Study of rainfall in Bangladesh using global precipitation measurement data

Abstract

Rainfall measurement is important but, rainfall is one of the most difficult atmospheric parameters to measure because of its large variations in space and time. In Bangladesh, only 35 rain gauge stations data are represent the rainfall throughout the country, which is insufficient. Inadequate rain gauge networks throughout the country sometimes provide incomplete information about the distribution of rainfall. To overcome this problem satellite based remote sensing device is a solution. The GPM (Global Precipitation Measurement) mission is an international satellite mission led by JAXA and NASA with the collaboration of many international partners that provide satellites carrying microwave radiometer instruments. Integrated Multi-satellitE Retrievals for GPM (IMERG) is the unified US algorithm that provides multi-satellite precipitation product. To study the rainfall and the possibility to use of IMERG rainfall data in Bangladesh, half-hourly IMERG data of spatial resolution 0.1degree×0.1degree and three hourly Bangladesh Meteorological Department (BMD) rain gauge data of 35 stations are used. The study period was from April, 2014 to May, 2015. It is found that IMERG data are highly correlated with BMD rain gauge data. The daily, monthly and seasonally correlation coefficients (CCs) between IMERG and rain gauge data are found 0.79, 0.96 and 0.99, respectively. The root mean square errors between IMERG and rain gauge are found 6.19 mm, 88.19 mm and 315.02 mm for the daily, monthly and seasonal rainfall, respectively. The standard deviations of IMERG data are found 7.04 mm, 113.83 mm and 468.70 mm for the daily, monthly and seasonal rainfall, respectively. On the other hand the standard deviations of BMD rain gauge data are found 9.86 mm, 183.21 mm and 773.43 mm, respectively. The IMERG and the rain gauge observed daily average rainfall is found 4.18mm and 5.61mm, during the study period, respectively. Monthly trend of rainfall variation of IMERG is almost same as that of rain gauge observation but most of the monsoon months IMERG is underestimated. The IMERG rainfall is overestimated by 7.93% (29.35mm) and 12.09% (3.36mm) during the pre-monsoon and winter seasons, respectively. On the other hand, IMERG rainfall is underestimated by 37.58% (629.33mm) and 3.54% (2.65mm) during the monsoon and post-monsoon seasons, respectively. The yearly underestimation of IMERG data was 27.9%. IMERG underestimated by 56.01%, 21.01%, 24.82%, and 11.63% for the very heavy rainfall (VHR, greater than 88 mm in 24 hour), heavy rainfall (88 ≥ HR ≥ 43.5 mm in 24 hour), moderate heavy rainfall (43.5 > MHR ≥ 22.5 mm in 24 hour) and moderate rainfall (22.5>MR ≥10.5 mm in 24 hour) events, respectively. On the other hand IMERG overestimated by 11.43% for the light rain (10.5>MR ≥2.5 mm in 24 hour) events. IMERG is able to detect light rainfall more precisely than the very heavy rainfall events. IMERG is underestimated by 0.34%, 0.43% and 0.66% for the 90th, 95th and 99th percentiles of rainfall, respectively and overestimated by 0.02% for the 10th percentile. Further studies are necessary to explore the potential of GPM-era IMERG data for meteorological, and hydrological research in Bangladesh.