Group Title: Virgin Islands Water Resources Institute annual technical report
Title: Virgin Islands Water Resources Institute annual technical report. FY 2002.
Full Citation
Permanent Link:
 Material Information
Title: Virgin Islands Water Resources Institute annual technical report. FY 2002.
Series Title: Virgin Islands Water Resources Institute annual technical report
Physical Description: Serial
Language: English
Creator: United States Virgin Islands. Water Resources Research Center
Affiliation: University of the Virgin Islands -- Caribbean Research Institute -- Water Resources Research Center
Publisher: United States Virgin Islands
Publication Date: 2003
Subject: Caribbean   ( lcsh )
Spatial Coverage: North America -- United States Virgin Islands
 Record Information
Bibliographic ID: CA01300598
Volume ID: VID00003
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.


This item has the following downloads:

FY_2002_Annual_Report ( PDF )

Full Text

Virgin Islands Water Resources Research Institute

Annual Technical Report
FY 2002


The University of the Virgin Islands, a land grant university and Historically Black College or University,
is committed to advancing the well-being of the people in the Virgin Islands community, and the broader
world-wide community through its academic, research and public service programs. The Water Resources
Research Institute (WRRI) is one of these programs. Receiving its primary support from the United States
Geological Survey, the WRRI in FY 2002 with its three-fold mission of information dissemination,
training and research the WRRI utilized guidance in setting its program priorities from an advisory group
composed of community stakeholders. This insured that WRRI sponsored projects were responsive to
community needs while fitting within overall program priorities established by the USGS and at the same
time maintaining consistency with the University's mission. Project principal investigators were
University of the Virgin Islands faculty and collaborators from other universities.

In its FY 2002 program, March 1, 2002 through February 28, 2003, the WRRI investigated climatic
effects on the water resources of the Virgin islands, environmental influences on the quality of cistern
water supplies and provided training on the restoration of wetland areas, critical to effective ecological
balances in island systems. These focus area activities were in addition to its usual role of serving as an
information resource to the community for water resources data, providing input and general guidance on
water quality and quantity and other related issues. This report summarizes progress made on these

Research Program

Seasonal to Century Scale Climate Effects on the Water
Resources of the U.S. Virgin Islands

Basic Information

: Seasonal to Century Scale Climate Effects on the Water Resources of the U.S.
Title: .
Virgin Islands

Project Number: 2002VI3B
Start Date: 4/1/2002

End Date: 4/1/2003

Funding Source: 104B

Not Applicable

Research Category: Climate and Hydrologic Processes

Focus Category: Climatological Processes, Drought, Surface Water

Descriptors: Rainfall modeling, Virgin Islands, precipitation patterns

Principal Amos Winter, Henry H. Smith


1. A working paper is being developed to be submitted to the Journal of Applied Meteorology.

Summary Report on

Seasonal to Century Scale Climate Effects
on the Water Resources of the U.S. Virgin Islands

Problem and Research Objectives

The U.S. Virgin Islands is very sensitive to precipitation fluctuations and lies in a region
that is susceptible to droughts and extreme precipitation events that can cause flooding
and land slides. Understanding the cyclic nature of these events will lead to better
preparedness for the population and emergency managers. Ultimately, research in this
area may lead to the ability to predict future precipitation events with reasonable skill.

The main purpose of this project is to understand the behavior of rainfall process of U.S.
Virgin Islands and its relations with climate general circulations. Five specific targets
will be accomplished:

Determine homogeneous climate zones using air temperature, and rainfall
Identify interactions between U.S. Virgin Islands rainfall changes with global
climate changes.
Identify significant correlations between rainfall behavior and meteorological
Identify time series models to predict rainfall process at each coop station and
compare with neural network prediction skills.
Design and train a neural network to perform monthly rainfall predictions.


The general methodology consists of seven major tasks: (1) Data collection (2)
Estimation of missing values (3) Development of time series models (4) Identification
of changes on the mean of the rainfall process (5) Identification of homogenous climatic
regions (6) Designing a neural network, and (7) Comparing forecasting skills between
neural network and time series models. This report describes the first four tasks. This
project is expected to finish by September 2003.

Principal Findings and Significance

(1) Data Collection:
The identified data set of monthly rainfall of the Virgin Islands includes 14 coop
stations and the records for most of them started in 1972. Our preliminary finding
shows that the oldest stations started in 1961 and these stations are: Wintberg on St.
Thomas and Annaly and Fountain on St. Croix. Table 1 shows the 14 coop stations.

Table 1. Coop Stations
Name I.D. Number Records
1 Wintberg, St. Thomas 24470 679450 1961 2001
2 Annaly, St. Croix 24424 670240 1961 2001
3 Fountain Valley, St. Croix 673150 1961 -2001
4 St. Croix FAA Hamilton 670198 1961 -2001
5 Beth Upper New Works, St. Croix 24426 670480 1972 2001
6 Caneel Bay, St. John 24431 671316 1972- 2001
7 Catheringburg, St. John 671348 1972- 2001
8 Charlotte Amalie Harbor, St. Thomas 678905 1972- 2001
9 Christiansted Fort, St. Croix 671740 1972- 2001
10 Cruz Bay, St. John 671980 1972- 2001
11 East End, St. John 672551 1972- 2001
12 East Hill, St. Croix 24444 672560 1972- 2001
13 Estate Fort Mylner, St. Thomas 672823 1972- 2001
14 Granard, St. Croix 24456 673677 1972- 2001

This work focuses on studying the monthly rainfall process of the longest four coop-
stations and some well-known meteorological indexes. Figure 1 shows the rainfall
observation for the most complete stations. Meteorological indexes were obtained
throughout the web pages of federal agencies. These indexes exhibit monthly
information during the period of 1961 to 2001. The studied meteorological variables are
the following: SST in the North Atlantic (5-200N, 60-30W), SST in the South Atlantic
(0-20S, 30W-100E), SST in Tropical Equatorial (10S-100N, 0-360). The SST in the
equatorial Pacific: el Niho 1-2 (0-10S, 90-80W), el Niho 3 (50N-50S, 150-90W), el
Niho 4, (50N-50S, 1600E-150W), and el Niho 3-4 (50N-50S, 170-120W). The data set
also includes the North Atlantic Oscillation index, the Artic Oscillation Index, the Sahel
Rainfall Index (20-8N, 20W-10E), the North Brazil Rainfall Index, Cold tong Index, and
Southern Oscillation Index.

Figure 1. Monthly Rainfall Records

(2) Estimation of missing values:
The available time series exhibited some missing values. Therefore, missing observations
were estimated to perform time series analysis and to develop the appropriate empirical
functions. A method to perform time and spatial interpolation was implemented to
estimate the missing values of the rainfall process (Ramirez et al., 2003). The
interpolation algorithm is a convex combination of spatial and time interpolation
methods. The convex combination can be written as follow:

y,,t = a,,t + (1 -a)A1,t, O a < 1 (1)

where a, is the convex coefficient at the ith station, z,,t is an estimate of the missing value
in the ith station at time t, K,, is the spatial interpolation in the ith station at time t obtained
by using the Kriging algorithm and A,,t is the time interpolation in the ith station at time t
obtained by the seasonal autoregressive integrated and moving average (ARIMA) model
(Storch and Zwiers,2001; Brockwell, and Davis, 1996; Matheron, 1979, Allard, 1998).

The convex coefficient was estimated using a cross-validation technique. A sufficient set
of rainfall observations with no missing values was selected to estimate the a,
coefficients. The selected data include about 120 observations, i.e., 10 years of data. The
selected time series were divided into two equal parts, the first one was used to fit the
ARIMA models and the second one was used to perform time and spatial interpolation.
The second part was also used to perform validation. Thus, 30% of the second part was
randomly eliminated for each series and the eliminated values were declared as missing
values. The Kriging algorithm and the ARIMA model were used to estimate the missing
values for each station. Since the complete data set is known the ai coefficients were
estimated using linear regression techniques. Table 2 shows the estimated convex
coefficient and the interpolation errors for each station. The interpolation errors are the
difference between the observed rainfall value and the corresponding interpolation value.

Table 2. Convex Coefficients and Interpolation Errors
Station Convex Average absolute
coefficient interpolation error
Wintberg 0.7523 2.7401
Annaly 0.7126 2.4516
Fountain 0.3316 2.0578
FAA, St. Croix 0.3361 2.0512

Table 2 shows that in Wintberg and Annaly stations the spatial-interpolation factor (K,,)
is more relevant than the time-interpolation factor ( A, ). However, the time interpolation
was dominated in the remaining stations. The average absolute error of the interpolation
method ranges from 2.05 to 2.74 inches. Wintberg, and St. Croix stations exhibit the
maximum and the minimum interpolation errors, respectively. Since the interpolation
algorithm provides a reasonable average absolute error it was used to perform the
interpolation tasks to estimate all missing values in the working data series.

(3) Develop time series models:
A monthly rainfall process for a given station can be considered as a sequence of random
variables. A sequence of random variables usually has deterministic and stochastic
components. The deterministic components are known as seasonal and trend
components. Figure 2 shows the periodogram for the rainfall of Wintberg stations. The
periodogram shows that the highest spectrum is given at the frequency 0.0833 i.e., the
period is 12. Periodogram also shows that at low frequency the spectrum is not
significant and consequently the trend component is not significant.

Figure 2. Periodogram for the Wintberg Station.

To model the rainfall process requires removing the seasonal component. Fitting a
periodic function and then subtracting it from the original process can remove this
component. A second alternative is to perform a twelve difference. Since the second
alternative requires the minimum computational effort, this method was implemented.
After removing the seasonal component the remaining process becomes a stationary
process. However, a mathematical transformation was needed to stabilize the variance of
the process. Finally the identified model includes a twelve and a first order
autoregressive process. The developed model for the Wintberg station is as follows:

(1 -B12)(1 (B12)(1 B)yt = zt

where y, is the natural log of rainfall observations at month t, z, is a white noise process
with constant variance and cero mean, B is the back shift operator, (1 B12) represents
the twelve difference, O and 0 are the twelve and first autoregressive parameters which
were estimated from rainfall observations. Table 3 exhibits the results of the parameter
estimation and Figure 3 presents the observations versus model fitting estimation. Figure
4 shows a one-year forecast for the Wintberg station.

Parameter Estimation

(Wintberg Station)
Std. Error

Figure 3. Rainfall Model Fitting at Wintberg Station

Figure 4. Rainfall Forecast for year 2002.

(4) Identify rainfall changes:
The main purpose of this task is to identify global climate changes throughout
meteorological index changes and test the interaction with changes on the U.S. Virgin
Islands rainfall. The available longest period of historical data is being collected for the
following variables: sea surface temperature, sea level pressure, rainfall process, and air
temperature. Most of the existing statistical tools to identify significant changes in the
mean or in the variance of sequential information have been designed for independent
time series (Cusum, Ewma, and Hotelling test). However, metereological time series are
characterized by being auto- and cross-correlated processes. New statistical techniques
are been tested and developed to identify significant changes on autocorrelated processes.
A statistical test has been tested and compared with recent techniques (Ramirez and
Sastri, 1997). Once the statistical test is selected, climate changes and Caribbean
interactions will be identified. For example, it is expected to test how the North Atlantic

Table 3.




Oscillation index is affecting the rainfall process in the U.S. Virgin Islands; how the
Sahel rainfall index interact with U.S. Virgin Island rainfall changes; how the Artic
Oscillation Index interact with U.S. Virgin Islands rainfall changes. etc.


a. Publications: A working paper is being developed to be submitted to the Journal of
Applied Meteorology.
b. Proposal submitted/granted: A proposal entitled: "Tropical Climate Research and
Meteorological Data Management" was submitted. This proposal was submitted to
the NASA program: Earth Science Research Education and Applications Solutions
Network (REASON: CAN-02-OES-01). This proposal is pending. A second
proposal entitled "Cloud Characterization and Time Series from Insolation
Measurements Predicting" was submitted. This proposal was submitted to NASA
IDEAS-ER Program and is pending. The Principal Investigator is Co-Investigator of
a proposal entitled "Simulating the Hydrologic Water Balance of Puerto Rico Using a
Coupled RAMS/LEAF-2/TOPMODEL/MODFLOW Modeling System." This
proposal was submitted to NSF.
c. Established partnership: Research collaboration with NASA Goddard Space Flight
Center was established. An official letter was received that research collaboration is
underway. A visit to this NASA Center is planned for Summer 2003.
d. Collaborations n i/h internal groups: Strong collaboration exists with remote sensing
and numerical modeling groups. The statistical group is working very closely with
Dr. Amos Winter, Dr. Ramon Vasquez, Dr. Jorge Gonzalez, Dr. Fernando Gilbes and
Dr. Joe Eastman.
e. Student Participation: Joan M. Castro is a student who is pursuing a Master of
Science Degree in the Electrical Engineering Department. He is performing some
statistical tests to detect climate change on meteorological indexes. Andrew S.
Garcia is an undergraduate student in the Mechanical Engineering Department.
He is organizing the meteorological indexes and rainfall data.


The major drawback that has been identified is that the longest records started in 1961.
The short records impose significant constraints to the proposed methodology.

General Future Projections

During the no-cost extension to this project, the remaining three major tasks will be
completed. It is expected that a set of time series models will be developed to predict the
expected rainfall one-year ahead at a specific station. The identification of significant
rainfall changes and their interactions with global meteorological changes is expected. It
is also expected that some correlations between meteorological indexes and rainfall
processes will be identified. These correlations may suggest the physical mechanism
that generates significant climate changes.


Allard D., 1998. Geostatistical Classification and Class Kriging. Journal of Geographic
Information and Decision Analysis, Vol. 2, No. 2, 77-90.
Box, G.E.P. and Jenkins, G.M. 1976. Times Series Analysis Forecasting and Control,
Hodlen-Day Oackland, California.
Brockwell, P.J., and Davis, R.A., 1996. Introduction to Time Series and Forecasting,,
Spring-Verlang New York Inc.
Matheron, G., 1971. The Theory of Regionalized Variables and Applications: Centre de
Geostatistique, Fontainebleau, France.
Ramirez-Beltran, N.D., and T. Sastri. 1977. Transient Detection with an Application to a
Chemical Process. Computers & Industrial Engineering, Vol. 32, No. 4, pp 891
Ramirez-Beltran, N.D., 2003. Time and Spatial Interpolation algorithm. A working
paper to be submitted to Apply Meteorology.
Storch, H.V. and Zwiers, W.Z. 2001. Statistical Analysis in Climate Research,
Cambridge University Press., Cambridge, UK.

Environmental Influences on Cistern Water Quality

Basic Information

Title: Environmental Influences on Cistern Water Quality
Project Number: 2002VI5B
Start Date: 3/1/2002
End Date: 2/28/2003
Funding Source: 104B

Congressional District: NA

Research Category: Water Quality
Focus Category: Water Quality, Water Supply, Treatment
Descriptors: Cisterns, Virgin Islands, Rainfall Harvesting
Principal Investigators: Henry H. Smith


Summary Report on

Environmental Influences on Cistern Water Quality

Problem and Research Objectives

About two thirds of the residents of the U. S. Virgin Islands are not served by public
water distribution systems. About eighty percent of Virgin Islands' homes rely on
rainfall harvesting as their main water supply. This practice generally involves using a
house's roof area for catchment of the rainfall, which is conveyed by gutters to a masonry
cistern that forms a part of the house's foundation. The water is pumped to a pressurized
tank for use upon demand.

Collection and storage of water by these means expose the harvested water to many
sources of contamination. These may include atmospheric and other depositions on the
catchment's surface, as well as contamination after entering the storage system which is
generally not tightly sealed. While the United States Environmental Protection Agency
has set water quality standards for all public water supplies, individual cistern systems are
not considered public water supplies and are not subject to these regulations. Cistern
water supplies then, may pose significant health risks to users in the Virgin Islands.

In this study, a broad survey of water quality in Virgin Islands cistern water supplies will
be conducted to assess the influence of surrounding environment on water quality of
cistern water supplies. Findings will provide information that will help to identify factors
that contribute to the degradation of cistern water quality. Having this information will
allow users of cistern systems to take steps to safeguard the water supply and reduce the
health risks associated with using water from these systems.


Homes located on St. Thomas and St. John were selected in a way that they represented
areas influenced by as many different factors as possible. These included population
density, varying degrees of annual rainfall, proximity to industry, shorelines and vegetal
cover. These homes were surveyed and information recorded on the physical
construction of the catchment, conveyance and containment structures, environmental
factors, and maintenance and management practices affecting the systems.

The most common form of contamination of cistern water supplies is by coliform
bacteria. Most of these bacteria are harmless and are free-living commensal organisms
that live in soil and water as well as the gut of animals. These bacteria then may be used

to indicate the general quality of water they are found in and are often used to estimate
the likelihood that the water is fecally contaminated. In this study, tests were made for
the presence or absence of total coliform and E. Coli bacteria. For sites where positive
results of any of these tests were obtained, the following months' tests involved
enumeration counts until negative results were obtained. The analysis procedure used
was the IDEXX's Quanti-tray enumeration method using the Colilert-18 and Entralert
media test kit for thermotolerant coliform and entercocci respectively. Media trays were
incubated for at thirty-five degrees Celsius (35C) for eighteen (18) hours. The results
were interpreted by a count of positive wells according to a most probable count table.

Physical characteristics of the samples recorded were temperature, pH and turbidity.
These were measured consistent with procedures proscribed (prescribed) by the
American Public Health Association and others in Standard Methods for the Examination
of Water and Wastewater.

Principal Findings and Significance

Sampling was conducted during the period June to September 2002 on St. Thomas and
St. John. While plans were initially to include St. Croix in the study, this was not due
primarily to logistically and related cost consequences. Most samples in the study were
taken three times, though this varied due to accessibility to the site, availability of
suitable water from harvesting in the system and so on. Not surprisingly, the cisterns
monitored tested generally positive for both Coliform and E. Coli during the study
period. While at some site chlorination was practiced, it was not done in such a
systematic manner that the stored water remained contaminant free during the monitoring
period. This is fairly typical for most Virgin Islands' homes.

Obvious roof contaminants were found most commonly to be leaf litter, droppings from
birds and other animals and dirt, soot and other sediments. Ponding due to blockages
were not unusual in the guttering systems. Cisterns examined revealed insects, dust and
debris floating on the water's surface and accumulation of sediments and other foreign
items on the bottom.

Management practices to maintain water quality were for the most part lacking. Roofs
were often easily accessible and overhung by vegetation. Cisterns too were easily
accessible due to poor-fitting manhole covers and inflow and overflow ports that were
inadequately screened. Devices to divert the first-flush of harvested water or in any other
way provide for rinsing the catchment's surface were not present and there were no
regular programs for adding disinfectants or cleaning conveyance or storage surfaces
with which the water came in contact.

The data gathered, as intended, not only provided a snapshot assessment of cistern water
quality and management practices associated with these supplies but also allowed
establishment of a database on several water quality parameters and site characteristics

that would be very helpful in future studies. Among these studies might be determination
of variances in cistern water quality with time, examination for correlation between
cistern water quality and overall public health in the Virgin Islands.

Information Transfer Program

Environmental Education and Hands-on Training on Mangrove
Restoration Techniques

Basic Information

Environmental Education and Hands-on Training on Mangrove Restoration

Project Number: 2002VI1B
Start Date: 3/1/2002

End Date: 2/28/2003
Funding Source: 104B

C gre l Not Applicable

Research Category: Not Applicable
Focus Category: Education, Non Point Pollution, Wetlands

Descriptors: Mangroves, vegetation restoration

Principal Richard Stephen Nemeth, Marcia Taylor


Summary Report on

Environmental Education and Hands-on Training on Mangrove Restoration

Problem and Objectives

Mangrove wetlands provide an important buffer between land and coastal marine
communities. They slow the flow of stormwater, filter runoff, and help remove
destructive pollutants before they enter fragile coastal environments. Unfortunately, the
number of mangrove areas in the Virgin Islands has significantly declined in the last fifty
(50) years because mangrove ecosystems have been routinely bulldozed to make space
for marinas and other coastal developments. Additionally, hurricanes have also taken
their toll on these wetlands.

Although mitigation has decreased the rate of mangrove habitat loss, limited awareness
and a poor understanding of the role played by mangroves continue to hinder these
efforts. There is a need to increase local awareness of the importance of mangroves in
the Virgin Islands and to demonstrate the benefits of restoring lost or damaged wetlands.

The objectives of the project are to train a target group of ten (10) high school students on
mangrove restoration techniques, and to increase awareness of the function of mangroves
in reducing non-point source pollution.

Methodology and Accomplishments

In order to achieve the objectives the staff of the Center for Marine and Environmental
Studies (CMES) first solicited for student interns to assist with the project (May 2002).
Putting a notice in the newspaper to alert high school students of the opportunity did this.
As a result, several students were given an orientation/training session on wetlands,
mangroves and the project (June 2002). Students were then given an on-site training
session at the University of the Virgin Islands (UVI) Wetlands and the area to be restored
was identified (June 2002).

The students assisted with the first mangrove planting in July 2002 by planting fifty (50)
red mangroves at the site. They also collected two hundred (200) black mangroves and
planted them in tree cones using a media of mud from the site and potting soil. The
germination rate was low, so additional seeds were planted in a media composed of a
higher percentage of potting soil. These seeds grew well and were put in the UVI
greenhouse for several weeks (July 2002-Sept. 2002). After the black mangrove
seedlings were about six (6) inches tall, they were removed from the greenhouse and
exposed to a greater amount of sunlight in an attempt to "harden" them to full exposure

(Oct. 2002). The black mangrove seedlings were planted at the site in Dec. 2002 and Jan.

Part of the program was to teach students how to conduct tours at the UVI wetlands. This
reinforced what they learned and also allowed other students to see their peers doing
outreach activities.

Students were given a training session on how to conduct wetland tours at the site and got
an opportunity to watch and assist with the tours. The students then conducted tours
while Center for Marine and Environmental Studies (CMES) staff watched. The students
were enthusiastic and seemed to enjoy the experience of giving tours (July 2002).

As an additional outreach component to this project a website was developed where basic
information on the project can be found (

Principal Findings

Monitoring conducted at the site showed that only a few red and black mangrove
seedlings had survived the transplant. A subsequent planting with differing conditions
(different elevations, level of shade) showed similar results (Jan. 2002).

Based on these results, additional research was done in an attempt to find a way to
increase mangrove survival at the site. The question was put out on the mangrove
discussion list and several responses were received. In addition, contact was made with
individuals who had been involved with mangrove restorations and discussed our results
with them. All researchers suggested an alternative approach to black mangrove
planting. They suggested that hand broadcasting of seeds be done repeatedly rather than
starting the plants growing in the greenhouse. They indicated that a low survival rate
should be expected, less than ten percent (<10%), which has still a higher survival rate
than trying to transplant them. It was determined that transplanting easily shocks black
mangroves and therefore survival is often low.

Because of the findings, a decision was made to request an extension of the grant in order
to do additional plantings. Although the transplant results were a bit disappointing, the
enthusiasm and initiative shown by the interns was inspiring. Thus the requested no-cost
extension will also make allowance for the recruitment a new group of local high school
students to this project and provide additional opportunities in conservation techniques
for Virgin Islands youth.

Future Plans/During Extension Period (Feb. 2003-Aug 2003)

- Identify additional interns available (Feb. 2003 March 2003)
- Collect and plant black and red mangrove seeds/propagules (March 2003)
- Monitor success of seedlings (March 2003)

- Collect additional seeds/propagules as needed
- Monitoring seedlings (March 2003-Aug.2003)
Train interns about UVI Wetlands (March 2003-April 2003)
Run Wetland tours with interns (March 2003-Aug. 2003, based on availability of tours)
Final report (Aug. 2003)

Students involved with the proiect to date and academic level

Lamin Jackson
Daniel Hodge
Amaris Chew
Candace Cornwall
Michael Camacho
Manuel Camacho
Amalee Lockhart


Good Hope School
Educational Complex
Educational Complex
Educational Complex
Central High School
Central High School
Educational Complex

USGS Summer Intern Program

Student Support

Student Support

Category Section 104 Section 104 NIWR-USGS Supplemental Total
Category Total
Base Grant RCGP Award Internship Awards
Undergraduate 1 0 0 0 1
Masters 1 0 0 0 1
Ph.D. 0 0 0 0 0
Post-Doc. 0 0 0 0 0
Total 2 0 0 0 2

Notable Awards and Achievements

Publications from Prior Projects

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs