Analysis of long-term precipitation changes in West Bengal, India: An approach to detect monotonic trends influenced by autocorrelations

Abstract

The precipitation of the entire Indian sub-continent is primarily driven by the monsoon wind system, which throws a tough challenge to model changes due to its natural variabilities. Additionally, the unique geographical location and diverse climatic condition of the Indian state of West Bengal (WB) accelerates this challenge even more. Such a situation can be more complicated if the assumption of randomness in the Mann-Kendall (MK) test is not taken care of. The present study inspects the dynamics of precipitation using the original MK test along with its three modified versions in WB. The modified versions are considered to incorporate all the significant autocorrelation coefficient (ACC) along with the short and long term persistence (STP and LTP) in the time series. Results showed that the presence of significant ACC was more common than the LTP behavior. The STP approach produced similar results to the original MK test, while the LTP approach reduced the number of significant trends. When the series was free from the scaling effect, consideration of all significant ACC gave better result in comparison to the STP approach. The annual precipitation is decreasing in large areas of WB, while the coastal areas are receiving increasing trends, which can intricate the estuarine and coastal processes. The Sub-Himalayan region and the western parts of WB are experiencing significant falling trend in monsoon season. Such a decreasing trend can enhance drought vulnerabilities, especially in the western parts of WB. However, the other three seasons (pre-monsoon, post-monsoon, and winter) witness the maximum number of non-significant trends.