Initial pilot study results suggest that climatology maybe used as an early warning system of Visceral leishmaniasis in South Sudan.Alexandra Sweeney1, Andrew Kruczkiewicz1, Caitlin Reid1, Jill Seaman5, Abdinasir Abubakar3, Koert Ritmeijer4, Constantino Doggale6, Katherine Jensen7, Ronny Schroeder7, Kyle C. McDonald7, Madeleine Thomson1, Dia Elnaiem2, Pietro Ceccato1
1International Research Institute for Climate and Society, The Earth Institute, Columbia University, Palisades, New York, U.S
2Department of Natural Sciences, University of Maryland Eastern Shore, U.S.
3Communicable Disease Surveillance and Response, World Health Organization, South Sudan Office, Juba, South Sudan
4Public Health Department, Médecins Sans Frontières (MSF-Holland), Amsterdam, The Netherlands
5Alaska Sudan Medical Project, Anchorage, U.S.
6Ministry of Health, Republic of South Sudan, Juba
7Department of Earth and Atmospheric Science, The City College of The City University of New York, U.S.
Leishmaniases are vector-borne diseases caused by the obligate intracellular parasitic protozoa of the genus Leishmaniasis (Leishman, 1903). The genus Leishmania consists of more than 20 described species that can cause human infection (Rougeron et al., 2011). Transmission occurs when the Leishmania are injected into the skin during the blood meal of the infected female sand fly vector of the genera Lutzomyia in the New World and Phlebotomus in the old world, including the Northern States of the Republic of South Sudan (Desjeux, 2004).
The epidemiology of Leishmaniasis is complex and dependent upon a multitude of entomological, geophysical, and socioeconomic variables (Murray et al., 2005). Some of these variables include the type of parasite species, the local environmental and climatic characteristics of the transmission sites, past exposure of the human population to the parasite, and human behavior (World Health Organization (WHO), 2013). As these contingencies are influenced by environmental and climatic variables, a causal relationship between these exogenous drivers and outbreaks of Visceral leishmaniasis (VL) has been hypothesized (Quate, 1961; Ashford and Thomson, 1991; Elnaiem et al., 1998).
There are three main clinical forms of Leishmaniasis: visceral, cutaneous, and mucocutaneous, together accounting for an estimated 2,357,000 new cases per year globally (WHO, 2013). VL, also known as kala azar in east Africa and India, is the most serious form of the disease and is caused by a parasite multiplication in the reticulo-endothelial tissues of the spleen, liver and bone marrow (El-Safi et al., 2006). Symptoms of VL may manifest in the form of recurring fevers, anemia, rapid weight loss and/or enlargements of the spleen and lymph nodes (Seaman, Mercer, and Sondorp, 1996; Kolaczinski et al., 2008). With a near-100 percent mortality rate if left untreated, VL presents an important health challenge in many countries, including South Sudan (El-Safi et al., 2006).
The World Health Organization (WHO) lists VL as a “Neglected Tropical Disease” though it is responsible for high mortality worldwide (WHO, 2012). Alvar et al. (2012) summarizes the reasons VL is often forgotten: “This consignment to critical oblivion results from its complex epidemiology and ecology, the lack of simple, easily-applied tools for case management and the paucity of current incidence data, often results in a failure on the part of policy-makers to recognize its importance.”
Although treatment is available, there is no vaccine to prevent VL and the emergence of symptoms varies depending on an individual’s immune response (MSF, 2013). This causes the incubation period to range from two to six months (Gage et al., 2008; Seaman et al., 1996), which leads to difficulty in associating the diseases’ transmission with specific environmental factors, although attempts have been made (Elnaiem et al., 2003).
Further complicating the issue, VL transmission cycles vary according to their reservoir host type, which is specific to certain geographical regions; anthroponotic (human hosts) in the Indian Sub-continent, zoonotic in Brazil, and zoonotic/anthroponotic (animal/human hosts) in East Africa (Tesh, 1995; Ashford, 2000; WHO, 2010). In East Africa, even though strong evidence exists for zoonotic cycles, the specific animal reservoirs are largely unknown (Elnaiem, 2011).
Disease vectors are known to exhibit acute sensitivities to climatic and environmental factors due to their small size and high ratio of exposed surface area, which makes it plausible that the parasite they carry will likely exhibit similar sensitivities to their environment (e.g., mosquito vector and malaria) (Lindsay et al., 1998). A similar relationship has been implicated for VL, but determining the causality of environmental and climatic variables on the sand fly and subsequently on VL is particularly difficult as multiple parts of the transmission cycle (reservoir – vector and human) are affected differently.
Furthermore, while the intensity of VL transmission is known to vary significantly within East Africa on small geospatial scales, it is unknown what drives this variability (Elnaiem, 2003). Before Earth-observing satellite measurements were made accessible, understanding the drivers of the distribution gradient of vector-borne diseases in a context that captures fluctuations of environmental factors was limited by the paucity of in situ data (Hay et al., 1996). In the last three decades, considerable advances have been made in this regard.
This study focuses on three states in northern South Sudan: Jonglei, Unity, and Western Upper Nile. Our goal is to describe the current context and initial findings of a pilot study investigating the relationship between VL and climatic and environmental drivers of transmission observed by health providers in these states in South Sudan. We seek to demonstrate how new advancements in remote sensing of Earth observations have afforded the opportunity to explore the distribution and glean the relationship between vector-borne diseases and environmental and climatic factors on a meaningful spatial scale (Thomson et al., 1997).
Endemic regions of VL exist within East Africa with a geographic hotspot in the northern states of South Sudan (Seaman et al., 1996). This region is known to experience seasonal fluctuations in cases that typically peak during the months of September, October, November, December, and January (SONDJ) (WHO, 2013; Gerstl et al., 2006; Seaman et al., 1996). In the northern states of South Sudan alone, VL epidemics have recently been observed with a reported 28,512 new cases from January 2009 to December 2012 (Ministry of Health – Republic of South Sudan, 2013). Without proper treatment, mortality in South Sudan is high, with numbers approaching 100,000 during one multi-year epidemic in the late 1980s and early 1990s (Seaman et al., 1996). In the northern states, the sand fly responsible for transmission of VL is Phlebotomus orientalis (Quate, 1964; Hoogstraal & Heyneman, 1969; Schorscher & Goris, 1992; Seaman et al., 1996; Kolaczinski, 2008).
The habitat of P. orientalis is determined by specific ecological conditions including the presence of black cotton soils, or vertisols, the presence of Acacia-Balanites woodlands, and mean maximum daily temperature (Thomson et al., 1999; Elnaiem et al., 1999; Elnaiem, 2011). Research has also documented associations between environmental factors and VL that may contribute to outbreaks of the disease including: relative humidity (Salomon et al., 2012; Elnaiem, Hassan, and Ward, 1997), precipitation (Gebre-Michael et al., 2004; Hoogstraal and Heyneman, 1969), and Normalized Difference Vegetation Index (NDVI) (Elnaiem et al., 2003; Gebre-Michael et al., 2004; Hoogstraal and Heyneman, 1969; Elnaiem et al., 1997; Elnaiem et al., 2003; Rajesh and Sanjay, 2013). Additionally, Ashford and Thomson (1991) suggested the possibility of a connection between a prolonged inundation (flooding) event in the 1960s and the corresponding 10-year drop in VL within the northern states of South Sudan. Although the importance of environmental variables in relation to the transmission dynamics of VL have been established, the lack of in situ data within the study region has led to inconclusive results regarding these relationships. By exploiting the advantages of sustained and controlled Earth monitoring via NASA’s Earth observations, this project explores the relationship between environmental factors and the spatiotemporal distribution of VL in the northern states.
Environmental Data Acquisition
Datasets from Earth-observing satellites were retrieved from the International Research Institute’s (IRI) Data Library (DL). The DL is a repository that catalogues climatic and environmental raw data for scientific research (Del Corral et al., 2012). Further, the Ingrid Data Analysis Language was used to construct specialized spatial and temporal analyses of these various products in the DL Expert Mode (Ceccato et al., 2006).
Nighttime and daytime land surface temperature (LST) time series were obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard the Aqua and Terra satellites from 2004-2012. Additionally, the Normalized Difference Vegetation Index (NDVI) from MODIS onboard Terra and precipitation from the Tropical Rainfall Measuring Mission (TRMM) were extracted for this same time period. Surface inundation datasets developed at The City College of New York (CCNY), as part of a NASA Making Earth System Data Records for Use in Research Environments (MEaSUREs) project supporting assembly of global scale Earth System Data Record (ESDR) and characterizing spatio-temporal attributes of inundated wetlands, developed new products to examine inundation dynamics (McDonald et al., 2011). We employed the Surface Water Microwave Product Series (SWAMPS) component of that ESDR derived from the QuikSCAT and AMSR-E active/passive microwave datasets to examine the period 2004-2009 for the South Sudan (Schroeder et al, 2010 and 2014). These three environmental variables (NDVI, precipitation, and inundation) aided in determining whether wet or dry years were more conducive to the transmission of Leishmaniasis. Lastly, relative humidity data was obtained from the NOAA NCEP dataset at the 1000 millibar pressure level from 2004-2012.
Landsat 8 images were downloaded from the U.S. Geological Survey (USGS) to validate the inundation fraction anomalies. Soil property maps were acquired from the Center for International Earth Science Information Network (CIESIN) and developed through the CIESIN African Soil Profiles Database. These maps showed the spatial distribution of seven different soil properties: bulk density, cation exchange capacity, clay content, soil organic carbon, pH (acidity), silt content, and sand content. These soil property maps were obtained from the International Soil Reference Information Center (ISRIC) at six different depths (0-5 centimeters (cm), 5-15 cm, 15-30 cm, 30-60 cm, 60-100 cm, 100-200 cm) for three predicted values (mean, lower limit, upper limit) (ISRIC, 2013). Use of these maps affords us the potential to expand on previous work that aimed to describe the relationship between soils and VL distribution, but lacked the geospatial and temporal resolution to do so (Gebre-Michael et al., 2004; Elnaiem et al., 2003; Thomson et al., 1999).
Epidemiological Data Acquisition
Epidemiological data from 29 medical centers were provided by their operators and compiled at the WHO office in Juba, South Sudan, with assistance from Jane Pita. Datasets were transferred across two temporal ranges: 2004-2009 and 2010-2013. Sixteen of 29 medical centers were found to have continuous reporting across both ranges and were selected for further analysis (Figure 1). Each medical center reported the number of confirmed cases of VL at monthly intervals. Cases of VL were reported throughout the year, however plotting of monthly values indicated that epidemics most commonly occur during the months of September through January. Considering a known two- to six-month incubation period, these patients would likely have been infected during the months of April, May, June, or July (Seaman, 1996).
It should be noted that there are additional challenges that are not directly revealed by the data. Uncertainties regarding population base, geographic origin of the patient and fluctuations in cases due to socioeconomic factors are unique to each individual center and are difficult to quantify. For example, it has been observed that the medical center in Old Fangak routinely receives patients from Keew and Atar, a travelling distance of 20 and 40 kilometers, respectively, therefore affording further deliberation regarding the area to average the climatic and environmental variables over.
Time Series Analysis of Environmental Factors and VL
Precipitation, inundation (flooding), humidity, NDVI, and temperature were geospatially averaged over a polygon approximately 275 by 385 kilometers (170 by 240 miles). It should be noted that even though satellite data allow for analysis of the variables at higher geospatial resolution, confounding socioeconomic factors such as nomadic movement of the population, variable service radii of medical centers, and lack of prolonged periods of political stability might limit the accurate capture of transmission dynamics. Given these challenges, analysis at a larger geographic scale would likely capture events that are less likely to be influenced by local factors.
To explore the relationship between environmental factors and VL, time series analyses were carried out in three ways. First, the seasonal anomaly of each environmental variable over the entire transmission period of April, May, June, and July (AMJJ) was compared to the sum of cases from the following September-January (SONDJ) documented within the polygon. Doing so allowed us to explore the general relationship between environmental factors during the transmission season and resulting cases of VL.
Second, the individual monthly anomalies during AMJJ of each environmental variable were computed and compared to the sum of cases in the following SONDJ documented within the polygon. These analyses allowed us to investigate the influence of environmental variables during specific months of the transmission season and resulting cases of VL.
Third, the individual monthly anomalies of each environmental variable over the entire transmission period, AMJJ, were compared to the sum of cases reported at each medical center in the following SONDJ period. This analysis allowed us to examine the influence of the environmental variables on the distribution of the number of cases between medical centers.
Analysis of Geospatial Distribution of Inundation
To explore the relationship between inundation and VL outbreaks, the total and individual monthly anomalies during AMJJ were compared to the sum of cases in SONDJ of the same year. Years where both negative (positive) inundation anomalies (greater than +/- 0.1) in AMJJ and a high (low) amount of cases reported at centers during SONDJ were flagged for analysis. The team downloaded Landsat 8 images from the months of highest and lowest inundation during the transmission season (AMJJ) of those flagged years. Photographic interpretation was used to compare the difference in location, size, turbidity and geospatial extent of inundation between months exhibiting the highest and lowest values (Frazier and Page, 2000). Unfortunately, due to persistent cloud cover over the short temporal scale, we were not able to affirmatively identify any changes in water bodies.
To explore the relationship between soil type and VL outbreaks, high-resolution (1-kilometer) static soil maps were overlaid on a map of the medical centers. This process aided in visualizing the relationship between the geospatial distribution of soil characteristics and cases of VL.
A photographic interpretation analysis was designed to indicate potential relationships and establishing a conduit for future analyses (Dent and Young, 1981; Mulder et al., 2011). It is important to note that no geospatial statistical correlation analyses were carried out in this study and thus the relationship between soils and VL cannot be quantified at a significant level.
Not all environmental variables exhibited a relationship with VL. Of those that did, some exhibited a stronger relationship with VL than others. The strongest relationships found were of precipitation and inundation. Furthermore, our analysis indicated that the influence of climatic variables varies from month-to-month during the transmission season. The limitations of a statistical analysis apply in that relationships identified may not be causal in nature (Holland, 1986).
In addition, it was found that below-average precipitation during particular months of the transmission season were better indicators of lagged reports of VL than others (Figure 3). During the VL epidemics of 2009, 2010, and 2011, the month of June exhibited below average precipitation. Furthermore, the two largest epidemics (2010 and 2011) were associated with years of below average precipitation in the month of April.
Inundation during AMJJ also exhibited a strong inverse relationship with VL cases in SONDJ. This relationship was best explored when comparing the VL case data of a specific medical center to the inundation anomalies. Results are typified by the Lankien Medical Center analysis where below average inundation during April displays an inverse relationship with VL cases in the following SONDJ (Figure 4).
Drought may lead to below average inundation, which could allow for soils to maintain their fissures resulting in a sustained breeding season for the sandflies (Quate, 1964). Above-average precipitation and inundation might have the inverse effect, thus eliminating their breeding sites within the soil.
LST Night, LST Day, and relative humidity did not show a particularly strong relationship with VL. Further research is needed, as these variables are known to exist across strong gradients within the northern states (Quate, 1964).
The analysis of the geospatial distribution of soil showed a possible connection between VL distribution and soils of higher clay content, but more research is necessary to reach significant conclusions. The integration of epidemiological data, remote sensing, Geographic Information Systems (GIS), and spatial statistics is paramount if we are to deem a geospatial pattern to be the result of a significant quantitative relationship rather than just random noise (Moore et al., 1999).
This pilot project has enabled the collation of significant epidemiological and environmental data and established a partnership between regional actors including the WHO office of South Sudan in Juba, Médecins sans Frontières, Columbia University, University of Maryland Eastern Shore, and The Alaska Sudan Medical Project. As a first step, we have undertaken an exploratory analysis of climate and environmental variables with VL cases obtained from facilities and aggregated to the lowest geospatial level where significance can be maintained. Initial results indicate that below average precipitation in the VL transmission period (AMJJ) may correspond with higher transmission. Below-normal inundation during this transmission period exhibited a similar association with VL cases. Particular months seem to have a stronger relationship with heightened cases of VL and should be investigated further in future studies.
This project was designed to address issues that had limited previous attempts to quantify the relationship between climatic and environmental factors with outbreaks of VL. While previous research relied on limited data from the few in situ locations within South Sudan; the use of recent advances in remote sensing in this project has enabled development of datasets that are geospatially and temporally significant. Integration of these datasets may allow for us to quantify the relationship between climatic and environmental factors with outbreaks of VL.
We have presented an analysis of how the environment and climate of three states in South Sudan may affect VL, but we acknowledge that further investigation is needed. Population vulnerability, which to a large extent is determined by immunity, health and nutrition status, and recent exposure to infection, complicates the transmission dynamics within the northern states and should be considered. Investigation into the breeding and resting sites of P. orientalis may aid in predicting outbreaks of VL (Elnaiem, 2011). Additionally, understanding the VL transmission cycle via reservoir hosts within the northern states of South Sudan is imperative to predicting epidemics and delivering necessary preventative materials and/or treatment. Research has shown that there is a relationship between Acacia seyal and Balanites aegyptica woodlands with increased reported cases of VL. This relationship should be analyzed with modified vegetation indices, but would ideally need to be performed at high-spatial resolution as even small patches of these woodlands might be responsible for some risk of transmission.
With the incorporation of newly developed soil maps, an investigation into the relationship between vector distribution and soil type will be possible at a high geospatial scale (1 kilometers) for the first time (ISRIC, 2013). Using these maps affords us the potential to expand on previous work that aimed to describe the aforementioned relationship (Gebre-Michael et al., 2004; Elnaiem et al., 2003; Thomson et al., 1999). Moving forward, a robust analysis of the relationship between soil type and both sand fly presence and VL case distribution should be conducted using a method similar to Nagendra (2010), but at higher geospatial and temporal scales.
An additional factor to consider when analyzing VL epidemiological data is the susceptibility of the population to develop the clinical disease after infection (Fernández-Guerrero et. al., 1987). This is to a large extent determined by the immune status of the population that in turn is based on exposure to infection in previous years and health and nutritional status of the at-risk population (Ibrahim et al., 2013; Collin et al., 2004).
Examining the climatic parameters during the height of sand fly activity (AMJJ) lends insight into potential transmission levels of VL. Comparing climate variables with epidemiological data indicate that interannual differences in environmental factors contribute to the transmission of the disease. Below-average precipitation and inundation in the transmission season are conducive to seeing more cases of VL in the northern states of South Sudan in the following SONDJ. These novel findings implied that the development of a Leishmaniasis Early Warning System could be developed based on climate variables, much like those produced for malaria (Ceccato & Connor, 2012).
This research was funded by a grant from NASA SERVIR (NNX12AQ70G) and the NASA DEVELOP National Program. Portions of this work were supported by NASA’S Making Earth System Data Records for Use in Research Environments (MEaSUREs) program. The RSS trips of Dr. Elnaiem was supported through a WHO mission organized Dr. Abdinasir Abubakar (WHO office in Juba, RSS), Dr. J. Alvar and Dr. Daniel Argaw Dagne (WHO HQ, Geneva), and Dr. Jose A. Postigo (WHO-EMRO). Other participants in the trip were Jane Pita (WHO, Juba, RSS), Dr. Abate Mulugeta (WHO-Ethiopia), Dr. Caryn Bern (UC-SanFrancisco) and Dr. Israel Cruz (Health Institute “Carlos III”, Madrid).
Ibrahim, M. K., Barnes, J. L., Anstead, G. M., Jimenez, F., Travi, B. L., Peniche, A. G., … & Melby, P. C. (2013). The Malnutrition-Related Increase in Early Visceralization of Leishmania donovani Is Associated with a Reduced Number of Lymph Node Phagocytes and Altered Conduit System Flow. PLoS neglected tropical diseases, 7(8), e2329.
Jones, M. O., Jones, L. A., Kimball, J. S., & McDonald, K. C. (2011). Satellite passive microwave remote sensing for monitoring global land surface phenology. Remote Sensing of Environment, 115(4), 1102-1114.
Kaufman, Y. J., & Gao, B. C. (1992). Remote sensing of water vapor in the near IR from EOS/MODIS. Geoscience and Remote Sensing, IEEE Transactions on, 30(5), 871-884.
Khalil, E. A. G., Zijlstra, E. E., Kager, P. A., & El Hassan, A. M. (2002). Epidemiology and clinical manifestations of Leishmania donovani infection in two villages in an endemic area in eastern Sudan. Tropical Medicine & International Health, 7(1), 35-44.
Kolaczinski, J. H., Hope, A., Ruiz, J. A., Rumunu, J., Richer, M., & Seaman, J. (2008). Kala-azar epidemiology and control, southern Sudan. Emerging infectious diseases, 14(4), 664.
Leishman, W. B. (1903). On the possibility of the occurrence of trypanosomiasis in India. British medical journal, 1(2213), 1252.
Lindsay, S. W., Parson, L., & Thomas, C. J. (1998). Mapping the range and relative abundance of the two principal African malaria vectors, Anopheles gambiae sensu stricto and An. arabiensis, using climate data. Proceedings of the Royal Society of London. Series B: Biological Sciences, 265(1399), 847-854.
McDonald, K.C., Chapman, B., Podest, E. , Schroeder, R., Flores, S., Willacy, K., Moghaddam, M., Whitcomb, J., Hess, L., Kimball, J.S. (2011). Monitoring inundated wetlands ecosystems with satellite microwave remote sensing in support of earth system science research. Conference Paper 34th International Symposium on Remote Sensing of Environment – The GEOSS Era: Towards Operational Environmental Monitoring, 4p
Médecins Sans Frontières (MSF). (2013). Kala azar. Retrieved from http://www.doctorswithoutborders.org/news/issue.cfm?id=2394.
Ministry of Health – Republic of South Sudan (2013). Annual Report of Communicable Disease Surveilence in South Sudan – 2012.
Moore, C. G., & Guptill, S. C. (1999, July). Methods and Data for Vector-Borne and Zoonotic Diseases. In Health Research Methods and Data Workshop, Turku, Finland.
Murray, H. W., Berman, J. D., Davies, C. R., & Saravia, N. G. (2005). Advances in leishmaniasis. The Lancet, 366(9496), 1561-1577.
Osman, O. F., Kager, P. A., & Oskam, L. (2000). Leishmaniasis in the Sudan: a literature review with emphasis on clinical aspects. Tropical Medicine & International Health, 5(8), 553-562.
Patz, J. A., Graczyk, T. K., Geller, N., & Vittor, A. Y. (2000). Effects of environmental change on emerging parasitic diseases. International journal for parasitology, 30(12), 1395-1406.
Peterson, A. T., & Shaw, J. (2003). < i> Lutzomyia vectors for cutaneous leishmaniasis in Southern Brazil: ecological niche models, predicted geographic distributions, and climate change effects. International journal for parasitology, 33(9), 919-931.
Quate, L.W. (1964). Phlebotomus Sandflies of the Paloich Area in the Sudan. Journal of Medical Entomology, 1(3), 213-268.
Rajesh, K. & Sanjay, K. (2013). Change in global Climate and Prevalence of Visceral Leishmaniasis. International Journal of Scientific and Research Publications, 3(1).
Ritmeijer, K., & Davidson, R. N. (2008). Royal Society of Tropical Medicine and Hygiene joint meeting with Médecins Sans Frontières at Manson House, London, 20 March 2003: field research in humanitarian medical programmes. Médecins Sans Frontières interventions against kala-azar in the Sudan, 1989-2003.
Rougeron, V., De Meeûs, T., Hide, M., Le Falher, G., Bucheton, B., Dereure, J., … & Bañuls, A. L. (2011). Multifaceted population structure and reproductive strategy in Leishmania donovani complex in one Sudanese village. PLoS neglected tropical diseases, 5(12), e1448.
Schorscher JA, Goris M. (1992). Incrimination of Phlebotomus (Larroussius) orientalis as a vector of visceral leishmaniasis in western Upper Nile Province, southern Sudan. Trans R Soc Trop Med Hyg. 1992 Nov-Dec;86(6):622-3.
Seaman, J., Mercer, A. J., & Sondorp, E. (1996). The epidemic of visceral leishmaniasis in western Upper Nile, southern Sudan: course and impact from 1984 to 1994. International journal of epidemiology, 25(4), 862-871.
Salomon, O.D.; Quintana, M.G.; Mastrangelo, A.V.; & Fernandez, M.S. (2012). Leishmaniasis and Climate Change – Case Study: Argentina. Journal of Tropical Medicine. doi: 10.1155/2012/601242.
Schroeder, R. M. A. Rawlins, K. C. McDonald, E. Podest, R. Zimmermann and M.
Kueppers(2010). Satellite Microwave Remote Sensing of North Eurasian Inundation Dynamics: Development of Coarse-Resolution Products and Comparison with High-Resolution Synthetic Aperture Radar Data. Environmental Research Letters, special issue on Northern Hemisphere high latitude climate and environmental change, 5 (2010) 015003 (7pp) doi:10.1088/1748-9326/5/1/015003.
Schroeder, R K C McDonald, S K Chan2, B Chapman 2, E Podest 2, T J Bohn, J Kimball, R Zimmermann and M Küppers (2014) Development and Evaluation of a Global Wetland Data Record Derived from Combined Active/Passive Microwave Remote Sensing Data, in preparation.
Tesh, R. B. (1995). Control of zoonotic visceral leishmaniasis: is it time to change strategies?. The American journal of tropical medicine and hygiene,52(3), 287.
Thomson, M. C., Connor, S. J., Milligan, P., & Flasse, S. P. (1997). Mapping malaria risk in Africa: What can satellite data contribute?. Parasitology Today,13(8), 313-318.
Thomson, M.C., Elnaiem, D.A., Ashford, R.W., & Connor, S.J. (1999). Towards a kala azar risk map for Sudan: mapping the potential distribution of Phlebotomus orientalis using digital data of environmental variables. Tropical Medicine and International Health, 4(2), 105-113.
World Health Organization (WHO). (2010). Control of the Leishmaniases. Report of a meeting of the WHO Expert Committee on the Control of Leishmaniases, Geneva, 22-26 March 2010. WHO Technical Report Series 949
World Health Organization (WHO). (2013). Neglected Tropical Diseases: The 17 neglected tropical diseases. Retrieved from http://www.who.int/neglected_diseases/diseases/en/