Content area
Full text
1. Introduction
Hydrological studies rely on the quality of rainfall estimates to produce meaningful modeling output. Rain gauges can deliver accurate point measurements, but their poor ability to describe the spatial structure of rainfall can be a major limitation when precipitation fields are required, for example, in distributed hydrological modeling applications. This problem is more severe in tropical regions because of high rainfall variability and scarce data conditions.
Satellite-based estimates such as the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA; also TRMM 3B42) product are becoming increasingly attractive as an alternative source of forcing data in data-sparse regions, although their application can be limited by quantitative inaccuracies (Zulkafli et al. 2014a; Anagnostou et al. 2010; Tian et al. 2007). Information from a large number of rain gauges is already assimilated as part of global/regional satellite algorithms; nevertheless, the rain gauge databases sourced by these procedures can exclude more extensive national networks where data accessibility is restrictive, as often is the case in developing countries. In these regions, the global precipitation product may be found to be unsatisfactory and requiring local adjustment (e.g., Lavado-Casimiro et al. 2009).
Satellite algorithms are known to internally perform gauge correction, for example, using a mean-field bias correction (MBC) and/or inverse-error-weighted averaging methods (Huffman et al. 1997; Grimes et al. 1999). Postanalysis merging methods have also been used on the products of these algorithms, for example, the spatial adjustment of TMPA using interpolations by inverse distance weighting (Lavado-Casimiro et al. 2009), double-kernel smoothing (DS; Li and Shao 2010), and the nearest neighbor method (Vila et al. 2009); correction through regression analysis between the satellite- and rain gauge-based precipitation at various temporal scales, for example, climatologies in Almazroui (2011) and monthly in Yin et al. (2008); and correction using probability distributions (Anagnostou et al. 1999). Geostatistical methods have also been used such as the kriging with external drift (KED) to combine gauge and 10-day (dekad) IR-based precipitation data from Meteosat (Grimes et al. 1999) and the cokriging approach to interpolate daily rain gauge data with the GPCP multisatellite precipitation estimates as covariates (Kottek and Rubel 2008). More recently, Heidinger et al. (2012) performed a wavelet analysis on the signals from daily rain gauge and TMPA time series and reconstructed a...