Douglas Schaefer

Introduction

Variations in temperature and precipitation are both components of climate variability. Based on coral growth rates measured near Puerto Rico, the Caribbean was 2-3°C cooler during the "Little Ice Age" during the seventeenth century (Winter et al. 2000). At the millennial scale, temperature variations in tropical regions have been inferred to have substantial biological effects (such as speciation and extinction), but not at the multidecadal timescales considered here. My focus is on precipitation variability in particular, because climate models examining effects of increased greenhouse gases suggest greater changes in precipitation than in temperature patterns in tropical regions.

Some correspondence between both the El Niño-Southern Oscillation (ENSO) and the Northern Atlantic Oscillation (NAO) and average temperatures and total annual precipitation have been reported for the LTER site at Luquillo (Greenland 1999; Greenland and Kittel 2002), but those studies did not refer to extreme events. Based on climate records for Puerto Rico since 1914, Malmgren et al. (1997) found small increases in air temperature during El Niño years and somewhat greater total rainfall during the positive phase of the NAO. Similar to ENSO, the NAO index is characterized by differences in sea-level atmospheric pressure, in this case based on measurements in Iceland and Portugal (Walker and Bliss 1932). Its effects on climate have largely been described in terms of temperature and precipitation anomalies in countries bordering the North Atlantic (e.g., Hurrell 1995).

Puerto Rico is in the North Atlantic hurricane zone, and hurricanes clearly play a major role in precipitation variability. The association between extreme rainfall events and hurricanes is discussed in detail in this chapter. I examine the degree to

20 Kilometers

Figure 8.1 The island of Puerto Rico (centered approximately 18°15' N and 66°30' W), showing the locations of the sites mentioned in the text.

which extreme rainfall events are associated with hurricanes and other tropical storms. I discuss whether the occurrence of these extreme events has changed through time in Puerto Rico or can be linked to the recurrent patterns of the ENSO or the NAO. I examine the 25-year daily precipitation record for the Luquillo LTER site, the 90-year monthly record from the nearest site to Luquillo with such a long record, Fajardo, and those of the two other Puerto Rico stations with the longest daily precipitation records, Manati and Mayaguez (figure 8.1).

To explore the relationship between extreme rainfall events and sediment export, I used daily sediment export data from the Mameyes watershed in the Luquillo Mountains, and two other watersheds with the longest available records, Loiza and Rosario (figure 8.1). The Loiza watershed (23,260 ha) is in the north-central area and provides much of the inflow to the Carraizo Reservoir, the major water supply for San Juan, Puerto Rico's largest city. The rapid sedimentation of this reservoir (as well as others in Puerto Rico) is being intensively studied (Gellis 1993; Larsen et al. 1993; Morris and Fan 1998; Larsen et al. 1999). The Loiza watershed is currently experiencing substantial land clearing at the lower elevations as a result of urban expansion (Larsen et al. 1999). The Rosario watershed (4,740 ha) is in far western Puerto Rico. Sediment production has been modeled there using a combination of ground and remotely sensed data (Cruise and Miller 1994). As with much of the island outside the Luquillo Mountains, both the Loiza and Rosario watersheds were largely deforested for agriculture and pasture prior to 1940 and have since experienced secondary forest regrowth (Birdsey and Weaver 1987). The Mameyes watershed (1,782 ha) is in an area of largely undisturbed forest in the Luquillo Mountains in northeastern Puerto Rico. The analyses of daily sediment export data from Puerto Rico watersheds presented here indicates that 75% of the export occurs during the 1% of the days that have the greatest rainfall. Of course, this is an intuitive conclusion, but the extreme nonlinearity of the response deserves consideration. This type of nonlinearity in sediment delivery has been reported for the continental United States as well. Meade and Parker (1984) found that 50% of sediment export takes place in 1% of days, and 90% in 10% of days.

Variability of Extreme Rainfall Events Hurricane Variability through Time

Since the first anecdotal record in the year 1515, 51 hurricanes have passed over the island for an average of about 1 per 9.5 years. The reliability of this record may be questioned prior to 1851, the first year of Atlantic hurricane tracks published by the National Oceanic and Atmospheric Administration, so that recognition of long-term patterns may be uncertain. At the same time, there is evidence for substantial variability over the last five centuries. From 1616 to 1737 (122 years), no hurricanes were reported to pass over the island. At the other extreme, there were 9 hurricanes from 1804 to 1819 (16 years). More recently, 3 hurricanes passed over Puerto Rico from 1989 through 1998 (10 years). The approximately 16 hurricanes that have passed over Puerto Rico since 1851 constitute too small a sample set to subject to an analysis of temporal variability.

The total number of North Atlantic hurricanes does show multidecadal trends during the last 150 years, but connections with Earth's climate system remain controversial. Just as controversial is the possibility that future global climate change will affect the number or intensity of hurricanes, or the length of the annual hurricane season. In any case, the island of Puerto Rico (figure 8.1) is only 175 by 55 km and thus presents a small "target" for the hurricanes (99% of them go elsewhere), so hurricane visits to the island must always have a large random component regardless of climate variability.

Other Extreme Rainfall Events

The Luquillo LTER has kept daily rainfall records since 1975. Data on all the days with 100 mm or more rainfall at Luquillo are presented in table 8.1. Even though hurricanes may be seen as the most extreme events because of the damaging winds, they are not the only source of extreme rain events in Puerto Rico. Since 1975, 65 rain events of 100 mm or more have occurred in Luquillo. Of those, 12 (or about 18%) were from hurricanes (including instances where the "hurricane eye" did not pass over the island directly), tropical storms, or tropical depressions (the latter being essentially precursor stages to the formation of hurricanes). The other 53 (about 82%) of the events occurred in the absence of cyclonic storms, either from localized convective cells or stalled low-pressure systems. Therefore, in Puerto Rico, most of the extreme rainfall events and sediment export occur in the absence of hurricane development. The annual average precipitation at this site is 3400 mm, so these extreme events represent about 13% of the total rainfall. So 75% of the sediment export was caused by 13% of the total rainfall, on less than 1% of the days. These extreme events have occurred with no unidirectional trend through time, so that during this period, no connection with unidirectional climate change

Table 8.1 Dates of all rainfall events in the Luquillo Mountains of Puerto Rico (1975-1999) with more than 100 mm rainfall

Date H/TS/TD >100 mm Name Others > 100 mm

16-Sep-75a 243 TS. Eloise

20-Nov-75 112 10-Dec-75 132

23-Apr-77 147

18-May-77 121

24-Nov-77 247

07-Mar-78 106

10-Apr-78 228 27-May-78 130

26-Oct-78 160 TS. Kendra

19-Jan-79 111 14-Feb-79 292 30-May-79 103 30-Aug-79 309 H. David

04-Sep-79 197 TS. Fredric

14-May-80 155 06-Dec-80 139

21-Apr-81 118 13-Dec-81 222 09-May-82 117

11-May-82 106

27-May-82 149

12-Sep-82 128 TD. Debby 21-Apr-83 127 06-Jul-83 150 02-Dec-83 133

15-May-85 161

18-May-85 104

20-Jul-85 104

13-Sep-85 118 06-0ct-85 243 TD. Isabel

13-Nov-85 102

16-Nov-85 120

02-May-86 117

14-May-86 130 14-Aug-86 117

08-0ct-86 104 12-Apr-87 389

19-Jun-87 101 27-Nov-87 318 08-Dec-87 501 11-Aug-88 109

25-Aug-88 153 TD. Chris

17-Feb-89 124

03-Jun-89 120

18-Sep-89 170 H. Hugo

14-Aug-90 122

17-0ct-90 107

08-Nov-91 114

Table 8.1 Continued

Date H/TS/TD >1GG mm Name Others > 1GG mm

Table 8.1 Continued

Date H/TS/TD >1GG mm Name Others > 1GG mm

30-Dec-92

11G

16-May-95

1G4

16-Sep-95

173

H. Marilyn

04-Apr-96

12G

07-Sep-96

540

H. Bertha

23-Nov-96

258

22-Aug-97

1G4

26-Nov-97

122

07-Mar-98

42G

16-Apr-98

16G

22-Aug-98

211

19-Sep-98

377

H. Georges

02-Dec-98

119

17-Jun-99

13G

21-Oct-99

101

TS. Jose

02-Dec-99

254

Count

12

53

Average mm

233

159

Maximum mm

540

5G1

a Hurricanes (H), tropical storms (TS), and tropical depressions (TD) and their dates and rainfall amounts are shown in bold.

a Hurricanes (H), tropical storms (TS), and tropical depressions (TD) and their dates and rainfall amounts are shown in bold.

can be detected (table 8.2). These events can occur during any month but are not common during the January-March dry season (table 8.3).

In terms of monthly total precipitation, a weak ENSO signal can be discerned at stations throughout Puerto Rico. The longest monthly precipitation time series near the Luquillo Mountains comes from Fajardo on the northeastern coast, starting in 1909, and it is used here as an example. Specifically, El Niño years have more rainfall and La Niña years have less rainfall than other (index) years, but during the month of May only (figure 8.2). Although it has long been recognized that precipitation totals for the month of May are more variable than any other month (F. Scatena, pers. comm., May 3, 1997), the connection between this variability and the ENSO cycle has not previously been recognized.

In light of this pattern, the 12 extreme rainfall events in May that are shown in table 8.3 for Luquillo merit further attention. Only two of them (in 1978 and 1995) occurred during El Niño years. Thus, El Niño years had 0.22 extreme rainfall events during May per year, whereas the other (index and La Niña) years had 0.30 May events per year. From the limited 25 years of daily data examined here, I cannot conclude that the ENSO cycle alters the occurrence of extreme rainfall events in Luquillo. Longer-term (100 years) daily rainfall data from other Puerto Rico stations are considered subsequently.

The cumulative frequency distributions of daily rainfall for the wettest year of record (1987), the driest (1994), and all other years together are compared in figure

Table 8.2 Annual occurrence of extreme rainfall events at the Luquillo LTER site in Puerto Rico

Year Number of rain events >100 mm

1975

3

1976

0

1977

3

1978

4

1979

5

1980

2

1981

2

1982

4

1983

3

1984

0

1985

7

1986

4

1987

4

1988

2

1989

3

1990

2

1991

1

1992

1

1993

0

1994

0

1995

2

1996

3

1997

2

1998

5

1999

3

Table 8.3 Seasonality of extreme rainfall events at the Luquillo LTER site in Puerto Rico

Month Number of rain events greater than 100 mm

Table 8.3 Seasonality of extreme rainfall events at the Luquillo LTER site in Puerto Rico

Month Number of rain events greater than 100 mm

Jan

1

Feb

2

Mar

1

Apr

7

May

12

Jun

3

Jul

1

Aug

7

Sep

8

Oct

5

Nov

7

Dec

7

Figure 8.2 Average monthly precipitation at Fajardo, Puerto Rico (northeastern coast), 1909-1996. Data are segregated into El Niño (solid black bars), La Niña (dark shaded bars), and Index years (light shaded bars) as described in the text. Errors are one standard deviation above and below the means. During May, El Niño years have significantly greater precipitation than Index years, and La Niña years have less. During other months, the differences were not significant.

Figure 8.2 Average monthly precipitation at Fajardo, Puerto Rico (northeastern coast), 1909-1996. Data are segregated into El Niño (solid black bars), La Niña (dark shaded bars), and Index years (light shaded bars) as described in the text. Errors are one standard deviation above and below the means. During May, El Niño years have significantly greater precipitation than Index years, and La Niña years have less. During other months, the differences were not significant.

8.3. In the "other years" category, 1% of days have 100 mm or more of rain, and this is the daily rainfall threshold associated with 75% of the sediment export in the Luquillo Mountains. The year 1987 had 2% of such days, and 1994 had none. During that drought year (1994), the Mameyes watershed had only 28% of its 6-year average sediment export.

The longest daily records available for the island are Manati (north-central coast) and Mayaguez (far western coast) both beginning in 1900. The latter data were obtained in digital form from the National Climate Data Center. Whereas 1% of the days at El Verde have 100 mm rain or more, that value is reached at 54 mm at both Manati and Mayaguez, based on their entire records. To determine whether such extreme days are associated with the ENSO cycle, data were segregated into El Niño, La Niña, and Index (all other) years. For this analysis El Niño years were 1919, 1926, 1940-1942, 1952, 1958, 1964, 1966, 1973, 1978, 1983, 1987, 19921994, and 1998. La Niña years were 1918, 1939, 1950, 1951, 1956, 1971, 1974, 1976, 1989, and 1999. The ENSO data and categorical segregation followed Trenberth (1984) with more recent data taken from the University Center for Atmospheric Research web site (www.ucar.edu). For these three stations, there is no clear association between the state of ENSO and the proportion of days with greater than the threshold amounts of rain (table 8.4).

Figure 8.3 Cumulative frequency distribution of daily rainfall at the Luquillo LTER site, 1975-1999. The highest curve represents 1994, a drought year, when no days had more than 100 mm of rainfall. The lowest curve represents 1987, a flood year, when about 2% of the days had more than 100 mm of rainfall. The intermediate curve includes data from all other years, for which about 1% of days exceeded 100 mm of rainfall. Note that most days (having less than 10 mm of rainfall) are outside the range of this figure.

Figure 8.3 Cumulative frequency distribution of daily rainfall at the Luquillo LTER site, 1975-1999. The highest curve represents 1994, a drought year, when no days had more than 100 mm of rainfall. The lowest curve represents 1987, a flood year, when about 2% of the days had more than 100 mm of rainfall. The intermediate curve includes data from all other years, for which about 1% of days exceeded 100 mm of rainfall. Note that most days (having less than 10 mm of rainfall) are outside the range of this figure.

To determine whether such extreme days are associated with the NAO cycle, NAO negative years were 1918, 1919, 1966-1972, 1979, 1980, and 1981. Positive NAO years were 1907-1910, 1914, 1922-1927, 1934-1939, 1946, 1954, 1955, and 1990-1995, and index years were all others. This segregation followed Hurrell (1995) with more recent NAO data obtained from the UCAR web site listed previously. In this case both Mayaguez and El Verde had less extreme rainfall days when NAO was negative than in index or negative years (table 8.5). But we must recall that in the short El Verde record, both drought years (1994 and 1995) were NAO positive years. Whether this relationship is casual (as opposed to causal) cannot be determined from a record of this length. Droughts in Puerto Rico during future NAO-positive periods would constitute very important observations.

Table 8.4 The El Nino-Southern Oscillation (ENSO) has not altered the frequencies (percentage of days) of extreme rainfall events in Puerto Rico.

ENSO State Manati Mayaguez El Verde

54a 54 loo

Index years (all other years) 0.89 0.85 1.10

Periods of record are in parentheses and rainfall thresholds (mm/day) used for each site are in bold.

Table 8.5 The Northern Atlantic Oscillation (NAO) may have altered the frequencies (percentage of days) of extreme rainfall events in Puerto Rico

NAO State

Manati (1900-1998) 54

Mayaguez (1975-1999) 54

El Verde (1975-1999) 100

Negative

0.94

1.52

1.37

Index years (all other years)

0.98

1.13

1.28

Positive

0.85

0.83

0.37

a Periods of record are in parentheses and rainfall thresholds (mm/day) are in bold.

a Periods of record are in parentheses and rainfall thresholds (mm/day) are in bold.

Finally, the two longest records were compared during consecutive 20-year periods to search for unidirectional trends. Manati was indifferent, whereas extreme rainfall events in Mayaguez may have been increasing since 1920 (table 8.6). Whether a developing "heat island" in Mayaguez could be related to more extreme rain events (as suggested for continental U.S. cities by J. Luvall, pers. comm., January 17, 1996) is at present simply an interesting hypothesis.

Effects of Extreme Rainfall Events

Extreme rainfall events, regardless of whether they are from hurricanes, have pervasive ecological effects on this tropical forest. The historical data used for this analysis do not lead to the conclusion that global climate variability will necessarily lead to increased numbers of such extreme rainfall events. Yet it must be stressed that because of the extreme nonlinearity of sediment export versus precipitation amounts, even a small increase in the number of extreme rainfall events would have a large impact on sediment flux, as well as on its associated ecological responses, as mentioned elsewhere. Longer term daily precipitation data from other tropical sites could provide additional insight into this issue.

Table 8.6 Multidecadal trends in the frequencies (percentage of days) of extreme rainfall events in Puerto Rico

Manati

Mayaguez

(1900-1998)

(1900-1998)

Interval

54

54

1900-1919

1.14

1.15

1920-1939

0.72

0.90

1940-1959

0.97

0.91

1960-1979

1.01

1.20

1980-1998

0.87

1.35

a Periods of record are in parentheses and rainfall thresholds (mm/day) are in bold.

a Periods of record are in parentheses and rainfall thresholds (mm/day) are in bold.

Effects on Stream Water Chemistry

Since Luquillo became an LTER site in 1988, two of the three Puerto Rico hurricanes have passed close enough to the forest to cause substantial wind damage (Hugo in 1989 and Georges in 1998). Stream water chemical export is one of the ecosystem responses to these events that we have examined. In brief, stream water export of potassium and nitrate ions increased markedly after Hurricane Hugo and remained elevated for 18 to 24 months until the canopy leaf cover returned. Other ions were much less affected (Schaefer et al. 2000). Hurricane Georges caused much less forest damage, as its eye passed south of the Luquillo Mountains, and our north-facing research areas were sheltered. Consequently, the effects on stream water chemistry were much smaller than after Hugo. One conclusion from these observations is that biogeochemical cycling in this forest is resilient to the moderate hurricane damage caused by Hugo (near-total canopy defoliation and 7-11% stem mortality; Walker 1991). If there is a higher damage threshold for biogeochemical cycling in this forest, it has apparently not been crossed during the last century.

Because of the positive concentration versus discharge relationships for potassium and nitrate ions (e.g., Schaefer et al. 2000), most of the stream export for these ions occurs during high flows associated with extreme rainfall events as well. However, those relationships appear to saturate at the highest stream flows, unlike those for sediment export, so watershed export of those dissolved ions will be less sensitive to the number of extreme rainfall events than is sediment export.

Effects of Sediment Export

Watershed sediment export is also studied at the Luquillo LTER and by the U.S. Geological Survey (USGS) at watersheds elsewhere in Puerto Rico. Stream sediment export is a particularly tangible example of several coupled ecosystem responses to extreme precipitation events. Where reservoirs are located downstream, sediment fluxes reduce reservoir capacity over time. Morris and Fan (1998) found that average sediment yields from the watersheds supplying the 14 major reservoirs in Puerto Rico ranged from 7 to 27 t ha-1 yr-1. These values represent an annual loss of reservoir capacity from 0.3 to 1.3%. According to Soler-Lopez (2001), the 14 major water-supply reservoirs in Puerto Rico have lost an average of 35% of their storage capacity since construction, which is twice the rate projected at their time of construction. With regard to the near-shore marine environment, Rogers (1990) summarized negative effects of sediment discharge on coral reefs and sea grass beds. Sediment export can carry adsorbed toxins such as heavy metals, pesticides, and other organic compounds (Meade and Parker 1984). With gully formation in particular, soil erosion is linked to the loss of arable cropland (Lal 1994). Riverine biota may be negatively impacted by sediment exports via reduction in food supplies, clogging of feeding structures, mechanical scouring, suffocation, and downstream relocation during the high flows (Newcombe and MacDonald 1991). The sediment fluxes during extreme rainfall events in the Luquillo Mountains of Puerto Rico are also associated with landslides (Larsen and Simon 1993). Their data indicate that 24-hour rainfall totals of 200 mm or greater trigger landslides.

Table 8.7 Annual sediment fluxes for three Puerto Rico watersheds in the years of record and the number of extreme events (defined as those with 75% of the total recorded export) occurring in each year

Loiza extreme

Rosario extremea

Mameyes extremeb

Water Year

(Oct-Sep)

t ha-1 yr-1

1 days

t ha-1 yr-1

days

t ha-1 yr-

days

1983-1984

10.5

3

1984-1985

20.1

7

1985-1986

33.3

7

1986-1987

4.5

3

7.0

3

1987-1988

32.1

9

21.1

9

1988-1989

8.2

3

10.9

6

1989-1990c

1990-1991

4.9

1

4.0

0

1991-1992

19.1

6

10.3

2

1992-1993

7.2

1

13.8

7

2.1

3

1993-1994

2.7

1

2.4

1

0.6

1

1994-1995

3.8

2

5.1

1

0.9

1

1995-1996

23.5

2

3.4

1

4.7

5

1996-1997

3.3

0

2.1

0

2.0

4

1997-1998

26.0

4

83.1

5

8.7

15

Average

14.2

3.6

14.8

3.4

3.2

4.9

CV (%)

128

75

157

78

100

108

"Data begin in 1986. b Data begin in 1992. c Data unavailable.

"Data begin in 1986. b Data begin in 1992. c Data unavailable.

Quantifying Sediment Export in Puerto Rico

Daily sediment export is currently being measured in more than 20 rivers in Puerto Rico (Diaz et al. 1984-1998). For this study, daily streamflow and sediment fluxes in the Loiza, Rosario, and Mameyes watersheds were taken from the USGS annual water resources data reports for Puerto Rico (Diaz et al. 1984-1998). Daily data were summed for annual totals, and those days totaling 75% of the sediment export (extreme days) were identified (table 8.7). In all three watersheds, 75% of the sediment export occurs in approximately 1% of the days. Sediment exports from these three watersheds average 3.2 t ha-1 yr-1 (Mameyes 1993-1998), 14.2 t ha-1 yr-1 (Loiza 1984-1998), and 14.8 t ha-1 yr-1 (Rosario 1987-1998).

By comparison, sediment yields in the continental United States range from 0.15 to 150 t ha-1 yr-1 (Vanoni 1975), and tropical rates range from 1 to more than 100 t ha-1 yr-1, with the higher rates being associated with severe land degradation (Lal 1990). In a compilation of data from 280 global rivers, Milliman and Syvitski (1992) provide annual sediment yields from the 16 largest tropical rivers (draining land areas of 430,000 km2 and greater) ranging from 0.1 to 14 t ha-1 yr-1. Their data for 73 smaller tropical rivers range from 0.1 to 360 t ha-1 yr-1. Sediment yields in excess of 300 t ha-1 yr-1 are observed in Taiwan (Li 1976) that exhibit an unfortunate combination of extensive land clearing, heavy rainfall (in part from tropical typhoons, the Pacific Ocean equivalent of hurricanes), and steep lands resulting from rapid rates of geological uplift.

Based on 1984-1990 data for the Loiza watershed, Morris and Fan (1998) reported that 65% of the sediment was delivered in 10 days (0.3% of the total days) and 17% of the total occurred in the single largest event (290,000 tons on 13 May 1986). That sediment discharge has since been exceeded by the 2-day total that occurred during Hurricane Georges (352,200 tons during 21-22 September 1998) and 396,000 tons during Hurricane Bertha (10 September 1996).

In a related study, the revised universal soil loss equation was applied to the Guadiana watershed in central Puerto Rico by Del Mar Lopez et al. (1998). They reported that, within that watershed, areas of open forest eroded at 26 t ha-1 yr-1 and closed forest at 7 t ha-1 yr-1. Based on their sediment delivery ratio (the fraction of eroded soil material that reaches the river channel; Trimble 1975) of 0.17, these cover classes yield 4.42 and 0.12 t ha-1 yr-1sediment to the river, respectively.

My selection of the 75% level of total stream sediment export was arbitrary, but convenient in that, based on 25 years of daily precipitation records at the Luquillo site, this occurred in precipitation events of 100 mm or more.

Effects of Numbers of Extreme Events on Sediment Exports

For all three watersheds, the extreme events, defined as producing 75% of the total sediment exports, were identified. In the Loiza watershed, there were 50 such events in 14 years (average 3.6 yr-1), of which 12 were associated with hurricanes, tropical storms, or tropical depressions (24% of total). In the Rosario watershed, there were 37 such events in 11 years (average 3.4 yr-1), of which 4 were associated with hurricanes, tropical storms, or tropical depressions (11% of total). In the Mameyes watershed, there were 29 such events in 6 years (average 4.8 yr-1), of which 3 were associated with hurricanes, tropical storms, or tropical depressions (10% of total). Analyses of how increasing the number of these extreme events could increase sediment exports were conducted as follows: Annual exports were calculated without these events, and the annual sediment export attributed solely to those extreme events was regressed against the number of events in each year (figures 8.4-8.6).

Sediment export from the Rosario watershed during Hurricane Georges appeared to be an outlier in this analysis, as the r2 of that regression was increased from 0.14 to 0.74 by its exclusion. For that reason the regressions for all three watersheds were performed both with and without Hurricane Georges, and both versions are presented in figures 8.4-8.6 (showing Loiza, Rosario, and Mameyes, respectively). This modification slightly improved the Loiza r2 (from 0.65 to 0.74), and reduced the Mameyes r2 (from 0.95 to 0.70). Based on these models, the effect of increasing numbers of extreme events was explored and those results are presented in table 8.8. For these analyses, all extreme events were treated as equals, even though they vary greatly in terms of sediment production. This simplification may be justified in two ways. First, the relationships between rainfall amounts or

0 2 4 6 3 10 0246S10

Extreme events per year Extreme events per year

Figure 8.4 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Loiza watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

0 2 4 6 3 10 0246S10

Extreme events per year Extreme events per year

Figure 8.4 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Loiza watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

Extreme events per year Extreme events per year

Figure 8.5 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Rosario watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

Extreme events per year Extreme events per year

Figure 8.5 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Rosario watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

intensities and sediment production are complex. Second, it was not deemed fruitful to specify the sizes of extreme events that would be added for these analyses.

Although climate variations (whether natural or anthropogenic) could alter the number of such extreme events, the analyses performed on long-term rainfall data from Puerto Rico provide scant evidence that this has occurred to date. Rather, this exercise demonstrated the sensitivity of total sediment export to the number extreme events that may occur. In Puerto Rico, one additional event per year could add 13 to 24% to the total sediment export, and doubling the number of extreme events occurring could increase sediment export by 61 to 95%.

Extreme everts per year Extreme events per year

Figure 8.6 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Mameyes watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

Extreme everts per year Extreme events per year

Figure 8.6 Regressions of the annual numbers of extreme events on the sediment exports during those events in the Mameyes watershed of Puerto Rico. Filled circles represent annual data, solid lines are regressions, and dashed curves are 95% confidence intervals. Regressions are performed with all data and excluding the 1998 Hurricane Georges (see text).

Conclusions

What preexisting conditions affect sediment export during extreme rainfall events? The condition of saturated soil is important, but not required. Surface flow of water is the minimum necessary condition for sediment mobilization. This can occur whenever the hydraulic conductivity of the surface soil is exceeded by the rainfall rate. When unsaturated, surface soils have higher hydraulic conductivities than when they are saturated (by previous rainfall). In effect, sediment mobilization begins at lower rainfall rates when soils are saturated (Lal 1990). The rainfall thresholds selected in this study exceed this requirement because of their observed effects on sediment export.

The lower limit was arbitrarily selected as being one of the larger rainfall events associated with the export of 75% of the sediment totals. In the Luquillo Mountains, this rainfall threshold is approximately 100 mm, and it is about 50 mm in the lower elevation rain records (Mayaguez and Manati) considered here. Although there is no fixed upper limit to the amount of rain that can fall within a 24-hour period, there are no records that it has exceeded 600 mm in Puerto Rico. The largest recorded flood (5-10 October 1970; Haire 1972) in Puerto Rico exhibited a maximum 1-day total of about 460 mm (5-day total of 970 mm), and the 2-day total for Hurricane Bertha (1996) in El Verde was 540 mm.

With respect to soil erosion and sediment deposition, conditions generally do not return to a previous state. The biota in the sediment source areas, the watershed itself, may persist (or not) in the remaining soil resource. River channels and coastal systems may be cleared of their sediments by subsequent water flow, but a persistent biological signal (perhaps best expressed as long-term variation in coral growth rates) may remain. In the very long term, geologic uplift and pedogenesis may keep pace with soil erosion—or it may not.

Table 8.8 Projection of the effects of adding one extreme sediment export event (defined as one of the class that produces 75% of the total export recorded) and of doubling the annual number of such events on the average annual sediment export (t ha-1 yr-1) in three Puerto Rico watersheds

Watersheds

Case 1. All dataa

Average export w/one 17.4 17.7 3.8

extreme event added (t ha-1 yr-1) (+23%) (+20%) (+19%)

Average export with 25.8 24.7 6.2

extremes doubled (t ha-1 yr-1) (+82%) (+67%) (+95%)

Case 2. Hurricane Georges excludedb

Average export w/one 16.3 9.4 3.0

extreme event added (t ha-1 yr-1) (+24%) (+20%) (+13%)

Average export with 24.4 13.2 4.2

extremes doubled (t ha-1 yr-1) (+85%) (+67%) (+61%)

a Analyses performed on all the total data sets.

b Analyses excluded Hurricane Georges (21-22 September 1998) sediment fluxes.

There is weak evidence, at best, that climatic cycles (ENSO and NAO) influence the occurrence of extreme rainfall events in Puerto Rico. One of the century-long daily rain records examined here (Mayaguez) may indicate an increase over the last several decades, and thus may merit further study. Sediment export records for other LTER sites (and elsewhere) also merit examination with respect to climate cycles and decadal (or longer) trends. The hypothetical analysis performed here on possible increases in extreme rainfall events is intended to highlight the high sensitivity of the sediment export process to these events. Doubling the number of events would not cause annual rainfall totals to fall outside the historic range of variation, but it could increase sediment export to unusually high levels.

Finally, even though hurricanes and their precursor stages (tropical depressions and tropical storms) constitute a portion of the extreme rainfall events, they are by no means the majority. If either natural climate variability or anthropogenic effects increase the number of extreme rainfall events, this could occur with or without changes in hurricane frequency or intensity.

Acknowledgments This research was supported by the U.S. National Science Foundation Long Term Ecological Research Program (BSR-8718396, BSR-8811902 and DEB-9411973) and a NASA Institutional Research Award (NAGW-4059) to the University of Puerto Rico.

References

Birdsey, R. A., and P. L. Weaver. 1987. Forest area trends in Puerto Rico. U.S. Department of Agriculture Forest Service Research Note SO-331.

Cruise, J. F., and R. L. Miller. 1994. Hydrologie modeling of land processes in Puerto Rico using remotely sensed data. Water Resources Bulletin 30:419-428.

Del Mar Lopez, T., T. M. Aide, and F. N. Scatena. 1998. The effect of land use on soil erosion in the Guadiana watershed in Puerto Rico. Caribbean Journal of Science 34:298-307.

Diaz, P. L., Z. Aquino, C. Figueroa-Alamo, R. J. Vachier, and A. V. Sanchez. 1984-1998. Water Resources Data for Puerto Rico and the U.S. Virgin Islands, U.S. Geological Survey, San Juan, Puerto Rico.

Gellis, A. 1993. The effect of Hurricane Hugo on suspended-sediment loads in the Lago Loíza basin, Puerto Rico. Earth Surface Process and Landforms 18:505-517.

Greenland, D. 1999. ENSO-related phenomena at long-term ecological research sites. Physical Geography 20:491-507.

Greenland, D., and T. G. F. Kittel. 2002. Temporal variability of climate at the U.S. Long-term ecological research (LTER) sites. Climate Research 19:213-231.

Haire, W. J. 1972. Flood of October 5-10, 1970 in Puerto Rico. U.S. Geological Survey Water Resources Bulletin 12. San Juan, Puerto Rico.

Hurrell, J. W. 1995. Decadal trends in NAO, regional temperatures and precipitation. Science 269:676-679.

Lal, R. 1990. Soil erosion in the topics: Principles and management. McGraw-Hill, New York.

Lal, R. 1994. Soil erosion by wind and water: problems and prospects. In R. Lal, editor. Soil erosion research methods. Soil and Water Conservation Society, Ankeny, Iowa.

Larsen, M. C., P. D. Collar, and R. F. Stallard. 1993. Research plan for the investigation of water, energy, and biogeochemical budgets in the Luquillo Mountains, Puerto Rico. U.S. Geological Survey Open-File Report 92-150.

Larsen, M. C., and A. Simon. 1993. Rainfall intensity-duration threshold for landslides in a humid tropical environment, Puerto Rico. Geografiska Annaler 75A:13-21.

Larsen, M. C., A. J. Torres-Sánchez, and I. M. Concepción. 1999. Slopewash, surface runoff, and fine-litter transport in forest and landslide scars in humid-tropical steeplands, Luquillo Experimental Forest, Puerto Rico. Earth Surface Processes and Landforms 24:481-506.

Li , Y. H. 1976. Denudation of Taiwan island since the Pliocene epoch. Geology 4:105-107.

Malmgren, A., A. Winter, and D. Chen. 1997. El Niño-southern oscillation and North Atlantic oscillation control of climate in Puerto Rico. Journal of Climate 11:2713-2718.

Meade, R. H., and R. S. Parker. 1984. Sediment in rivers of the United States. National water summary. United States Geological Survey, Reston, Va.

Milliman, J. D., and J. P. M. Syvitski. 1992. Geomorphic/tectonic control of sediment discharge to the ocean: The importance of small mountainous rivers. Journal of Geology 100:525 -544.

Morris, G. L., and J. Fan. 1998. Reservoir sedimentation handbook: Design and management of dams, reservoirs and watersheds for sustainable use. McGraw-Hill, New York.

Newcombe, C. P., and D. D. MacDonald. 1991. Effects of suspended sediments on aquatic ecosystems. North American Journal of Fisheries Management 11:72-82.

Rogers, C. S. 1990. Response of coral reefs and reef organisms to sedimentation. Marine Ecology Progress Series 62:185-202.

Schaefer, D. A., W. H. McDowell, F. N. Scatena, and C. E. Asbury. 2000. The effects of hurricane disturbance on long-term stream water concentrations in eight tropical forest watersheds of the Luquillo Experimental Forest, Puerto Rico. Journal of Tropical Ecology 16:189-207.

Soler-Lopez, L. R. 2001. Sedimentation survey results of the principal water-supply reservoirs of Puerto Rico. In W. F. Sylva, editor. Proceedings of the sixth Caribbean Islands Water Resources Congress. Mayagüez, Puerto Rico, February 2001.

Trenberth, K. E. 1984. Signal versus noise in the southern oscillation. Monthly Weather Review 112:326-332.

Trimble, S. W. 1975. Denudation studies: Can we assume steady state? Science 188:12071208.

Vanoni, V. A. 1975. Sedimentation engineering. American Society of Civil Engineers, New York.

Walker, G. T., and E. W. Bliss. 1932. The North Atlantic oscillation. Memoirs of the Royal Meteorological Society 44:53-83.

Walker, L. R. 1991. Tree damage and recovery from Hurricane Hugo in the Luquillo Experimental Forest, Puerto Rico. Biotropica 23:379-385.

Winter, A., H. Ishioroshi, T. Watanabe, T. Oba, and J. Christy. 2000. Caribbean Sea surface temperatures: Two-to-three degrees cooler than present during the Little Ice Age. Geophysical Research Letters 27:3365-3368.

0 0

Post a comment