Monitoring sugar cane crops through dtw-based method for similarity search in NDVI time series.

AUTOR(ES)
FONTE

INTERNATIONAL WORKSHOP ON THE ANALYSIS OF MULTI-TEMPORAL REMOTE SENSING IMAGES

DATA DE PUBLICAÇÃO

2011

RESUMO

Brazil is an important sugar cane producer, which is the main resource for ethanol production, a renewable source of energy. Due to the strategical importance of this agricultural commodity, it is necessary to improve models that assist the crops monitoring process. Recently, remote sensing images have also been used to crops monitoring. Vegetation index images obtained by operations between satellite channels, for instance, can be taken over a season, showing the development of crops. Specialists in agrometeorology need methods which aim at understanding and mining these datasets to discover interesting patterns and knowledge. Accordingly, this paper presents a methodology to analyze NDVI time series using a distance function based on dynamic time warping distance (DTW) to perform similarity search. The experiments were done for NDVI multi-temporal images from seven harvests regarding the period from April/2001 to March/2008. NDVI time series was generated from NOAA-AVHRR images of a relevant sugar cane producer region in Brazil. Two different distance functions were compared and DTW reached better results than Euclidean distance. The proposed method allowed comparing harvests in different regions and in the same time series. Results of similarity search on NDVI time series demonstrate the efficacy of the use of distance function to similarity search in remote sensing data. This approach is appropriate to assess patterns in a long time series of multi-temporal images and can assist in the process of decision making by agricultural entrepreneurs.

ASSUNTO(S)

sensoriamento remoto agrometeorologia imagens multitemporais de ndvi imagens noaa-avhrr cana-de-açúcar agricultura remote sensing sugar cane agrometeorology

ACESSO AO ARTIGO

Documentos Relacionados