Printer Friendly

GeoFIS flood insurance system for Trinidad: a case study for San Juan downstream.


Flood is one of the most common natural disasters resulting in threats to life and property throughout the world (Sharma and Priya 2001). Flooding occurs when heavy and continuous rainfall exceeds the absorbing capacity of the soil or the flow of the water is greater than the normal carrying capacity of a stream channel. Statistically, streams equal or exceed the mean annual flood level once every 2.33 years (Leopold et al. 1964) and cause streams to overflow their banks onto flanking lands. Flood often accompanies other natural disasters such as brief torrential rain, monsoonal rain, cyclones, hurricanes, or tidal surges (Brakenridge et al. 2004). In addition, increasing impermeable layers, such as roads, residential buildings, and industrial complexes, reduce the land's natural ability to absorb water, which increases runoff as well as disturbs the natural water flow, thus increasing the risk of flooding (Ramroop 2005).

In Trinidad, flood is one of the major hazards affecting the country every year and during all seasons (Ramroop 2005). In recent years, the number of flood occurrences has increased throughout the country. In addition to the previously mentioned common causes, factors contributing to flood occurrences in Trinidad are particularly indiscriminate dumping into streams and improper or illegal hillside land development and agricultural practices (WRA/MIN. Env. 2001). Flood damages can be categorized as physical damages to houses and infrastructure, casualties of people and livestock as a result of drowning, spreading of diseases, scarcity of clean drinking water because of water contamination, and damages to food crops (Mileti 1999). According to Mileti (1999), flood hazards severely impede the economy of the United States; translated into the context of Trinidad, damage caused by flooding events in 1993, 2002, and 2006 are $580,000, $3,300,000, and $2,500,000, respectively (WRA/MIN. Env. 2001, Brakenridge et al. 2003, Brakenridge et al. 2007).

After a decade of economic growth, mainly driven by the energy sector (IMF Country Report 2005), housing development in Trinidad has increased considerably even in flood-prone areas. Economic values of houses have increased with the use of costly fixtures, which further add to the losses. Unfortunately, flood insurance has not kept up with housing development and insurance providers lack the tools to properly predict potential losses and recommend mechanisms to benefit both parties in the insurance market. The insurer, more often than not, is an agent in a chain of transfer of premiums in return for potential compensation. This kind of risk transfer is depicted in Figure 1.

However, potential clients are not readily purchasing flood-insurance policies because of high premiums (Browne and Hoyt 2000, Miller 1997, Preist et al. 2005). Thus, implementing flood insurance for private households with affordable premiums is in the best case difficult and in the worst case plainly not profitable (Miller 1997). For these reasons, it is very important to classify areas based on their flood risk. Geographic information systems (GIS) can be used to categorize flood-risk zones by analyzing complex spatial data sets from different sources (Gangai et al. 2003). In this study, GIS forms the basis for a private household flood-insurance system for Trinidad to calculate premiums based on household exposure to flood risk and to speed up the underwriting process.



Trinidad is situated at the southernmost end of the Caribbean island chain located at latitude 10.5[degrees] N, longitude 61.5[degrees] W, and is approximately 5,126 km2 in size. The climate of Trinidad is tropical wet, with an average rainfall of 2,200 mm (WRA/MIN. Env. 2001) and its monsoonal character results in high-intensity rainfall and subsequent frequent flooding (Bryce 2007).

The flood history of Trinidad shows that the frequency and intensity of flooding events is increasing (Bryce 2007). Based on information collected from newspaper articles (Maharaj 2006), the Water and Sewerage Authority (WASA), and the Office for Disaster Preparation and Management (ODPM), we mapped more than 100 locations in Trinidad that have been flooded in 1986-2006 (see Figure 2). In four locations, floods have occurred ten or more times within the past 20 years. More than 30 of these locations are in high-density settlement areas and floods in these areas cause significant economic damages. Typically, they occur in brief storms associated with sheet or surface flow (Baban and Kantarsingh 2005).

It is widely documented (e.g., Chan 1997, Smith 1991, Baban and Canisius 2007) that alluvial planes prone to flooding also are often densely populated and contain highly built-up areas vulnerable to flooding. Figure 2 shows that this holds true for Trinidad as well.


Based on Figure 2, significant areas in Trinidad are flood prone and coincide with residential developments. Therefore, a need exists for introducing a flood-insurance system for Trinidad to cover financial losses caused by flooding. The adaptation of the British flood-insurance system has proven unsuitable, for many householders who are living out of a flood-prone area would have to pay higher insurance premiums. This is because UK insurers traditionally determine flood-risk premiums on the basis of administrative boundaries/postcode bands rather than on particular addresses (Ordnance Survey 2007).


The GeoFIS flood-insurance system simplifies the process of risk assessment of private households by integrating GIS, allowing insurers to verify and evaluate the flood-risk level of a property and to fix a premium. Based on a GIS, the operator may zoom in on the house to be insured for a visual clarification. There are five main components to this system (see Figure 3): (1) spatially identify a particular property located in a flood-prone area; (2) analyze the vicinity of flood boundaries to predict future chances for flooding; (3) classify the flood-risk level of the house based on the flood-prone area and considering previous flood-event statements by clients and number of insurance claims; (4) estimate area, age, and number of stories of the house and calculate the house's value, including other house information, such as construction of the house and permanently installed fixtures; and (5) calculate the premium based on the flood-risk class.


To classify flood risk, a house's location is identified to determine whether it is located inside or outside the flood zone. If the house is identified as lying in a flood zone, the flood-recurrence interval is analyzed in a second step. If the house is located outside the flood zone, the likelihood for flooding is determined by calculating the elevation difference between the property and the nearest flat plain, where river and drainage channels pass through.


Three ArcView Avenue scripts implement the outlined approach: (1) identification of a property location on a floodplain, (2) calculation of the distance to the floodplain, and (3) determining the floodplain in the first place:

To identify whether a house is located inside a floodplain, first retrieve the address polygon using the address ID. Next create x and y coordinates for the retrieved address polygon and create a point feature for it. Then intersect the created house point with the flood boundary and determine whether the house is located inside the flood boundary. Finally, check the flood-recurrence interval of the house that was identified inside the flood boundary.

If the house is located outside the flood boundary, then determine the elevation difference between the house elevation and the nearest flood boundary elevation. Obtain the elevation of the house by intersecting the house point with the average elevation. Then find the elevation of the flood boundary by intersecting the flood polygon with the buffered house polygon by the calculated minimum distance and obtain the smaller value of the two. Next create a point as described previously to intersect the point with the average elevation and determine the elevation of the flood boundary. Finally, calculate the elevation difference by subtracting the elevation of the house from the elevation of the flood boundary.

The location of a house in a floodplain, where river and drainage channels pass through, is identified in the following steps. First, find that the house is located in a floodplain by intersecting the house point calculated in (1) or (2) with the floodplain polygon. Then intersect the river or drainage channels polygon and retrieve the flat plain polygon to ensure that the river or drainage channel crosses the identified floodplain.


In the assessment of the house value, its size, age, and number of stories are used. With the assessed house value, the value of permanent fixtures (built-in dishwasher, hot-water heaters, shelving and cabinetry, plumbing fixtures, stoves, ovens, refrigerator, and air conditioner) and the construction of the house (varieties of wall, floor, roof, and window) are added to calculate the total value of the house. Using MS Access, derive the area of the house and multiply the derived area by the number of stories to obtain the total area of the house. Next, multiply the total area of the house by the market price of square feet. At this point, considering the age of the house, add the percentage of the house value and calculate the total value of the house. The area, number of stories, age, and total value of the house are subjected to verification by the client. Extra tables and procedures are encoded to update market prices and changes in the variability of the price of square feet, permanently installed fixtures, and construction of the house, and the percentage of the premium, the percentage of the discount of the flood-risk classes, and the percentage of the house value with regard to the age of the house.


Case Study of San Juan Downstream

We selected San Juan as a study area for GeoFIS; it is the third largest city of the country and undergoes sizable developments, even in floodplain areas.

Data Collection

Flood data, houses, roads, rivers, elevation data, and aerial photos (shown in Figure 4) were collected from the Department of Surveying and Land Information, University of the West Indies. The 1994 and 2003 aerial photos were used to update house data and to estimate the number of stories and the ages of houses. A site visit was performed for some ground truthing. This included getting experts to estimate the square-foot market value. In addition, personal-level information about the client and the house from the application files was obtained.

Acquire Area, Age, and Number of Stories of House

To calculate the area of each house, we updated our files based on a 2003 aerial photograph mosaic that we created using ERMapper software. The house data then were digitized and updated using ArcView (see Figure 5A).


To determine the age of a house, we employed multidate aerial photographs. A RGB color composite was developed using multidate aerial photographs obtained in 1994 and 2003 (red: 1994, green: 2003, blue: 1994). We then classified houses as either less than ten years old (green color house in Figure 5B) or more than ten years old (white color in Figure 5B).

To identify the number of stories of a house, we developed a stereo model using two consecutive aerial photographs taken directly one after the other with about 60 percent overlap of the area. This was done by DVP digital photogrammetry software, adjusting interior, relative, and absolute orientations. The height of the house was measured from the 2003 stereo model (Figure 5C). A height of less than 3.5 m was considered a single story and each 2.5 m above a single story was considered one additional story. These houses' heights were further confirmed during field visits to the study area. The area, number of stories, and age of the house were subjected to cross-check with the information provided by the house owner before calculating the house value.


With the GeoFIS flood-insurance system, insurance premiums for a house are calculated based on its flood-risk class in relation to the house's location (see Figure 6). To identify the flood-risk classes, the following four criteria were used: (1) The house is located inside the flood boundary; (2) the flood-recurrence interval of the flood boundary is less than or equal to five years; (3) the elevation difference between the house elevation and the elevation of flood is less than two meters; and (4) a waterway crosses flat land (less than 1 percent slope). If flooding in a particular area is very frequent (the flood-recurrence interval of the flood boundary is less than or equal to five), the houses in the flood boundary are classified as very high or high risk. In our study area, flooding is very frequent; therefore, the houses in the flood boundary are classified as very high risk. This procedure is summarized in Figure 6.

When we applied our criteria to actual flooding data (Figure 7 and Table 1), we found that 1.47 percent of very low risk, 7.59 percent of low risk, 17.8 percent of medium risk, and 36 percent of very high risk classes were flooded in the past. These percentages are encouraging, although we would obviously prefer to get a better handle on those judged to be low risk. We assume that a significant number of these houses were flooded because of other reasons such as improper drainage or drainage blocks that were not considered in this study.



The GeoFIS flood-insurance system was developed to determine the flood risk of private properties. The system requires high-resolution satellite and aerial imagery to derive a detailed flood map, which would be expensive to implement for the entire country. However, in Trinidad, the frequency of flooding, subsequent financial loss, and rapid development of built-up areas mandate that this system be implemented.

According to the Federal Citizen Information Center (FCIC) in the United States, about 25 percent of all flood-insurance claims come from outside the Federal Emergency Management Agency (FEMA) classification of high-risk areas. Available flood maps of Trinidad do not have the necessary resolution to truly represent the actual probability of flood danger of each individual private home. Ramroop (2005) recommends that the National Emergency Management Association (NEMA) be authorized to develop maps of flood-prone areas. SAR data is one possible source for the development of flood maps (Canisius et al. 1998). Nongovernmental organizations (NGOs) in Trinidad also showed their interest in developing flood maps using hydrological models.


The system will require regular updates. For instance, in the ten-year period from 1994 to 2003, the Bamboo Grove settlement increased by about 20 percent and the expansion of a highway may have changed floodplain conditions. This updating, however, will not affect the core function of the system, where separate lookup tables are used for variable parameters.


Insurance is a business of transferring risk. Understanding insurance in general and using GIS data in particular provides valuable input to realistically analyzing flood risk. Higher accuracy in risk assessment will help to prepare for likely increases in flood events that will enable all parties to make use of flood insurance for their advantages. The GeoFIS flood-insurance system was developed by integrating GIS into a general-purpose home-insurance system to improve processing and calculate fair premiums based on the flood-risk class of each property. Not only is this system useful for premium calculation but it also educates and prepares the entities of the insurance market about future flood perils.

The system has classified five flood-risk classes; they are: very high, high, moderate, low, and very low. By this classification, the system has provided clients a fair premium discount according to the vulnerability of their houses. This system offers advantages for both parties of the flood-insurance market: Clients can obtain the flood insurance and pay premiums based on the vulnerability of the flooding of their respective homes; insurers, on the other hand, can promote and sell their flood insurance to those homeowners who promise a long-term profit.


We express our sincere thanks to Dr. Jacob Opadeyi and Dr. Bheshem Ramlal of the Department of Surveying and Land Information, University of the West Indies, Trinidad and Tobago, for providing the data used in this study.


Baban, S., and F. Canisius. 2007. GIS methodology for identifying and mapping flood prone areas in Trinidad. In S. Baban, Ed., Enduring geohazards in the Caribbean, Chapter 9. Trinidad and Tobago: UWI Press, 2007.

Baban, S. M. J., and R. Kantarsingh. 2005. Mapping floods in the St. Joseph watershed, Trinidad, using GIS. International Association of Hydrological Sciences 295: 254-64.

Brakenridge, G. R., E. Anderson, and S. Caquard. 2003. Global active archive of large flood events, 2002 global register of extreme flood events. Hanover, NH: Dartmouth Flood Observatory, Hanover NH,

Brakenridge, G. R., E. Anderson, and S. Caquard. 2004. Global and regional analyses, world atlas of large flood events. Hanover, NH: Dartmouth Flood Observatory,

Brakenridge, G. R., E. Anderson, and S. Caquard. 2007. Global active archive of large flood events, 2006 global register of extreme flood events. Hanover, NH: Dartmouth Flood Observatory,

Browne, M. J., and R. E. Hoyt. 2000. The demand for flood insurance: empirical evidence. Journal of Risk and Uncertainty 20(3): 291-306.

Bryce, R. 2007. Trinidad and Tobago report. Caribbean Land and Water Resources Network (CLAWRENET) and Hydrologist at the Ministry of Agriculture, Land and Marine Resources (MALMR), Trinidad and Tobago,

Canisius, F. X. J., H. Kiyoshi, M. K. Hazarika, and L. Samarakoon. 1998. Flood monitoring in the central plain of Thailand using NOAA/AVHRR and JERS-1 SAR data. 24th Annual Conference and Exhibition of the Remote Sensing Society, UK, September 9-11, 1998.

Chan, N. W. 1997. Increasing flood risk in Malaysia: causes and solutions. Disaster Prevention and Management 6(2): 72-86.

Gangai, J., J. B. Lee, Dewberry and Davis. 2003. A case study: utilizing GIS tools to aid in the production of flood insurance rate maps for coastal communities, Proceedings of the 3rd Biennial Coastal GeoTools Conference, Charleston, SC.

IMF Country Report. 2005. Trinidad and Tobago: selected issues. International Monetary Fund Report No. 05/6. Washington, D.C.: International Monetary Fund, Publication Services.

Leopold, L. B., M. G. Wolman, and J. P. Miller. 1964. Fluvial processes in geomorphology. San Francisco, CA: W. H. Freeman.

Maharaj, A. N. 2006. Methodology for identifying and mapping flood prone areas in Trinidad using GIS. BSc Research Project, University of the West Indies, Trinidad and Tobago.

Miller, J. 1997. Floods: people at risk, strategies for preservation. New York: United Nations.

Mileti, D. S. 1999. Disasters by design. Washington, D.C.: NAS Joseph Hentry Press.

Ordnance Survey. 2007. Case studies. Great Britain's national mapping agency, oswebsite/business/sectors/insurance/news/casestudies/ raisingstandardfloodrisk.htm.

Priest, S. J., M. J. Clark, and E. J. Treby. 2005. Flood insurance: the challenge of the uninsured. International Journal of Geographical Information Science 37(3): 295-302.

Ramroop, S. 2005. Proposed flooding analysis research using GIS for sample areas in Trinidad and Tobago. American Congress on Surveying and Mapping, California Land Surveyors Association, Nevada Association of Land Surveyors, Western Federation of Professional Surveyors, Conference and Technology Exhibition, Nevada, March 18-23, 2005.

Rosenbaum, W. 2005. The developmental and environmental impacts of the national flood insurance program: a review of literature. Washington, D.C.: American Institutes for Research.

Sharma, V., and T. Priya. 2001. Development strategies for flood prone areas, case study: Patna. India Disaster Prevention and Management 10(2): 101-9.

Smith, K. 1991. Environmental hazards: assessing risk and reducing disaster. Routledge Taylor and Francis Group.

WRA/MIN. Env. 2001. Integrating the management of watersheds and coastal areas in Trinidad and Tobago. Water Resource Agency, The Ministry of the Environment.

Dr. Francis Canisius is currently a visiting scientist at Canada Centre for Remote Sensing, Natural Resources Canada, and he was attached with the Department of Surveying and Land Information, University of the West Indies, Trinidad and Tobago.

Ms. Sophia Nancy received her BSc. in Information Systems and Management from University of London, UK and she is a licensing specialist at Adobe Systems Inc. Ottawa, Canada.
Table 1. Classification of Houses into Risk Classes

Flood-risk No. of Houses No. of Flood-ed % Flooded
Class Houses

Very low 681 10 1.47
Low 580 44 7.59
Moderate 680 121 17.8
High 0 0 0
Very high 100 36 36
Total number 2041 211 10.34
of houses
COPYRIGHT 2009 Urban and Regional Information Systems Association (URISA)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Canisius, F.; Nancy, C.
Publication:URISA Journal
Article Type:Case study
Geographic Code:5TRIN
Date:Jan 1, 2009
Previous Article:Mapping the future success of public education.
Next Article:St. Kitts land resource analysis.

Related Articles
Upstream changes and downstream effects of the San Marcos River of central Texas.
Upstream changes and downstream effects of the San Marcos River of central Texas.
Nature conservancy takes on China's dams.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |