In a letter published in this issue, Hay and Tatem raise several points regarding the use of remote sensing (RS) in two of our papers that focused on urban malaria in Africa.1,2 We appreciate the points raised by Hay and Tatem and the opportunity provided by the editor of the American Journal of Tropical Medicine and Hygiene to respond to them.
With regard to our first paper,1 it was pointed out that optical sensors should not be the only approach used for water-body discrimination and that synthetic aperture radar (SAR) imagery affords significant advantages. We agree that optical sensors have limitations, particularly due to cloud cover. However, SAR imagery does not solve all the problems. Analysis of a multi-sensor approach3 combining optical (Landsat™, 30 meters) and microwave (Japanese Earth Resource Satellite- JERS-1–18 meters, and European Remote Sensing Satellite ERS-1–26, 1–25 meters) data concluded that 1) SAR did a poor job in separating vegetation classes, but improved the identification of standing and flowing water; 2) urban image classification using only optical data resulted in overestimation of built-up areas; 3) the use of only microwave data for land cover/land use classification was inadequate; and 4) the multi-sensor approach resulted in accurate classification of the urban scene. Current SAR systems (10–40 meters) are too coarse for monitoring urban expansion and identifying potential larval and mosquito habitats, which were our primary concerns. Synthetic aperture radar imagery with a spatial resolution of 1 meter will be available for the first time via the TerraSAR-X radar satellite to be launched in 2006 ( Spot Image http://www.spotimage.fr/html/_167_240_570_684_.php). At that time, it would be desirable to adopt a multi-sensor surveillance strategy for the applications we discussed.1
Additionally, we agree with Hay and Tatem that not all RS in urban areas is, or should be, focused on facilitating identification of larval habitats for operational staff. The RS data with different spatial resolution answer a myriad of questions and needs to be incorporated into strategies for disease control at distinct decision-making levels.4–7 Indeed, the Mapping Malaria Risk in Africa8 initiative is an excellent example of the importance of coarser spatial resolution for district, national, and global level analysis. At the coarser scales, an average risk, influenced by environmental factors, can be appraised. However, local specificities, which have the potential for improving surveillance and control efforts, are overlooked. For malaria control in Dar es Salaam and a diversity of other urban planning situations,9–11 high spatial resolution is necessary. For the landscapes in Dar es Salaam, initial evaluations12 show that a resolution >8 meters would fail to identify multiple important features.
In view of the points raised in the letter by Hay and Tatem and a recent misinterpretation of Hay and others13 regarding our second paper,2 it is important to clarify the following issues. First and foremost, our estimates of malaria incidence among urban dwellers in Africa were not derived from urban surface estimates. We used urban population estimates from the United Nations.14 We then calculated the annual number of malaria cases in urban Africa using the estimated log linear relationship between the entomologic inoculation rate (EIR) and malaria incidence. The EIRs were assigned on the basis of a three-division stratification system for urban environments.
A second point concerns the role of nighttime satellite imagery15 as part of a strategy for estimating the size of urban areas. In this regard, it is important to note that there is no gold standard for even defining what is meant by urban, making cross-country comparisons on the extend of urban areas or estimates of the total urban area difficult.16 Minimal population sizes to qualify an area as urban can be as low as 2,000 (Benin) or 10,000 (Angola). While several Asian cities have tight city boundaries (e.g., Bangkok, Jakarta, or Manila) and are actually much larger than their recorded sizes,16 in Africa a large fraction of the area within administrative city boundaries is not densely populated or built up. Urban agriculture is a common phenomenon. For example, in Kampala, Uganda, 56% of the official city area is devoted to agriculture.17 Thus, estimates of the size of urban areas will depend on which individual or combination of data source is used.
We agree with Hay and Tatem that bright nighttime lights may overestimate city sizes due to blooming. The most common approach to minimize this problem is the thresholding technique.18 Although this method shows good results for the United States,19–21 it is uncertain how big the threshold should be for less developed countries. For example, with the exception of urban areas in South Africa and Zimbabwe, most urban areas in sub-Saharan Africa have electrification levels <30% (e.g., Ethiopia = 12.9%).21 The annual electricity consumption per capita is, on average, only 21 kilowatt hours in Ethiopia compared to 24,248 kilowatt hours in Norway.22 For example, a threshold of 40% resulted in a minimum detectable population of 8,308 in Tanzania and 6,088 in Kenya.23 Considering that the thresholding technique eliminates night lights from small urbanized areas, but also from areas that were affected by black outs or that did not have a large percentage of cloud-free observations,15 we used raw data, rather then the thresholded data. Since many people in sub-Saharan Africa do not have electricity, even in and around the cities, we chose a correction factor of 2–3, which yielded a final estimate of 1.7–2.6% of urban area for Africa. We hope that others will take up the challenge of urban area estimation in Africa and improve upon our methods and estimates.
Castro MC, Yamagata Y, Mtasiwa D, Tanner M, Utzinger J, Keiser J, Singer BH, 2004. Integrated urban malaria control: a case study in Dar es Salaam, Tanzania. Am J Trop Med Hyg 71 (2 Suppl):103–117.
Keiser J, Utzinger J, Castro MC, Smith TA, Tanner M, Singer BH, 2004. Urbanization in sub-Saharan Africa and implication for malaria control. Am J Trop Med Hyg 71 (Suppl):118–127.
Bähr H-P, 2001. Image segmentation for change detection in urban environments. Donnay J-P, Barnsley MJ, Longley PA, eds. Remote Sensing and Urban Analysis. London: Taylor & Francis, 95–114.
Wilson ML, 2002. Emerging and vector-borne diseases: role of high spatial resolution and hyperspectral images in analyses and forecasts. J Geogr Syst 4 :31–42.
Beck LR, Lobitz BM, Wood BL, 2000. Remote sensing and human health: new sensors and new opportunities. Emerg Infect Dis 6 :217–226.
Snow RW, Craig MH, Deichmann U, le Sueur D, 1999. A continental risk map for malaria mortality among African children. Parasitol Today 15 :99–104.
Jensen J, Cowen D, 1999. Remote sensing of urban/suburban infrastructure and socio-economic attributes. Photogramm Eng Remote Sens 65 :611–622.
Jensen JR, 2000. Remote Sensing of the Environment: An Earth Resource Perspective. Upper Saddle River, NJ: Prentice Hall.
Castro MC, 2004. Hidden diversity in an urban/rural dichotomy: a case study for Dar es Salaam, Tanzania. Boston: Annual Meeting of the Population Association of America.
Hay SI, Guerra CA, Tatem AJ, Atkinson PM, Snow RW, 2005. Urbanization, malaria transmission and disease burden in Africa. Nat Rev Microbiol 3 :81–90.
United Nations, 2002. World Urbanization Prospects: The 2001 Revisions. New York: Population Division Department of Economics and Social Affair of the United Nations.
Elvidge CD, Baugh KE, Kihn EA, Kroehl HW, Davis ER, 1997. Mapping city lights with nighttime data from the DMSP operational linescan system. Photogramm Eng Remote Sens 63 :727–734.
Cohen B, 2004. Urban growth in developing countries: a review of current trends and a caution regarding existing forecasts. World Dev 32 :23–51.
Maxwell DG, 1996. Highest and best use? Access to urban land for semi-subsistence food production. Land Use Policy 13 :181–195.
Imhoff ML, Lawrence WT, Stutzer DC, Elvidge CD, 1997. A technique for using composite DMSP/OLS “city lights” satellite data to map urban area. Remote Sens Environ 61 :361–370.
Elvidge CD, Imhoff ML, Baugh KE, Hobson VR, Nelson I, Safran J, Dietz JB, Tuttle BT, 2001. Nighttime lights of the world: 1994–95. ISPRS J Photogramm Remote Sens 56 :81–99.
Sutton PC, 2003. A scale-adjusted measure of “urban sprawl” using nighttime satellite imagery. Remote Sens Environ 86 :353–369.
United Nations Development Program, 2003. Human Development Report. Millennium Development Goals: A Compact Among Nations to End Human Poverty. New York, Oxford University Press.
Sutton PC, Roberts D, Elvidge CD, Baugh KE, 2001. Census from heaven: an estimate of the global human population using night-time satellite imagery. Int J Remote Sens 22 :3061–3076.