Publications

Abstract : In light of the rapidly growing agricultural sector, spotting signs of climate change is crucial. Even a slight change in climate behaviour can hinder crop development. As a result, it is necessary to investigate these climatic signals at a localised regional scale. The current methods mainly emphasise on outliers when identifying abnormalities in geographical and temporal data. This disregards the possibility of negative impacts brought on by climatic changes at each location. This paper introduces a novel mathematical approach, named the probabilistic convolution neighborhood technique (PCNT), that finds the temporal and spatial coherence at each point. Its ability to examine distribution at all grid points makes it easy to identify even a slight change in climatic variables. This new approach has been applied to study the variations in parameters, namely humidity, precipitation, mean, and maximum temperatures, that are collected from the Indian Monsoon Data Assimilation and Analysis (IMDAA) for the time period 1980 to 2020. The datasets are daily single-level datasets from which annual and seasonal (Kharif, Rabi, and Zaid) data are obtained. Spatio-temporal properties, along with inter-vicennial and inter-decadal variations of climatic variables with regions having low, high, and normal probability distributions, are studied. The findings showed a pattern in the temporal features between Kharif and Rabi. The mean temperature for the annual and seasonal data showed decreasing trends for tail diagnostics. The study brings out an understanding in the probable change in various metrics, which are crucial in the context of crop growth and agricultural practice.