Abstract
For elucidating process and responding to red-soil runoff from sugarcane field, it is required to detect the harvest date precisely at each land parcel. Because satellite remote sensing provides very limited frequency of observation, it is not appropriate for this case. In this study, we observed sugarcane field continuously by using a digital time-lapse camera, and developed a new method to detect the harvest dates automatically. Color information of red, green, and blue (RGB) digital counts were derived from the images that were taken 3 times a day from January to May. By statistical analysis of RGB values, images taken under bad weather were excluded, and a Green Ratio index (GR) was calculated for each pixel. The values of GR reflected the greenness of plant leaves, and changed in relation to the increasing and decreasing of plants. The maximum ratio of decreasing in the temporal variation of GR indicated the date of harvest, and the pixel based analysis enabled to detect spatial distribution of the dates of harvest at each land parcel. Thus image analysis under high temporal and spatial resolution by use of digital camera is cost-effective and applicable to all over the place for monitoring various ecosystems and fields.
