
a short period was accomplished through cooperative
tion over the entire study region. These were the same sites
efforts of 21 organizations that conduct routine monitor-
sampled by Noble et al. (1999, 2000) during two previous
ing of southern California’s beaches. Each participating
dry weather regional water quality surveys along the
laboratory used their established analytical methods for
shoreline of the SCB (Figure 1). These sites were selected
sampling processing, which include membrane filtration
using a stratified random approach, with strata corre-
(MF), multiple tube fermentation (MTF), and the defined
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
25
Rachel T. Noble et al.
|
Figure 1
|
Journal of Water and Health
Regional microbiological studies
|
01.1
|
2003
Map of the Southern California Bight depicting the shoreline, counties, and land features of southern California and sampling sites (black dots).
substrate technology test kits, Colilert® and Enterolert
Methods (1995) or following manufacturer’s instructions.
(IDEXX Laboratories, Inc., Portland, ME). All analyses
Comparability among laboratories and among methods
were performed using techniques as outlined in Standard
was confirmed prior to the study through a series of
intercalibration studies (Noble et al. 2003). To enhance
reliability of comparisons between studies conducted
Table 1
|
State of California single sample daily bacterial indicator thresholds
during
wet
and
dry
conditions,
each
laboratory
processed samples from the same sites as they did in the
Daily limits
Indicator
(cfu or MPN per 100 ml)*
Total coliforms
10,000
Fecal coliforms
400
Enterococci
104
two previous dry weather regional surveys (Noble et al.
1999, 2000).
Total coliform:fecal coliform ratio
When total coliforms are
> 1,000, and TC:FC≤10
The assessment of shoreline condition focused on
estimating the percent of shoreline miles that exceeded a
threshold of concern. The State of CA daily single-sample
water quality standards for TC, FC, EC and the TC:FC
ratio were used as thresholds (Table 1). The percent of
shoreline exceeding the thresholds was estimated for each
strata and for the shoreline as a whole using a ratio
*cfu: colony forming units; MPN: most probable number
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
estimator (Thompson 1992).
26
Table 2
Rachel T. Noble et al.
|
|
Journal of Water and Health
Regional microbiological studies
|
01.1
|
2003
Rainfall quantity for locations in Southern California (in centimetres) and duration (in hours) for the storm sampled during the Storm Study
(2/20/00–2/21/00). Also included is antecedent rainfall information
Location
Time rain
started
(2/20)
Time rain
stopped
(2/21)
Duration
of storm
(h)
Rainfall
(cm)
San Ysidro
6:00 am
9:00 pm
40
7.19
Plaza Bonita Rd
6:00 am
4:00 am*
47
3.91
Fashion Valley
5:00 am
8:00 pm
40
5.11
San Onofre
5:00 am
9:00 pm
41
Encinitas
5:00 am
1:00 am*
Carlsbad
5:00 am
Oceanside
Days since
last rain
Duration of
most recent
storm (h)
0.41
9
0.41
23
3
0.71
23
2.79
3
0.30
21
46
2.69
3
0.41
14
12:00 am
44
4.09
3
0.51
21
5:00 am
4:00 pm
36
3.63
3
0.30
15
Santa Ana River
6:00 am
4:00 pm
34
4.09
NA
0.00
Coyote Creek
6:00 am
3:00 pm
33
3.23
NA
0.00
Point Vicente
6:00 am
8:00 pm
39
3.91
3
1.30
14
Malibu
7:00 am
5:00 pm
35
5.61
3
1.30
18
Oxnard Airport
5:00 am
11:00 am
31
7.06
3
0.43
14
Ventura
4:00 am
12:00 pm
33
7.21
3
0.84
14
Sea Cliff
4:00 am
12:00 am
45
7.47
3
1.09
13
Lechuza Patrol
6:00 am
12:00 pm
31
7.80
3
1.30
17
Point Hueneme
4:00 am
11:00 am
32
4.75
3
0.33
17
Santa Barbara
4:00 am
3:00 am*
48
6.88
3
0.94
13
UCSB
5:00 am
10:00 pm
42
7.34
> 19
0.00
31–48 h
2.69–7.8 cm
3 days
0–1.3 cm
Overall range
3
Rainfall from
most recent
storm (cm)
2.5
9–23 h
*Rain stopped on 2/22.
NA: No data available
RESULTS
of the rainfall event averaged 39 h. A smaller storm
that produced rainfall quantities between 0.1 and 1.25 cm
The rainfall event that preceded sampling deposited
preceded this storm event by 3 days.
between 2.5 and 7.0 cm throughout the study area,
More than half (58%) of the SCB shoreline exceeded
with the highest quantity measured near the Los
at least one of the indicator bacteria thresholds (Table 3).
Angeles–Ventura County border (Table 2). Duration
Beach areas immediately in front of perennially flowing
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
27
Rachel T. Noble et al.
Table 3
|
|
Journal of Water and Health
Regional microbiological studies
Fecal
coliforms
Total
coliforms
Total:fecal
ratio<10
Any
indicator
Ephemeral point zero
52
26
11
22
52
Ephemeral
38
13
3
11
38
Rocky
34
19
6
7
34
Sandy
59
42
31
18
62
Perennial
67
28
20
17
67
Perennial point zero
87
43
33
30
87
All SCB
56
36
24
16
58
|
01.1
|
2003
Percent of shoreline miles that exceeded State of California water quality standards in the Southern California Bight
Enterococci
Table 4
|
Percent of shoreline that exceeded single or multiple bacterial indicator standards
Any
indicator
Only one
indicator
Any two
indicators
Any three
indicators
All four
indicators
Ephemeral point zero
52
19
11
19
4
Ephemeral
38
13
14
8
3
Rocky
34
8
19
4
5
Sandy
62
14
10
33
5
Perennial
67
29
13
18
8
Perennial point zero
87
40
3
27
17
All SCB
58
15
12
26
5
urban runoff outlets had the highest frequency of
highest frequency of multiple indicator threshold failures
threshold failures (87%). The rocky shoreline strata had
occurred at the perennial point zero sites (Table 4).
the lowest frequency of failures (34%).
The vast majority of water quality exceedences,
EC was the indicator bacteria that exceeded state
regardless of indicator type, were significantly above the
water quality standards most often, with nearly 100%
water quality thresholds. Using method-specific estimates
of the samples that failed standards exceeding for EC
of laboratory variability developed during the intercalibra-
(Table 3). EC exceeded water quality standards at twice
tion exercise (Noble et al. 2003), we found that 77% of the
the frequency of FC. Approximately three-quarters of the
samples exceeding water quality standards for EC did so
samples failed water quality standards for more than one
by more than one standard deviation of measurement
bacterial indicator during the Storm Study (Table 4). The
error. Similarly, 42% and 53% of the TC and FC failures
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
28
Rachel T. Noble et al.
Table 5
|
|
Journal of Water and Health
Regional microbiological studies
Comparison of the percentage of total shoreline miles that failed State of
California water quality standards in Mexico and the United States following a storm event
Enterococci
Fecal
Total
coliforms
coliforms
Mexico
Table 6
|
|
01.1
|
2003
Median indicator concentrations in the United States and Mexico following a
storm (reported as MPN or cfu/100 ml)
Enterococci
Total
Fecal
coliforms
coliforms
Beach
Sandy beach
63
32
11
Mexico
330
Point zero
80
50
20
United States
130
900
80
Entire shoreline
66
36
15
Mexico
310
1,450
515
United States
228
1,400
80
66
42
31
Point zero
87
43
33
Entire shoreline
61
36
24
220
Urban Runoff Outlets
United States
Sandy beach
490
Kachel, in press). Moreover, southern California has an
arid environment with a short rainy season and long dry
periods when the rivers provide minimal runoff. Thus,
exceeded the standard by more than a standard deviation
bacteria and other contaminants accumulate on land
of measurement error.
between storms, enhancing runoff quality concerns com-
The failure of California’s water quality standards
pared to temperate areas where rainfall is more frequent.
along the Mexican shoreline were similar to that found in
The storm effect on water quality is well illustrated by
the United States (Table 5). For example, 63% and 66% of
comparison with results from the two dry weather regional
the shoreline along beaches failed the EC threshold in
surveys that sampled at the same sites (Noble et al. 1999,
Mexico and the United States, respectively. Median indi-
2000). The extent of shoreline that exceeded water quality
cator concentrations of samples that failed standards were
standards during this study was nearly 10 times higher
also similar between the United States and Mexico, except
than in the two dry weather studies (Figure 2). This
for FC. FC concentrations were noticeably lower in the
increase was observed across all shoreline types and
United States at both urban runoff outlets and beaches
among all bacteria indicator types (Figures 2 & 3). More-
(Table 6).
over, the magnitude of the exceedences was much greater
during this study. During dry weather, two-thirds of the
threshold failures were attributable to failure of a single
bacterial indicator and most of those failures were barely
above the indicator threshold. In contrast, two-thirds of
DISCUSSION
the threshold failures during wet weather were for
Non-point runoff concerns are exacerbated in southern
multiple indicators in which at least one indicator was
California because its rivers are highly modified storm-
twice the allowable standard (Figure 4). Because we used a
water conveyance systems that are independent of the
regionally applied, stratified-random sampling design, we
sewage
flows
have not only further demonstrated the importance of
unimpeded to the ocean. When storm events occur, runoff
rainfall as a component of urban runoff, but we have also
plumes can become large oceanographic features that
demonstrated that rainfall events have region-wide
extend for many kilometers (Bay et al. 1999, Hickey and
impacts on coastal water quality in southern California.
treatment
systems,
so
urban
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
runoff
29
Rachel T. Noble et al.
Figure 2
|
|
Regional microbiological studies
Journal of Water and Health
|
01.1
|
2003
The extent of water quality threshold exceedences in the Southern California Bight during the summer and winter studies (dry) compared with the present storm study (wet),
by shoreline type.
Figure 3
|
The extent of water quality threshold exceedences among indicator bacteria in the Southern California Bight during the summer and winter studies (dry) compared with the
storm study (wet).
Another difference between wet and dry weather con-
between Mexican and the United States. While the better
ditions was the comparability in water quality between
dry weather water quality in the US is probably a reflec-
Mexican and US waters. During dry weather, water quality
tion of their more extensive sewage treatment systems, the
standards were exceeded five times more often on
comparable wet weather water quality probably reflects
Mexican beaches than on US beaches (Noble et al. 2000).
the lack of urban runoff treatment in either country. Still,
In contrast, we found that during wet weather there was
there were some differences between the countries during
no difference in the percentage of impacted shoreline
wet weather. Median FC and EC levels were higher along
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
30
Rachel T. Noble et al.
Figure 4
|
|
Journal of Water and Health
Regional microbiological studies
|
01.1
|
2003
Relative frequency of single and multiple bacterial indicator threshold exceedences in the Southern California Bight during the summer and winter studies (dry) compared
with the storm study (wet).
Mexican shoreline, regardless of whether samples were
evaluate the health effects of wet-weather urban runoff are
taken at an open beach or near an urban runoff outlet
advisable to further support these management actions.
(Table 6). These results could be due to the fact that
Mexican runoff contains contributions of fresh human
fecal contamination from untreated sewage (Noble et al.
CONCLUSIONS
2000).
•
The public health risk of the high indicator bacteria
concentrations observed in this study are unclear, particularly if the source material has a large animal contribution.
beach water quality of southern California.
•
source of bacteria is human sewage rather than urban
weather.
•
conducted in Santa Monica Bay, CA and was limited to
The indicator bacteria, enterococci, exceeded State
of California water quality standards more often
runoff. The only epidemiological study that focused on the
human health concerns associated with urban runoff was
During large storm events, indicator bacteria levels
are orders of magnitude higher than during dry
Most studies relating bacterial indicator levels to illnesses
rates have been conducted at locations where the primary
Storm events have a dramatic regional effect on the
than total coliforms or fecal coliforms.
•
Urban runoff outlets, both in Mexico and the United
States, are primary sources of contaminated runoff,
assessing health effects of dry-weather runoff (Haile et al.
with 90% of sites near urban runoff outlets failing
1999). Currently, most public health agencies in southern
water quality standards during a storm event.
CA issue countywide warnings to avoid recreational water
contact following all storms of 1.25 cm or greater. Our
findings of high, spatially extensive indicator bacteria
counts suggest that warnings on large spatial scales are
appropriate, but additional epidemiological studies to
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
REFERENCES
Bay, S., Jones, B. H. & Schiff, K. C. 1999 Study of the Impact of
Stormwater Discharge on Santa Monica Bay. Executive
31
Rachel T. Noble et al.
|
Regional microbiological studies
Summary Report prepared for the Los Angeles County
Department of Public Works, Alhambra, CA. USC Sea Grant
Program, Los Angeles, CA. (USCSG-TR-02-99).
Boehm, A. B., Grant, S. B., Kim, J. H., Mowbray,S. L., McGee, C. D.,
Clark, C. D., Foley, D. M. & Wellman, D. E. 2002 Decadal and
shorter period variability and surf zone water quality at
Huntington Beach, California. Environ. Sci. Technol. 36,
3885–3892.
Dwight, R. H., Semenza, J. C., Baker, D. B. & Olson, B. H. 2002
Association of urban runoff with coastal water quality in
Orange County, California. Wat. Environ. Res. 74, 82–90.
Haile, R. W., Witte, J. S., Gold, M., Cressey, R., McGee, C. D.,
Millikan, R. C., Glasser, A., Harawa, N., Ervin, C., Harmon, P.,
Harper, J., Dermand, J., Alamillo, J., Barrett, K., Nides, M. &
Wang, G. 1999 The health effects of swimming in ocean water
contaminated by storm drain runoff. J. Epidem. 104, 355–363.
Hickey, B. & Kachel, N. (in press) The influence of river plumes in
the Southern California Bight. Cont. Shelf Res.
Jiang, S., Noble, R. & Chu, W. 2001 Human adenoviruses and
coliphages in urban runoff-impacted coastal waters of southern
California. Appl. Environ. Microbiol. 67, 179–184.
Lipp, E. K., Kurz, R., Vincent, R., Rodriguez-Palacios, C., Farrah S.
R. & Rose, J. B. 2001a The effects of seasonal variability and
weather on microbial fecal pollution and enteric pathogens in a
subtropical estuary. Estuaries 24, 266–276.
Lipp, E. K., Schmidt, M., Luther, M. E. & Rose, J. B. 2001b
Determining the effects of El Nino-Southern Oscillation events
on coastal water quality. Estuaries 24, 491–497.
Mallin, M., Williams, K., Esham, E. & Lowe, R. 2000 Effect of
human development on bacteriological water quality in coastal
watersheds. Ecol. Appl. 10, 1047–1056.
Messer, J. J., Ariss, C. W., Baker, J. R., Drouse, S. K., Eshleman, K.
N., Kaufman, P. N., Linthurst, R. A., Omernik, J. M., Overton,
W. S., Sale, M. J., Shonbrod, R. D., Stanbaugh, S. M. &
Tutshall, J. R., Jr. 1986 National Surface Water Survey:
National Stream Survey, Phase I-Pilot Survey.
Downloaded from http://iwaponline.com/jwh/article-pdf/1/1/23/393255/23.pdf
by guest
Journal of Water and Health
|
01.1
|
2003
EPA-600/4-86-026. U. S. Environmental Protection Agency.
Washington, D. C.
Noble, R.T. & Fuhrman, J. A. 2001 Enteroviruses detected in the
coastal waters of Santa Monica Bay, California: Low relation
to bacteriological indicators. Hydrobiol. 460, 175–184.
Noble, R. T., Dorsey, J. H., Leecaster, M. K., McGee, C. D., Moore,
D., Orozco-Borbon, V., Vainik, P. M. & Weisberg, S. B. 1999
Southern California Bight 1998 Regional Monitoring Program:
Volume II. Winter Shoreline Microbiology. Southern California
Coastal Water Research Project. Westminster, CA.
Noble, R. T., Dorsey, J., Leecaster, M., Reid, D., Schiff, K. C. &
Weisberg, S. B. 2000 A regional survey of the microbiological
water quality along Southern California Bight shoreline.
Environ. Monit. Assess. 64, 435–447.
Noble, R. T., Weisberg, S. B., Leecaster, M. K., McGee, C., Ritter, K.,
Vainik, P. & Walker, K. 2003 Comparison of methods for
measuring bacterial indicators of beach water quality. Environ.
Monit. Assess. 81, 301–312.
Overton, S. W. 1987. A sampling and analysis plan for streams, in
the national surface water survey conducted by EPA. Technical
Report No. 117. Department of Statistics, Oregon State
University, Corvallis, OR.
Schiff, K. & Kinney, P. 2001 Tracking sources of bacterial
contamination in stormwater discharges from Mission Bay.
Wat. Environ. Res. 73, 534–542.
Schiff, K. C., Morton, J. & Weisberg, S. B. (in press) Retrospective
evaluation of shoreline water quality along Santa Monica Bay
beaches. Mar. Environ. Res.
Solo-Gabrielle, H. M., Wolfert, M. A., Desmarais, T. R. & Palmer, C.
J. 2000 Sources of Escherichia coli in a coastal subtropical
environment. Appl. Environ. Microbiol. 66, 230–237.
Standard Methods for the Examination of Water and Wastewater
1995 Edited by A. D. Eaton, L. S. Clesceri & A. E. Greenberg.
18th Edition. American Public Health Association (APHA),
Washington, DC.
Thompson, S. K. 1992. Sampling. Wiley & Sons, New York.