Natural hazards like droughts give hard hit to economies, hydrological circle and human lives. Climatic fluctuations have exacerbate the chances of such calamities in future. The consequences can be abbreviated to some degree by predicting it and doing forward planning for such situations. Several drought indices evaluate the intensity of droughts but are incapable of covering most of the important drought evoking factors and lacks universality. The presented paper proposes an energy conserving cloud-assisted system for drought severity evaluation and prediction. The framework intelligently deploys sensor only to the locations that does not add redundant information. An energy conserving sleep scheduling algorithm is applied at fog layer that economize the power consumption but maintaining maximum accuracy of the system at the same time. Spatio-temporal analysis of wide spectrum of drought inciting factors by kernel based methods give the current and future degree of severity at the cloud layer. Kernel K-means clustering and Support Vector Regression are used to evaluate the situation for current and predict it for subsequent time periods respectively. Experimental results prove the system efficiency in assessing and predicting droughts with energy saving at the sensor node level.

干旱等自然灾害给经济，水文圈和人类生活带来沉重打击。气候波动加剧了此类灾难的可能性。通过对结果进行预测并针对此类情况进行预先计划，可以在某种程度上简化后果。几种干旱指数评估了干旱的强度，但无法涵盖大多数重要的干旱诱发因素，并且缺乏普遍性。本文提出了一种用于干旱严重程度评估和预测的节能云辅助系统。该框架仅将传感器智能地部署到不添加冗余信息的位置。节能睡眠调度算法应用于雾层，可节省功耗，同时保持系统的最大准确性。通过基于核的方法对广泛的干旱诱因进行时空分析，可以得出云层当前和未来的严重程度。内核K均值聚类和支持向量回归用于评估当前情况并分别在后续时间段进行预测。实验结果证明了该系统在评估和预测干旱方面的效率，并在传感器节点级别节省了能源。