Introduction
Remote Sensing has brought so many advances to many different scientific disciplines, including the study of epidemiology. For as long as life has been on the Earth, disease has always been factor in the success of any species. Pandemics throughout the history of time have devastated the populations of both animals and humans. This paper looks to explore the use of remote sensing in the fight against the spread of disease in human populations that threaten our way of life on a daily basis.
Main Body
Scientists are adopting new remote sensing (RS) techniques to study a variety of vector borne diseases. Environmental data collected from satellites such as temperature, sea surface temperatures, land cover, vegetation density and vector density are being used to characterize vector habitats. The use of RS to map vector borne diseases has ‘evolved significantly’ over the past 25 years (Kalluri, 2007). Sadly, until the 20th century vector borne diseases were responsible for more human deaths that all other causes (Kalluri, 2007). Interesting much of the focus in the study of disease is on the natural cycles of life, but there are many other factors that influence the spread of diseases, that include floods and other natural disasters. RS may be able to offer more answers to health scientists and epidemiologists in the field of diseases caused by disasters aside from the research that is already being conducted.
The beginnings of spatial analysis occurred more than 100 years prior to any satellite being launched. In September of 1854 a cholera outbreak occurred in the SoHo district of London. At that time a medical doctor, John Snow, was fairly convinced the disease was caused by a water borne pathogen, even though his theory went against popular belief. To further his theory, Dr. Snow created a map of the area that included the location of public water pumps as well was the concentrations of cholera fatalities. There were 521 cholera deaths within 250 yards of the Broad Street pump. The map confirmed Dr. Snow’s theory that the Broad Street pump was the culprit of the outbreak. In response, the handle of the pump was broken by skeptical local authorities and the outbreak quickly subsided. (David A. Vaccari, 2006)
Although this example is not a study in RS, it does demonstrate the importance of analyzing data spatially in the fight against disease. In an article published in the National Institute of Health’s Library of Medicine, the techniques of Remote Sensing and Epidemiology ‘have the potential to revolutionize the discipline of epidemiology and its application in human health’. The article explains in great detail the definition of remote sensing, electromagnetic radiation, atmospheric transmission, spectral response, active and passive remote sensing, as well as many other parameters of remote sensing. But mostly the author also explains how the applications of remote sensing can be applied to the study of disease. Such as the use of spectral vegetation indices with reference to chlorophyll and carotenoid pigments, land surface temperature indices and atmospheric moisture indices (Hay, 2000). Interestingly the article, which was published in 2011, seemed to infer that the combination of remote sensing in the field of epidemiology is a potential application, even though studies have been published as early as 1970. Others share Hay’s enthusiasm as seen in an article in the American Journal of Epidemiology where the author describes the use of aerial photography and other remote sensing techniques as ‘New Eyes for Epidemiologists’ (Cline, 1970).
Although the use of RS appears to be a useful tool in the study of human health, the study is not without it’s critics, as explained in a 2007 Health & Place journal article. The authors explain that much of the research conducted invloves the use of pre-processed spatial data and low cost images, which limits the adaptability of the data for biological uses. Other criticicms included extrapolating local data to regional areas and the lack of supported field work. The problems where are further exacerbated by the lack of acknowledgemet of these limitations in many of the papers cited in the article.
Though the criticisms of the use of remote sensing and epidemiology may be justified, the research that has been conducted would certainly have value. This has been demonstrated in several studies that use the data from environmentally remote sensed images in the study of disease. This is demonstrated in a study involving the creation of software used to combine RS data with earth science data profiled in the journal ‘Environmental Modelling & Software’ (Liu Y, 2014). The software titled ‘EASTWeb’ is an open source application, that requires client input that automatically connects to earth science data archives and processes and summarizes remote sensing data sets. The user inputs a time and a place which is in turn integrated with the data sets. The goal of the software is to predict high risk factors and for the support of early warning systems specifically for malaria and West Nile outbreaks in many parts of the world, including the United States (West Nile) and Ethiopia (malaria). The result is a ‘high level architecture’ program.
The RS components that are implemented for the software include MODIS, TRMM and ETo data, which are used to create environmental indices that are used to calculate a variety of metrics. Based on the results, a statistical model is developed for use to forecast and respond to threats to human health by the specific vector borne diseases mentioned above.
Limitations to the software, according to the authors of the article, are the large volumes of data that are necessary and need to be stored and processed on local computer systems. The software is also highly specialized to respond only to threats specifically from malaria and West Nile. Other software programs used in epidemiology are more generalized, which make those programs more adaptable. Even so, there is value in specialization since diseases and factors affecting their spread can be varying. The future of the software may include integrating other forecasts, such as epidemiological and entomological data, which should increase their effectiveness.
Another example of the RS and human health research is being done is by the European Space Agency, Copernicus. The space agency was working to minimize the exposure and increase the response to the outbreaks of Ebola that occurred in 2014. Parameters such as water bodies, wind or dust and land cover were and continue to be monitored to aid in the prevention and timely response to outbreaks in areas prone to epidemics. The sentinel-2 is a mission that has been used to achieve these directives. The mission comprises of twin polar-orbiting satellites in the same orbit, phased at 180° to each other. It has a wide swath width and a high revisit time of either 10 days at the equator with one satellite or five days with two satellites. The mission monitors changes to vegetation in the growing season. The coverage limits are from latitudes 56° south and 84° north, which would include the study area of Kwendin, Liberia. An article in the ‘Copernicus’ journal outlines a particular study concerning the fruit bat, which is a vector of Ebola (Tracing the Outbreak of Ebola, 2014). The following image details the land cover, such as roads and settlements within areas of dense tropical forests and oil palm cultivations that support fruit bat populations.
When viewing the aerial image and highlighting the proximity of the environmental factors to populations, high risk areas are easily identified. A similar study was conducted in Tanzania and the outbreak of disease involving the nematode, Ascaris lumbricoides. This infectious agent is mainly prevalent in the tropics and affects one-quarter of the world’s population. It can cause respiratory and gastrointestinal problems.
The remote sensing data for this was obtained from ‘The Shuttle Radar Topography Mission’ (Farr, 2007), which employed two synthetic aperature radars with a C band (5.6 cm) radar and a X band (3.1 cm) radar in the shuttle payload bay, as well as secondary C and X band attenas in an 60 meter mast, used to obtain digital elevation models.
The elevation data obtained from the Shuttle Radar Topography mission was used to calculate slope, which contributed to climatic conditions that contributed to the prevalence of disease cause by the nematode (Schule, 2014). Conclusions from the study attributed the main environmental factors in the spread of the disease to rainfall, land surface day temperatures and vegetation density (positive correlation).
Human health is not only affected by vector borne diseases, but often by the methods used to eliminate the vectors. This is demonstrated in a study used to link pesticides to human health (VoPham T., 2015). The study used Landsat images captured since 1972, and compared them to rarely updated land use crop data to reconstruct past exposure to specific pesticides. The ultimate goal was to estimate pesticide exposure based on crop fields near residences. Supervised classification of was implemented to classify various types of crops that would satisfy the four pieces of information necessary to the study; Landsat images, ground truth, pesticide application data and geocoded locations. The land use classes were identified by a sample of 1990 NDVI signatures and patterns.
The particular usefulness of using RS data as compared to GIS data is based on informational input involving the changes occurring in planted crops that may not be reported. Changes in landscape features such as crop rotations and land use conversions would be captured by RS sensors such as Landsat. The data captured would ultimately have to be validated in an Accuracy Assessment, as compared to Land Use surveys.
The Center for Disease Control (CDC) also recognizes the importance of RS in the field of epidemiology. An article published in the ‘Emerging Infectious Disease’ Journal highlights many studies that are being conducted, such as cholera in Bangladesh and Lyme disease the northeastern United States. Interestingly, the Lyme disease study used local, municipal data from veterinarians to measure the risk to human health, based on the assumption dogs are more likely to be bitten by ticks near their owner’s property. The Landsat thematic mapper measured and derived indices of vegetation greenness and wetness. The deer population, in relation to forest size and density was also a factor and was recorded. Ultimately it was determined that greenness and wetness were positively correlated with tick abundance (Beck, 2000).
The second study highlighted in the article concerns outbreaks or cholera in Bangladesh. Similar to another study discussed in class, this research correlated sea surface temperature (SST) to phytoplankton. Which concludes as SST temperatures increase, phytoplankton populations increase and cholera outbreaks increase. Positively the CDC looks to the future of RS by stating; “In the next 15 years, new sensors will provide valuable data for studies of infectious diseases similar to the ones described here. For Lyme disease, new sensors could provide similar information about ecotones, human settlement patterns, or forests. These sensors include ARIES-1, scheduled for launch by Australia; CCD and IR/MSS sensors onboard CBERS, launched by China and Brazil in late 1999; Ikonos, a commercial satellite with 4-m spatial resolution; LISS III, onboard the orbiting Indian IRS-1C and -1D satellites; and ASTER, onboard the recently launched Terra satellite. Information from these sensors could also be used to address other vector-borne diseases, such as malaria, schistosomiasis, trypanosomiasis, and hantavirus, whose patterns are likewise influenced by environmental variables.’ (Beck, 2000).
Much of the focus in the study of RS and epidemiology involves precipitation and vegetation density. Recently health scientists have made the connection between meningitis and dust. This is demonstrated in a study by The International Research Institute for Climate and Society in Sahel, Africa (Cuevas, 2011). The mineral dust may enhance meningococcal by damaging the mucous barrier in the epithelial lining in the upper respiratory tract aiding bacterial penetration. A more controversial theory suggests that the dust particles may actually be a carrier for the bacteria. The NMMB/BSC dust model was used in the study.
Remote sensing techniques were aided by providing total column dust observations, although it cannot be assumed that dust near the ground actually affect the population. Even so, the focus of the study was primarily on dust, which did correlate to an increased incident of meningitis.
It is worth reiterating that although the future of RS is promising, limitations of the data collected should be acknowledged and minimized as profiled in an article in Environmental Health Perspectives (Seltenrich, 2014). The author acknowledges the great promise in satellite data in projecting the outbreaks of infectious disease, but cautions that ‘satellite imagery does not provide a complete or perfectly reliable picture of what’s happening on the surface and should be verified by and blended with ground-sensed data’.
Conclusion
The marriage between RS and Epidemiology has only begun to hit its stride. While the use of RS has been used since the 1970s, there are still many limitations, primarily based on a lack of focus specifically relating to the study of disease and human health. Programs and software that are currently utilized were not designed for epidemiology, which may present challenges. Fortunately, many of the vector borne diseases have a strong correlation with environmental conditions, which is a field of study that is very compatible with RS and by extension epidemiology. It is worth noting that none of the studies were able to rely upon RS exclusively. Even though RS has been a useful tool in the study of human health, all of the studies required the collection of other data from other disciplines to be useful. Whether it was a confirmation from ground truth data, health samples or municipal data, no one has been able to look at an image created from a sensor and be able to ascertain any kind of risk or response to human or animal health.
Ultimately, for RS to be used in the mainstream study of epidemiology, many factors need to be improved. Factors such as higher spatial and spectral resolutions, increased satellite coverage, cost management and increased availability of data from new sensors should bring RS to the forefront in the surveillance and management of disease control.