CHAART Logo

CHAART Training Exercise 1: Processing Steps

Identifying high mosquito-producing rice fields in California

Mosquito








Objectives

The objective of this exercise is to measure the minimum distance from each rice field to a pasture. There are 104 rice fields in this study area, for which there are total seasonal (late May - early September) An. freeborni larval abundance data. The last step in this exercise is to determine the correlation between these mosquito larval counts and the distances to pasture.

Pasture locations will be extracted from the classified image produced during preprocessing, where the remotely sensed spectral information was changed into land cover / land use categories.

Small TM false-color -----> Small TM classification

Steps

This exercise consists of the following five sections or steps:
  1. Isolate the pastures
  2. Compute the distance from the pastures
  3. Extract the average distance
  4. Analyze the distance-to-pasture values
  5. Consider other land covers

The following spatial datasets will be used in this exercise:

  • Geo-registered Landsat Thematic Mapper (TM) data acquired in June 1990
  • Landcover classification produced from the TM data (including rice fields and pastures)
  • Mosquito larval count data

These data may downloaded if you want to do the processing yourself.

Processing Steps

  1. Isolate the pastures from the classified image

    The classified image contains six different land cover / land use classes. To determine the distance from the pasture, the pasture have to be isolated. This is done by recoding the other classes, so only the pasture remains.

    Pasture image

    For this exercise these pasture pixels were then filtered, so only the pastures larger than 2 ha were left. Small blocks of pasture pixels were ignored because they could be vacant lots or other pieces of land that were part of some other land cover. For example, the broken, narrow strip of pasture pixels down the middle of the study area.

    Pasture image, 2 ha

  2. Compute the distance from the pastures

    To determine the distance of each rice field from the pasture, the distance of each pixel from the pasture pixels is needed. This is also known as the proximity. The distance image will then be combined with the rice field image to get the average distance within each field.

    Distance from pasture

  3. Extract the average distance to each rice field

    The unique field values are then used to extract the mean distance from each of the pastures to each of those fields.

  4. Analyze the distance-to-pasture values

    The distance to pasture data can then be imported into a spreadsheet program, with larval count data, to compute the correlation between the two data sets.

    • Distance v. Larval Count: R2 = 0.118
    • Distance v. ln(Count): R2 = 0.344

  5. Consider other land covers

    Examine the land cover classification to consider other landscape factors that may have influenced the observed differences in mosquito production among rice fields but were not used above.

    Small Classification
    Land cover map

    For example, if distance to orchards and distance to pasture are used as independent variables and the ln(Count) is the dependent variable, R2 = 0.371, a very small improvement.

Answers to Study Questions

Congratulations!
You have completed Training Exercise 1.


Last updated: 23 Jul 1999