The Use of Phenology Attributes Derived from PROBA-V 100m NDVI Imagery in Mapping of Peanut Crops Haerani haerani (a*), Armando Apan (b), Badri Basnet (b)
a) Agricultural Engineering Department, Universitas Hasanuddin, Jl. Perintis Kemerdekaan km. 10, Makassar, 90245, Indonesia
*Email: haerani[at]agri.unhas.ac.id
b) School of Civil Engineering and Surveying, University of Southern Queensland, West Street, Toowoomba, 4350, Australia
Abstract
Mapping of peanut crops will support the estimation of their production in securing the market stability of this commodity. Advances in recent development of satellite technologies provide many benefits for crop area estimation, especially in terms of capturing changes over time in a particular area, reducing cost, and increasing time effectiveness. PROBA-V satellite has a spatial resolution of 100m, which is an intermediary spatial resolution between high spatial resolution imagery (e.g. 30m resolution of Landsat) and low spatial resolution imagery (e.g. 250m resolution of MODIS). Each crop has its unique phenology which describes crop^s development stages over the growth period. Time-series data has the ability to capture this crop phenology, which is useful in discriminating crops with similar spectral behavior. With a study area in South Burnett, Queensland, Australia, the objective of this study was to examine the use of phenology attributes derived from PROBA-V 100m NDVI imagery in classifying peanut crops. The phenological parameters of PROBA-V NDVI time-series dataset was produced by employing TIMESAT program. The parameters were classified using Maximum Likelihood Classification (MLC), Spectral Angle Mapper (SAM), and Minimum Distance Classification (Min) algorithms. The study results show that MLC outperformed other algorithms, with overall accuracy of 79.53%. In general, MLC algorithm also gave good results of producer and user accuracies of each classified class, i.e. >70%, except for mungbean class which had a producer accuracy of 52%. In relation to peanut class, all classification algorithms provide good results of producer and user accuracies, i.e. >75%. The highest producer accuracy of peanut class was provided by MLC, i.e. 87.93%, while the highest user accuracy of peanut class was provided by SAM, i.e. 90.91%. This study has successfully mapped peanut crops by using phenology attributes derived from PROBA-V imagery.
