Estimation of phenology using different types of data (TARBIL)

Transition times between phenological stages of a crop and the elapsed time of a phenological stage effects several parameters that are used in most of the yield evaluation methods. This study is about using the meteorological data and images from ground stations that are provided by TARBIL (Turkish Agricultural Monitoring and Information Systems) Project's to identify the transition dates between phenological stages of the specific crops. There are more than 300 working stations which are built next to agricultural fields in all over Turkey, and these stations have sensors that collect meteorological parameters that are reported in 10 minutes intervals from each site. Also the image data with different zoom rate is acquired in RGB format at 30 minutes period. In this study, an algorithm designed to observe the phenological course and detect the changes. Each stage has a different course than previous stage and the after one. 5 major stages are considered in this study to detect the transition dates. Two different process are applied and then their outcomes used for determining the transition dates of crops. Firstly, the daily images that are collected from TARBIL stations converted to HSV (Hue, Saturation, Value) color space then the green rate of these images are calculated. In a date base timeline, these calculated values create a green rate/date graph. To increase the correctness of the graph and decrease the noise level of the data, a noise filter is used on the dataset then the graph recreated. In the second step; GDD values are calculated from the meteorological data that collected from TARBIL stations. With respect to these GDD values, chosen crop's estimated phenological calendar is calculated. In the green rate/date graph there are refraction points that associates with a change in the phenological state. According to these refractions a new phenological calendar of the chosen crop is made.