Trend analysis of pre-monsoon flash floods for the Northeast Haor Region of Bangladesh to assess the impact on boro productivity

Introduction: Monsoon rains in August ۱۹۱۷ in Bangladesh affected close to ۶.۹ million people in Bangladesh. Rising waters began on August ۱۱ and by August ۱۵, one third of the country was submerged.The floods killed at least ۱۱۵ people and forced close to ۲۰۰,۰۰۰ people from their homes. In some parts of the country, up to ۸۰% of sanitation and water facilities were affected. Many shelter centres lacked the ability to provide adequate food, water, hygiene and protection, putting children, adolescent girls and women at risk. Most of the affected people were farm labourers, relying on agriculture to meet their basic needs. The extent of the disaster increased the risks of serious disease outbreaks; children dropping out of school; and violence, neglect, abuse and exploitation of children and women.Pre-monsoon (March-May) flash flood in the northeast Haor region of Bangladesh has drawn much attention due to its early onset, high frequency, and adverse impact on the Boro crop. To understand its past changes and future occurrences, a trend analysis is carried out on the observed ۳ - hourly water level data and daily rainfall data of the Haor region using the Mann-Kendall test, Trend-Free Pre-Whitening test, and Sens slope estimator.Material and methods: This study was conducted in the north-eastern region of Bangladesh covers approximately ۲۴,۵۰۰ km۲, bounded by the international border with India to the north and east, the Old Brahmaputra River to the west, and the Nasirnagar to Madhabpur and Meghna River to the south. This region is comprised of the floodplains of the Meghna River and its tributaries. The larger portion of this region is the Haor basin and is characterized by numerous large, deeply flooded depressions.The Mann Kendall Trend Test is used to analyze data collected over time for consistently increasing or decreasing trends (monotonic) in Y values. It is a non-parametric test, which means it works for all distributions, but your data should have no serial correlation. If your data does follow a normal distribution, you can run simple linear regression instead. The test can be used to find trends for as few as four samples. However, with only a few data points, the test has a high probability of not finding a trend when one would be present if more points were provided. The more data points you have the more likely the test is going to find a true trend (as opposed to one found by chance). The minimum number of recommended measurements is therefore at least ۸ to ۱۰.Results: Later, statistical trend analysis is conducted using the Mann-Kendall test and Sen’s slope. However, Lag-۱ autocorrelation was determined before statistical trend analysis. The stations having no autocorrelation were directly investigated by the MK test, whereas the stations showing significant autocorrelation were analyzed through TFPWMK. Here, TFPWMK was preferred to PWMK because PWMK deals with the influence of serial correlation on the MK test but does not address the interaction between a trend and autocorrelation process. There can be a case where a trend exists in a time series even though the time series does not comprise an autocorrelation process, and in such a situation, the use of PWMK can be erroneous.Conclusion: Although trend analysis can be extremely helpful in many applications—from climate change to sociological analysis—it’s important to keep in mind that it is not foolproof. In particular:All data (unless gathered through a population census) is liable to sampling error. The extent of this problem will increase when coarse sampling methods (e.g. convenience sampling) are used. Data is likely subject to measurement error; random, systematic, or external; trends in this error may be mistaken as trends in the actual data. “Phantom”, short term trends exist even in the most random of number sequences, so trends should be followed out as long as possible. Also, finding no trend may mean there is no trend, but it may just as likely mean that your data is insufficient to illuminate a trend which does in fact exist.A statistically significant increasing trend is found for the relative water level. The trend in rainfall is increasing, though it is not statistically significant. From the observed record, the peak of the flash floods is found to be arriving early in late March-early April (instead of late April-early May), coinciding with the harvesting period of the Boro crop. The early arrival of the flash flood can cause catastrophic damage to the Boro crop in future flash floods. None of the current Boro varieties BRRI dhan۲۸, BRRI ۳۶, BRRI dhan۶۹, BRRI dhan۸۸ are safer to save Boro from early flash floods experienced in recent years. To escape the Boro crop from an early flash flood, Boro varieties with a shorter growth duration should be introduced. This helps crop productivity. 

Keywords:

Agriculture , Auto-correlation , BRRI dhan۲۸ , Crop Productivity , Mann-Kendall , Onset , Sen’s slope , Trend Free Pre-Whitening