No abstract

Although coastal and inland areas of Bangladesh exhibit distinct physiographic and climatic characteristics, spatiotemporal variation of extreme climatic events is poorly understood in these two areas. This study was an attempt to understand the trends in extreme climatic events in coastal and inland areas over the period 1968–2018. The missing data in daily maximum and minimum temperature, and daily rainfall datasets were imputed using the multiple imputation by chained equations technique and implementing a predictive mean matching algorithm. The imputed datasets were then tested for inhomogeneity using the penalized maximal t and modified penalized maximal F tests. A quantile matching algorithm was then applied to homogenize the datasets, which were then used for generating 13 extreme temperature and 9 extreme rainfall indices. The trends were assessed using the Trend Free Pre‐whitened Mann–Kendall test and the magnitudes of the changes were determined using the Thiel–Sen slope estimator. Additionally, standardized anomalies were calculated to understand the seasonal variability of temperature and rainfall over the past five decades. Results suggested that both coastal and inland areas were warming significantly but coastal areas exhibited a higher rate of warming. Although most of the extreme rainfall indices showed statistically non‐significant changes for coastal and inland stations, there is evidence of localized dryness and increased rainfall at individual stations. In particular, the drought‐prone northwestern region of the country experienced decreased rainfall, which is discordant to the results of previous studies. Findings from this study highlighted important local and regional‐scale changes in extreme climate events that were previously overlooked. The findings of this study can help undertake targeted climate change adaptation strategies to save population and resources.

A high-resolution (1 km × 1 km) monthly gridded rainfall data product during 1901–2018, named Bangladesh Gridded Rainfall (BDGR), was developed in this study. In-situ rainfall observations retrieved from a number of sources, including national organizations and undigitized data from the colonial era, were used. Leave-one-out cross-validation was used to assess product’s ability to capture spatial and temporal variability. The results revealed spatial variability of the percentage bias (PBIAS) in the range of −2 to 2%, normalized root mean square error (NRMSE) <20%, and correlation coefficient (R2) >0.88 at most of the locations. The temporal variability in mean PBIAS for 1901–2018 was in the range of −4.5 to 4.3%, NRMSE between 9 and 19% and R2 in the range of 0.87 to 0.95. The BDGR also showed its capability in replicating temporal patterns and trends of observed rainfall with greater accuracy. The product can provide reliable insights regarding various hydrometeorological issues, including historical floods, droughts, and groundwater recharge for a well-recognized global climate hotspot, Bangladesh.

No abstract