Volume:9
Issue:02
Year:
2009
Abstract
Dengue fever over the last 40 years has become one of the most serious epidemic diseases in Malaysia. In 2005, the number of dengue cases increased dramatically and it became the worst epidemic in the nation’s history. Recently, remote sensing and Geographic Information System (GIS) technologies have provided an alternative potential tool for infectious disease surveillance and the control of many types of vector borne diseases. In this study, these technologies were used for the surveillance of infectious diseases particularly on a possible dengue outbreak that had been initiated in 2003. The selected study area was Subang Jaya, an area of rapid urbanisation located about 20 km from Kuala Lumpur. The aim was to identify high risk areas for a dengue outbreak using remote sensing and related datasets in a GIS database. The weighted overlays function was used in the analysis and the modeling process to identify the dengue risk areas.
The environmental factors derived from remote sensing data include land cover/use, topography, Land Surface Temperature (LST) and Normalised Difference Vegetation Index (NDVI), temperature, population density and clinical data of dengue cases collected from various agencies. The results showed that the high risk areas for dengue outbreak were associated with areas of high population density, topographically low land areas and high land surface temperatures (LST). Most of the victims were in residential and commercial areas near construction sites and epidemics usually emerged after days of heavy rainfall followed by high temperature. The environmental factors identified from remote sensing
data provide sound indicators of areas that are susceptible to dengue outbreaks and provide the dengue distribution pattern in the study area.
Key words:
Dengue outbreak; environmental health; environmental factors; Graphical Information Systems; GIS; remote sensing.
Introduction
Dengue fever has been epidemic in Malaysia since the early 1970s and the number of cases has continued to rise, especially in the late 20th century. Outbreaks usually happen during the rainy season especially when temperatures are high. Barbazan et al., (2002) found two main patterns, which might describe the fluctuations of Dengue Haemorrhagic Fever (DHF) incidences. In a cyclic pattern, incidences of dengue are high during the annually hot and rainy season and this corresponds to seasonal variations of transmission. A non-cyclic pattern shows increases of DHF cases for a variable duration separated by periods of fewer cases lasting two to five years.
It is a challenge for the public health system in Malaysia to make sure an epidemic of this scale does not reoccur. The Malaysian Ministry of Health has reported that the high incidence of dengue in this country was probably owing to an increase in potential mosquito breeding places such as construction sites, clogged drains and accumulated rubbish heaps. The number of construction areas may influence the increase of dengue cases in urban areas. The incidence of dengue is greatest in developed states with a high population density, rapid development and many construction sites.
According to the World Health Organization (WHO, 1997) the interactions between temperature and rainfall are important as determinants in a dengue transmission, as cooler temperatures affect the survival of adult mosquitoes, thus influencing transmission rates. Furthermore, rainfall and temperature may affect patterns of mosquitoes feeding and reproduction, and hence the population density of vector mosquitoes.
Figure 1.0 Dengue cases reported in Malaysia for year 2000-2005.
Previous studies have identified a number of key factors that contribute to a dengue outbreak. WHO (1997) mentioned that in areas of high human population density, many people might be exposed, even if the mosquito house index shows a low-density value (low mosquito numbers). Distances between houses may thus be of an epidemiological significance, especially in densely packed housing areas. The intensity of dengue transmission varies with the population density of the vectors present, the numbers of non-immunised people and the number of individuals ill with the disease.
The aims of this study are (i) to identify the environmental factors that contribute to the dengue outbreak using remote sensing data; (ii) to correlate the identified environmental factors with dengue occurrence pattern; and (iii) to predict potential high risk areas for a dengue outbreak.
Remote sensing and GIS
Recent advances in remote sensing technology have provided crucial information on Dengue transmission. Remote sensing provides up to date information on soil moisture, vegetation type, land cover/use, urban planning, crop monitoring, forestry, water and air quality that influence the vector borne disease occurrences. Maynard (2002) found that many of the environmental factors connected to the public health issues are observable through remote sensing, such as air and water quality, thermal extremes, ultraviolet radiation, oceanic harmful algal blooms, as well as pollutant/pathogen transport and deposition via the atmosphere, oceans, ice and rivers. The application of remote sensing together with GIS in health studies has increased especially in the monitoring, surveillance or risk mapping of vector-borne diseases. Most of these studies
used remote sensing data to explore the environmental factors that might be associated with disease-vector habitats and human transmission risk.
A Geographic Information System (GIS) is a computeraided database management and mapping technology that acquires, organises stores and integrates large amounts of multi-purpose information from different sources, programmes and sectors. GIS adds the dimension of geographic analysis to information technology by providing an interface between the data and a map. Tasks such as temporal modelling of climate changes, environmental degradation, disease transmission and other factors relevant to an outbreak can then be easily analysed. This provides rapid information to key decisionmakers quickly, efficiently and effectively.
Methodology
The methodology used in this study comprised five major parts; data acquisition, pre-processing, processing, creating a GIS database and spatial analysis and modeling. Satellite data of IKONOS, SPOT, Radarsat and Landsat-TM were processed and analysed to produce information on landcover/landuse changes, Land Surface Temperature (LST), Normalised Difference Vegetation Index (NDVI), and Digital Elevation Model (DEM). Environmental factors such as rainfall, temperature, population density and clinical data of dengue cases were integrated into the GIS database. This data was then correlated using the GIS spatial analysis, Weighted Overlay function and modeling techniques in order to identify potential high risk areas for a dengue outbreak. The Weighted Overlay function is a technique for applying a common measurement scale of values to diverse and dissimilar inputs in order to create an integrated analysis.
Study area
Subang Jaya is a district in the state of Selangor with an area size of 181 km2. It is located from from 3° 05’ 48.74” N 101° 33’ 02.39” E to 2° 58’ 22.93” N 101° 44’ 39.69” E. The Subang Jaya Municipality was selected as the study area because of the high occurrence of dengue disease in that area. It is surrounded by areas of rapid development and has a high population density ratio of 437,121 per km2 (Population and Housing Census of Malaysia 2000).
Environmental parameters
The first major phase of the data collection was to identify the environmental factors, which had significantly influenced the dengue distribution pattern (e.g. land use, slope, vegetated or non vegetated areas and housing types). All of the environmental data was generated from remote sensing data. This technique has been tested successfully by Connor et al., (1996) Four environmental parameters were identified (land use, NDVI of vegetated and nonvegetated areas, LST and population density).
Land use map
The land use map was derived from IKONOS (1 meter spatial resolution) imagery using supervised classification. It was classified into nine classes, which includes cleared land, construction, and industrial, commercial, recreational, residential, green area, oxidation pond and water body.
Normalised Difference Vegetation Index (NDVI)
The NDVI map was derived from the Landsat-TM image by using the visible and near infra-red bands. NDVI information was used to differentiate vegetated areas from non-vegetated areas. Topographical information was used to indicate low land areas, which were prone to forming stagnant water pools. These stagnant water pools could provide suitable mosquito breeding sites. To detect such topographical high or low land areas, satellite radar data was used to generate a Digital Elevation Model (DEM) of the study area.
Land Surface Temperature (LST)
Land Surface Temperature (LST) information was derived from the Landsat-TM dataset using the thermal band (band 6) that has the radiance value of the land surface. Band 6 is converted into radiance with the
following formulae (NASA, 2003):
(equations to follow)
Epidemiological data:
The second phase involves data collection on dengue incidence cases of the year 2006 from MPSJ and the Malaysian Ministry of Health. Sample cases with detailed information such as name, gender, address, serology status, date admitted in hospital and so on were included
in the analysis. This data included information about all the suspected and confirmed DF/DHF cases reported during the year 2006 in the MPSJ district. Data on population density was obtained from the Statistical Department.
Data Analysis:
Each of the four variables was tested using the weighted overlay function technique in the ArcGIS software. This technique is usually used for applying a common measurement scale of values to diverse and dissimilar inputs in order to create an integrated analysis. The priority value was ranked as low, medium and high (1 to 3) for each variable. A low value means the sub variable had a low intensity influence; a medium value equated to a greater risk influence to the outbreak and a high value equated to a very significant influence on the dengue outbreak pattern. The detailed weighting values for all four environmental variables that were identified as an indicator factor of dengue outbreak are presented in Table 1.0.
For the information obtained, the following algorithm was used to develop the dengue risk zone from each environmental indicator as shown below.
Dengue Risk Zone
(Land use (A)) + (Population Density (B)) + (NDVI (C)) + (LST (D))
Results and discussion
This study found that areas at a high risk of having a dengue outbreak could be identified through the integration of environmental factors derived from remote sensing with other data from GIS analysis and modeling. The weighted overlay function was chosen to create the risk area map. The following sections provide a detailed analysis of the impact of each variable:
The correlation of environmental factors to dengue distribution
Land cover and land use classification
Figure 2.0 shows the land cover and land use information obtained from the IKONOS (1m resolution) images. From the classification result, it was found that most cases occurred in urban areas, followed by mixed horticulture areas and some in construction areas. Just a few cases were reported in the industrial and forested areas. The reason why dengue cases occur mostly in urban areas can be explained by a number of factors. For an urban area, proper infrastructure such as a good drainage system is very important. A poor drainage system will create pools of stagnant water, which are suitable breeding grounds for mosquitoes. The same problem occurs in construction areas and squatter areas. In areas with a high population density (such as flats, apartments and condominiums), the population density per square metre is high. As a result dengue transmission can and will occur rapidly.
Table 1.0 The Weighting value for Environmental Risk Indicator
Dengue fever had also been reported in low-density areas (such as high class residential areas and small housing estates). This has been attributed to the fact that as most people like to create mini landscapes or gardens with ponds in their compounds, the ponds can become suitable breeding grounds for mosquito whenever proper preventive methods are not taken. Other areas that also have reported cases of dengue outbreak are areas that are not cleanly maintained, areas that have been left idle, areas with rapid development and areas having temporary structures.
Land surface temperature (LST)
The temperature profile of the land surface over the study area is shown on the LST map in Figure 3.0 which was derived from the Landsat TM thermal band (channel 6). The areas with high levels of LST (in red) can be correlated to urban areas, while the white area equate to either a vegetated area or a water body. The LST map shows that the temperature ranges from 21.0ºC to 30.0ºC. Most of the reported dengue cases occurred in the areas with a temperature range from between 25.0ºC to 30.0ºC. This temperature range is very conducive to the mosquito breeding cycle as an
increase in the number of times that the mosquito breeds will also increase the likelihood of the emergence of the dengue outbreak.
Normalised Difference Vegetation Index (NDVI): NDVI is usually used to derive the vegetation index from satellite images. For this study, it was used to identify the green areas over the study area. Figure 4.0 shows that the range of NDVI values in the study area are between 21 and 218. Built-up areas are shown in white, while the green areas refers to the vegetation density of an area. This study found that the vegetation density (‘greenness’) of an area was not a major factor in influencing the number of dengue incidences.
Figure 2.0 Land cover vs. dengue cases of Subang Jaya
Figure 3.0 The distribution of Land Surface Temperature from Landsat TM.
The relationship between other ancillary data with the dengue incidence
Analysis of dengue incidence with population density
Population density in any urban area is another factor that has to be taken into account in preventing an outbreak of dengue fever. Therefore in this study we used IKONOS images to determine areas with a dense population in the district of Subang Jaya. These areas have their own unique characteristics that are easily identifiable and can be located visually. Population data was also used to verify the result. A densely populated area stands a higher chance of experiencing a dengue outbreak even if the mosquito house index in that area is low. This is because the Aedes aegypti mosquito does not have to travel far to search for its victims. Therefore an outbreak of dengue fever can and will be able to spread rapidly in such an area.
Figure 4.0 Normalised Difference Vegetation Index (NDVI) map
Figure 5.0 Population density map of Subang Jaya
Identification of Dengue High Risk Areas in Subang Jaya
Identification of areas with a high risk of having a dengue outbreak requires the input of the above stated parameters for this analysis. The analysis results of the parameters were then given a specific priority value based on the requirements of this study. The priority values were ranked as ‘low’, ‘medium”, ‘high’ and ‘very high’. The contribution of the these values in every spatial layer were given a value between 1 (low) to 4 (very high), where a value of 1 means a very low contribution while a value of 4 means a very high contribution to the dengue outbreak. Areas with the highest score can then be identified as being areas with a very high risk of having a dengue outbreak.
The potentially high risk areas for the occurrence of dengue incidences over the study area are shown in Figure 6.0. Reported dengue cases data obtained from MPSJ was used to verify the above result. The identified ‘very high risk’ areas are Seri Serdang, Seri Kembangan and USJ. The result shows a strong correlation between locations of reported dengue cases with the potential high risk area map, which was created based on environmental factors used to identify dengue outbreak risk areas.
Figure 6.0 Risk area of dengue fever outbreak
Conclusions
- Remote sensing satellite data such as Landsat TM, SPOT, Radarsat and IKONOS are capable of providing information on the environmental factors: land cover/use, land surface temperature (LST), NDVI and topography, which are influential to a dengue outbreak. The high risk areas for a dengue outbreak are significantly correlated with environmental factors obtained from remote sensing data which were then integrated with rainfall, temperature, humidity and population density data.
- The study found that most of the victims were staying in the densely populated commercial areas near the construction sites, which were located at topographically low land areas with surrounding high values of land surface temperature (LST). The epidemic normally occurred in these types of areas after days of heavy rainfall followed by high temperature.
- Remote sensing and GIS technologies were found to be an important tool for the effective surveillance and prediction of the dengue outbreak in order to reduce the number of dengue cases. GIS analysis has the ability to model a risk map of dengue distribution through the use of the weighted overlay function, which enabled the users to easily identify high risk areas in a short time period. This initial finding points the way to the wider application of this technology by the relevant authorities to improve monitoring of potential future dengue outbreaks.
Acknowledgements
The authors would like to thank the Director of Majlis Perbandaran Subang Jaya (MPSJ) for providing ground data on dengue cases for this research work. The contribution of meteorological data from the Malaysian Meteorological Department is also duly acknowledged. Appreciation is also extended to Professor Abu Hassan Ahmad, Dean, School of Biological Sciences, Universiti Sains Malaysia, Malaysia, for his advice.
References
Alharthy A (unpublished). Role of GIS in Dengue Control Management Strategy at Jeddah. Jeddah Municipality, Jeddah, Saudi Arabia. Available online at: www.saudigis.org/FCKFiles/File/33_E_AbdullatifAlharthy_KSA.pdf [accessed 28/09/09]
Barbazan P, Amrehn J, Dilokwanich S, Gonzalez J-P, Nakhapakorn K, Oneda K, Thanomsinra A, Yoksan S (unpublished). Dengue Haemorrhagic Fever (DHF) in the Central Plain of Thailand. Remote sensing and GIS to identify factors and indicators related to dengue transmission. Available online at: http://std.cpc.ku.ac.th/delta/conf/Acrobat/Papers_Eng/Volume%201/Barbazan.pdf [accessed 28/09/09].
Department of Statistics Malaysia (2001). Population and Housing Census of Malaysia 2000. Available online at: http://www.statistics.gov.my/portal/index.php?lang=en [accessed 28/09/09]
Cringoli G, Taddei R, Rinaldi L, Veneziano V, Musella V, Cascone C, Sibilio G, Malone J B (2004). Use of remote sensing and geographical information systems to identify environmental features that influence the distribution of paramphistomosis in sheep from the southern Italian Apennines. Veterinary Parasitology, 122, 15-26.
Gong Peng X B (2006). Remote sensing and geographic information systems in the spatial temporal dynamics modeling of infectious diseases. Science in China Series C: Life Sciences, 49, 573-582.
Haja Andrianasolo S Y P (2001). Remote sensing in unravelling complex associations between physical environment and spatial classes of emerging viral disease. 22nd Asian Conference on Remote Sensing. Singapore, 5-9 November 2001.
Krishna Prasad Bandari P R (2008). Application of GIS modelling for dengue fever prone areas, based on sociocultural and environmental factors – A case study of Delhi City Zone. The International Archive of The Photogrammetry, Remote Sensing and Spatial Information Sciences , XXXVII, 165-170.
Maynard N G (2002). Remote sensing for the public health surveillance and response. Earth Observation Magazine. Special NASA Earth Science Enterprise Issue, pg 43-45. Available online at: www.aarsacrs.org/acrs/proceeding/ACRS2007/Papers/TS27.4.pdf [accessed 28/09/09].
Napier M (2003). Application of GIS and modeling of Dengue Risk Areas in the Hawaiian Islands. Pacific Disaster Center, 590 Lipoo Parkway Suite 259 Kinei, HI. Available online at: http://www.pdc.org/iweb/pdc_archives.jsp [accessed 28/09/09].
Openshaw S (1996). Geographical information systems and tropical diseases. Transactions Of The Royal Society Of Tropical Medicine And Hygiene, 90, 337-339.
Ratana Sithiprasasna K J (1997). Use of GIS to study the epidemiology of Dengue Haemorrhagic Fever in Thailand. Dengue Bulletin, 21, 68-73.
Schroder W (2006). GIS, geostatistics, metadata banking, and tree-based models for data analysis and mapping in environmental monitoring and epidemiology. International Journal of Medical Microbiology, 296, 23-36.
Song G T l (2000). The use of GIS in Ovitrap Monitoring for Dengue control in Singapore. Dengue Bulletin, 24, 110-116.
Tripathi K N (2005). An information value based analysis of physical and climatic factors affecting dengue fever and dengue haemorrhagic fever incidence. International Journal of Health Geographics, 4, 4-13.
Tzai-Hung Wen N H H C D (2006). Spatial mapping of temporal risk characteristics to improve environmental health risk identification: A case study of a dengue epidemic in Taiwan. Science of the Total Environment, 367, 631-640.
Valerie Crossa A F (2000). Fuzzy objects for geographical information systems. Fuzzy Sets and Systems, 113, 19-36.
Wong Ngai Szea L C (2007). An alert system for informing environmental risk of Dengue infectious. GIS for Health and the Environment, Part 7, 171-183, Berlin, Springer.