• No results found

3.   STUDY AREA, PHYSICAL BASIN CHARACTERISTICS AND DATA

3.4.   Climatic characteristics 62 

3.4.2.   Rainfall 63

3.4.2.2. Satellite rainfall data 68

In large scale southern African river basins (such as the Zambezi Basin) a consistent platform for data collection and transmission is currently under development (WMO and USAID, 2012). Modelling the hydrology of the Zambezi Basin is a challenging task because of its size and heterogeneity, but mostly because of the lack of reliable input data for calibration (Asante et

al., 2008; Liechti et al., 2011). With regard to hydrological model performance, the type and

quality of the input rainfall data are considered as equally – or even more – important than the choice of the hydrological model (Shrestha et al., 2008; Liechti et al., 2011). Satellite data is a viable option for use in data-scarce regions such as the Zambezi River Basin. However, the application of satellite derived rainfall data has not been adequately evaluated in the Zambezi River Basin (Winsemius et al., 2006). For most satellite rainfall products, data from two types of sensors are commonly used in the estimation algorithms – these are the Passive Microwave (PM) and the Visible and Infrared Radiance (VIS/IR).

The PM sensors identify the precipitation particles by the scattering caused by large ice particles present in the clouds (Kummerow et al., 1998; Xie et al., 2007). These sensors are

installed on Earth-orbiting satellites which offer only intermittent coverage of a given region of interest (currently approximately ten observations per day). Therefore, the estimation of rainfall from proxy parameters (such as cloud top temperature which can be inferred from geo-stationary observations) has been developed (Kummerow et al., 1998). The algorithms based on Infrared Radiance (IR) data relate rainfall to the cloud top temperatures and cloud optical properties through a precipitation index. The indexing method assigns a fixed rain rate to each identified cloud type (Kidd, 2001). This assumption is most effective for convective conditions but can yield crude estimates because of the weak link between cloud properties and precipitation (Asante et al., 2008; Liechti et al., 2011). With the multiple products currently available (Table 3.2), it is important to evaluate their precision and uncertainty, as well as their advantages and drawbacks, before opting for a specific application. Several studies have been conducted aiming at comparing rainfall estimates derived from satellite observations against locally observed data (Demirtas et al., 2005; Layberry et al., 2006; Winsemius et al., 2006). Given that the national networks of ground-based rainfall observations are sparse in the Zambezi River Basin, the potential for using relatively easily accessible satellite rainfall products – towards improving streamflow forecasting and early warning systems in the Zambezi River Basin – was investigated. Similar work has already been done in the basin by several researchers (for example: Winsemius et al., 2006; Beilfuss and Dos Santos, 2001; Beilfuss et al., 2009). There are several methods for correcting satellite rainfall (which are being used to derive final operational satellite rainfall datasets) that have been reported in literature (Sawunyama and Hughes, 2009; Thiemig, 2012).

A detailed evaluation of each of the sources and the methods used for all the datasets is beyond the scope of this study – reference can be made to the literature sources given in Table 3.2 for more details. In this study the focus was on the Climate Prediction Center (CPC) Rainfall Estimates product (CPC-RFE 2.0), and Tropical Rainfall Measurement Mission (TRMM 3B42). These methods were chosen for further analysis because of their spatial resolution and wider coverage in Africa. Two studies – conducted by ICIMOD and USAID (2008) (in the Hindu Kush-Himalayan Region) and Liechti et al. (2011) (for Southern Africa) – showed that the CPC-RFE 2.0 provided reasonable rainfall estimates when compared with the TRMM 3B42 and other satellite products – but CPC-RFE 2.0 still needed to be improved before being used for operational flood forecasting. Similar results were found in this study. Sawunyama and Hughes (2009) demonstrated that the original satellite estimates needed to be corrected, especially in areas where rainfall spatial variability is high because of topographic influences. The satellite data correction algorithm – developed by Sawunyama and Hughes (2009) at a monthly time step – and/or the Linear Interpolation Estimator (LIE) method (Morrissey et al., 1995) may be a valuable contribution toward introducing an integrated early

warning system for the Zambezi Basin. To validate these data, both visual verification and comparisons of maps of satellite estimates with observations within the Zambezi Basin were conducted (Chapter 5).

Table 3.2: An example of available global satellite rainfall products and their temporal and spatial coverage

Product Provider Spatial

coverage Temporal coverage

Spatial resolution

Temporal

resolution Reference

RFE 2.0 NOAA-CPC 20° W - 55°E 40°N - S, since 01.01.2001 0.25° 24h Xie et al., 2002 TRMM 3B42

v6 NASA 50°N - S, globally since 01.01.1998 0.50° 3h Kummerow et al., 1998

PERSIANN of Arizona University 60°N - S, globally since 01.03.2000 0.25° 6h Sorooshian et al., 2000 PERSIANN- CCS University of Arizona 50°N - S, globally since 01.03.2000 0.04° 0.5h Sooroshian et al., 2000

CMORPH NOAA-CPC 60°N - S, globally since 06.12.2002 0.25° 3h Dinku et al., 2008 CMORPH NOAA-CPC 60°N - S, globally 1.1.2006 - 31.12.2008 0.08° 0.5h Joyce et al., 2004

The Climate Prediction Center (CPC) Rainfall Estimates product (CPC-RFE 2.0)

The CPC-RFE 2.0 is computed by the National Oceanic and Atmospheric Administration Climatic Prediction Center (NOAA/CPC) (Herman et al., 1997). The African Rainfall Estimation Algorithm Version 2 (CPC-RFE 2.0), which became operational from 1 January 2001, and integrates PM estimates is used. The output daily rainfall data is a combination of PM and IR precipitation estimates, merged with daily rain gauge data from the Global Telecommunication System (GTS) records and is available from the website

ftp://ftp.cpc.ncep.noaa.gov/fews/fewsdata/africa/rfe2/bin/. The algorithm has at times,

however, resulted in rare high spikes in the precipitation estimates. The Tropical Rainfall Measuring Mission (TRMM 3B42)

The TRMM 3B42 is computed jointly by the National Aeronautics and Space Administration (NASA) and the Japan Aerospace Exploration Agency (JAXA, Kummerow et al., 1998). The data are available from http://disc.sci.gsfc.nasa.gov/precipitation/trmm3b42.

3.4.3. Potential evaporation estimation method and data sets