Analysis of trend in temperature and rainfall time series of an Indian arid region: comparative evaluation of salient techniques

This study investigated trends in 35 years (1979–2013) temperature (maximum, T_max and minimum, T_min) and rainfall at annual and seasonal (pre-monsoon, monsoon, post-monsoon, and winter) scales for 31 grid points in a coastal arid region of India. Box-whisker plots of annual temperature and rainfall time series depict systematic spatial gradients. Trends were examined by applying eight tests, such as Kendall rank correlation (KRC), Spearman rank order correlation (SROC), Mann-Kendall (MK), four modified MK tests, and innovative trend analysis (ITA). Trend magnitudes were quantified by Sen's slope estimator, and a new method was adopted to assess the significance of linear trends in MK-test statistics. It was found that the significant serial correlation is prominent in the annual and post-monsoon T_max and T_min, and pre-monsoon T_min. The KRC and MK tests yielded similar results in close resemblance with the SROC test. The performance of two modified MK tests considering variance-correction approaches was found superior to the KRC, MK, modified MK with pre-whitening, and ITA tests. The performance of original MK test is poor due to the presence of serial correlation, whereas the ITA method is over-sensitive in identifying trends. Significantly increasing trends are more prominent in T_min than T_max. Further, both the annual and monsoon rainfall time series have a significantly increasing trend of 9 mm year⁻¹. The sequential significance of linear trend in MK test-statistics is very strong (R² ≥ 0.90) in the annual and pre-monsoon T_min (90% grid points), and strong (R² ≥ 0.75) in monsoon T_max (68% grid points), monsoon, post-monsoon, and winter T_min (respectively 65, 55, and 48% grid points), as well as in the annual and monsoon rainfalls (respectively 68 and 61% grid points). Finally, this study recommends use of variance-corrected MK test for the precise identification of trends. It is emphasized that the rising T_max may hamper crop growth due to enhanced metabolic-activities and shortened crop-duration. Likewise, increased T_min may result in lesser crop and biomass yields owing to the increased respiration.