Training Reports for TWF Investigators

Bretas photo

Gustavo Bretas, M.D., Ph.D.


Institute of Social Medicine
Universidad de Estado do Rio de Janeiro

Project and Training Dates

Malaria in Rondonia, Brazil
August - September 1997

Planned Objectives for the Visit

  1. Use of satellite imagery and maps to build the baseline data for the study and control of malaria in Porto Velho, Rondonia.
  2. Acquisition of available data, derived from AVHRR, and forest fire information to be used for modeling malaria risk throughout the Amazon.
  3. Classification of vegetation and land use patterns in Candeias, Rondonia, to be correlated with malaria incidence.
  4. Acquisition of books and software for image processing and spatial analysis.
  5. Contacts with other researchers using satellite imagery to study diseases and/or environmental aspects of Amazon.


Four sheets of maps at 1:100k scale from the area of Porto Velho, were used to digitize the relief, hydrography, street and road network. A Landsat image was processed and the following two main thematic maps produced: one for land use and another for types of water. Rivers in the Amazon usually have three main different types of water: black (tea like, acid water), brown (muddy with sediment), and white (clear). Anopheles darlingi, the main vector in the area, is associated with clear, relatively large and still water. These areas were marked in the image and vector studies will be carried in the area to verify if it is possible to find the main breeding sites around urban areas using imagery. The cases of malaria in the urban and periurban areas will be plotted by their address over the street network and the spatial relationship with the breeding sites studied.

Four grids were built to model the relationship between deforestation, forest fires, and malaria. Using a half-degree pixel, data on the number of fires (AVHRR) for 1994, NDVI in for the previous year, and rate of malaria (measured at municipality level) for 1995 were analyzed. The assumption was that the areas with high NDVI and a large number of forest fires would be the ones where the rate of malaria would be highest. A relationship could not be shown within this dataset at the scale used. The plan is to work with a time-series and to use a more precise gridding for the malaria data and also to incorporate data on CO2 flux and methane emissions to the model.

Dr. Bretas reclassified SPOT and Landsat TM images of the study area that he used in his Ph.D. work. With a program developed at CHAART, the percentages of different landscapes in buffers around each house were extracted. They will be used in the modelling of the relationship of malaria and the environment in that area.

Images from 1994 and 1992 were used to create a thematic image of the changes in the area, deforestation, and regrowth of deforested areas. The amount of deforestation around houses was measured using ArcInfo AMLs and a C program developed at CHAART. The data will be used as above.

Future Activities

  1. Finish the analysis of the Ph.D. data.
  2. Work with the prediction model throughout the Amazon.
  3. Study the Porto Velho area.
  4. Work with health impacts of large engineering work.
  5. Finish an article with the thesis data.
  6. Write a booklet on Geomatics and Health.

Last updated: February 2002