Paper/Poster/Presentation Title
A Computer System for Forecasting Malaria Epidemic Risk Using Remotely- Sensed Environmental Data
Keywords
malaria, early warning systems, remote sensing, geoinformatics
Start Date
1-7-2012 12:00 AM
Abstract
Epidemic malaria is a major public health problem in the highlands of East Africa. Identifying the climatic triggers that increase malaria risk affords a basis for developing environmentally-driven early warning systems. Satellite remote sensing provides a wide range of environmental metrics that are sensitive to temperature, rainfall, and other climatic variables. The goals of this study were to develop a computer application for automatically acquiring and processing remote sensing data, and to test the utility of these data for modelling and forecasting malaria epidemics in the Amhara region of Ethiopia. The application was programmed using JAVA for user interface development and overall system control. Spatial analyses were carried out using Python scripts to call ArcGIS geoprocessing functions, and PostgreSQL was used to store and manipulate the resulting data summaries. Remotely-sensed variables included land surface temperature from MODIS/Terra, vegetation indices computed using MODIS nadir BRDF-adjusted reflectance, precipitation estimates from the Tropical Rainfall Measuring Mission, and actual evapotranspiration modelled using the simplified surface energy balance method. Historical remote sensing data from 2000-2010 were summarized at the district level by 8-day MODIS composite periods and transformed to deviations from their 11-year means. Time series of monthly malaria outpatient cases were collected for 19 districts in the Amhara region and used to compute risk indices for the main epidemic season from September- December. Malaria epidemics during this season were associated with a higherthan- normal number of malaria cases in May-June, higher-than normal rainfall in January-May, and warmer-than-normal temperatures in May-June. A crossvalidated statistical model containing these variables predicted more than 50% of the variability in malaria relative risk. Continued environmental monitoring using satellite remote sensing will allow us to forecast the environmental risk of malaria epidemics in future years and validate these initial results.
A Computer System for Forecasting Malaria Epidemic Risk Using Remotely- Sensed Environmental Data
Epidemic malaria is a major public health problem in the highlands of East Africa. Identifying the climatic triggers that increase malaria risk affords a basis for developing environmentally-driven early warning systems. Satellite remote sensing provides a wide range of environmental metrics that are sensitive to temperature, rainfall, and other climatic variables. The goals of this study were to develop a computer application for automatically acquiring and processing remote sensing data, and to test the utility of these data for modelling and forecasting malaria epidemics in the Amhara region of Ethiopia. The application was programmed using JAVA for user interface development and overall system control. Spatial analyses were carried out using Python scripts to call ArcGIS geoprocessing functions, and PostgreSQL was used to store and manipulate the resulting data summaries. Remotely-sensed variables included land surface temperature from MODIS/Terra, vegetation indices computed using MODIS nadir BRDF-adjusted reflectance, precipitation estimates from the Tropical Rainfall Measuring Mission, and actual evapotranspiration modelled using the simplified surface energy balance method. Historical remote sensing data from 2000-2010 were summarized at the district level by 8-day MODIS composite periods and transformed to deviations from their 11-year means. Time series of monthly malaria outpatient cases were collected for 19 districts in the Amhara region and used to compute risk indices for the main epidemic season from September- December. Malaria epidemics during this season were associated with a higherthan- normal number of malaria cases in May-June, higher-than normal rainfall in January-May, and warmer-than-normal temperatures in May-June. A crossvalidated statistical model containing these variables predicted more than 50% of the variability in malaria relative risk. Continued environmental monitoring using satellite remote sensing will allow us to forecast the environmental risk of malaria epidemics in future years and validate these initial results.
