Assessment of mapping of annual average rainfall in a tropical country like Bangladesh: remotely sensed output vs. kriging estimate

Das, S. and Islam, A. R. M. T. (2021) Assessment of mapping of annual average rainfall in a tropical country like Bangladesh: remotely sensed output vs. kriging estimate. Theoretical and Applied Climatology, 146, pp. 111-123. (doi: 10.1007/s00704-021-03729-3)

Abstract

The knowledge about spatial variation of annual rainfall is important for many applications ranging from agriculture planning to flood risk management in a tropical low-lying country like Bangladesh. The remotely sensed data has emerged as a suitable addition to the data source which is often suggested for use at ungauged conditions. This study investigates whether the remotely sensed outputs on its own or its incorporation as a covariate can outperform the mapping estimate of annual average rainfall. The work primarily considers a multivariate kriging approach, kriging with external drift (KED), which can take covariates to good effect for the spatial interpolation. Other than remotely sensed annual average rainfall (RAAR), the study includes easily accessible: geographical coordinates (LON, LAT) and elevation as potential covariates. The suitability of the KED model is assessed against the widely used classical univariate, ordinary kriging (OK), and the inverse distance weighting (IDW) methods. The annual average rainfall calculated at 34 stations based on observed daily rainfall data from 1970 to 2016 was used for the assessment. Based on cross-validation techniques, the KED with LON is identified as the best interpolation method. The IDW performed poorly and came last among all the interpolation methods. The performance of remotely sensed outputs on its own is not as good as the interpolation estimate; in fact, it is outperformed by the IDW quite convincingly. The integration of RAAR as a covariate with the KED performed superior to IDW but could not outperform the chosen KED (LON) model. Overall, remotely sensed data could be served better with the integration of an appropriate kriging approach rather than to be used as model outputs.

• Item Type: Articles
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Das, Dr Samiran
• Authors: Das, S., and Islam, A. R. M. T.
• College/School: College of Science and Engineering > School of Engineering
• Journal Name: Theoretical and Applied Climatology
• Publisher: Springer
• ISSN: 0177-798X
• ISSN (Online): 1434-4483
• Published Online: 23 July 2021