Mining Climate and Remote Sensing Time Series to Discover the Most Relevant Climate Patterns

Huge amounts of climate and remote sensing data have been stored by several institutions in the past few years. Properly analyzed, these databases can reveal useful information, which can help researchers to monitor and esti­ mate the production of agricultural crops. Recently, the information and knowl­ edge discovered from these data have also been used for research on climate changes as well as to increase the sustainable use of the soil, making farms more productive. Data mining techniques are the main tool to analyze and ex­ tract useful information, relationships and meaningful patterns. Accordingly, this paper presents a new unsupervised algorithm, called CLIPSMiner (Climate Patterns Miner) that aims at discovering relevant patterns with specific con­ straints in climate and remote sensing time series. This new algorithm works on multiple time series of continuous data, identifying all patterns or the most relevant ones according to a relevance factor, which can be tuned by the user. Experiments with synthetic and real data were performed. The results show that the algorithm detects some patterns that are known in climatology, as expected, indicating the correctness and feasibility of the proposed algorithm. Moreover, patterns detected using the highest relevance factor are coincident with extreme phenomena as many days without rain or heavy rain. This new algorithm can be used by climatology specialists to mine and discover knowledge from their long series of past and forecasting data.