International Journal of Remote Sensing 7725
0.73. The results underscore the superiority of the TRMM products, especially for TRMM 3B43, and suggest that the goal of the algorithm was largely achieved.
Rainfall distribution information and the structure of precipitation systems from large areas of Indonesia are important for TRMM data validation. This study observes the
Indonesian rainfall variability determined by TRMM 3B43 products, showing the capa- bility of these products to contribute to the analysis of climatic-scale rainfall in Indonesia.
To validate the results, the rainfall estimated from satellite data was compared with gauge observations over Indonesia, and we sought to determine how well the 3B43 product is an
adequate representation of monthly rainfall in Indonesia.
2. Study area
Research was conducted in the archipelago of Indonesia, which is composed of 17,508 islands of various sizes. Spatial data covered 8
◦
00
′
N to 13
◦
45
′
S and 92
◦
00
′
E to 141
◦
30
′
E. Figure 1 indicates the distribution of Indonesian topography, and six north– south cross-section lines were used to compare values of rainfall and elevation. Indonesia
is located between two continents and oceans, with a population of 237,641,326 in 2010. Sumatra, Kalimantan, Jawa Java, Sulawesi, and Papua are the five major islands,
with diverse topographical distributions. Several important mountains in Indonesia are Jayawijaya in Papua, Bukitbarisan in Sumatra, Kendeng in Jawa, Fenema and Gorontalo
in Sulawesi, and Muller in Kalimantan. Jawa Island is the most populated island and the most important industrial and agricultural region in Indonesia. Meanwhile, Sumatra,
Kalimantan, Sulawesi, and Papua are important islands that have tropical rainforests. Small islands in Indonesia are also unique and important, such as Bali, Lombok, and Halmahera.
3. Data and methods
Monthly rainfall data from 1998 to 2010, measured and collected by TRMM 3B43 satel- lite data, were employed to observe Indonesian rainfall variability. The TMPA is a
calibration-based sequential scheme for combining precipitation estimates from multiple satellites and gauge analyses where feasible, providing global coverage of precipitation
spatially over the 50
◦
S–50
◦
N latitude belt at 0.25
◦
× 0.25
◦
at three-hourly temporal
96 ° 00′
12 ° 30
′10 ° 00
′ 7°
30 ′
2° 30
′ 2°
30 ′
7° 30
′ 5°
00 ′
0° 00
′
5° 00
′
12 ° 30
′ 10
° 00 ′
7 ° 30
′ 2
° 30 ′
2 ° 30
′ 7
° 30 ′
5 ° 00
′ ° 00
′ 5
° 00 ′
102 ° 00′
108 ° 00′
114 ° 00′
120 ° 00′
126 ° 00′
132 ° 00′
138 ° 00′
2800 −4650 m
2000 −2800 m
750 −1100 m
Elevation
−120 m 120
−300 m 300
−500 m 500
−750 m 1100
−1500 m 1500
−2000 m 96
° 00′ 102
° 00′ 108
° 00′ 114
° 00′ 120
° 00′ 126
° 00′ 132
° 00′ 138
° 00′
Figure 1. The study area and Indonesian topography. Lines A-B to K-L indicate the south–north
cross sections used to compare values of rainfall and elevation. Black dots indicate the rain gauge locations.
Downloaded by [103.29.196.19] at 07:45 03 September 2013
7726 A.R. As-syakur et al.
resolution for 3B42, and monthly temporal resolution for 3B43 Huffman et al. 2007. The TMPA estimates are produced in four stages: 1 microwave estimates of precipita-
tion are calibrated and combined; 2 infrared precipitation estimates are created using the calibrated microwave precipitation; 3 microwave and infrared estimates are com-
bined; and 4 monthly data is rescaled and applied Huffman et al. 2007, 2010. The TMPA retrieval algorithm used for this product is based on the technique by Huffman
et al. 1995, 1997 and Huffman 1997. The TMPA data sets consist of 45 precipitation from passive microwave radiometers TRMM-TMI, Aqua-Advanced Microwave Scanning
Radiometer Aqua-AMSR and the Defense Meteorological Satellite Program Special Sensor Microwave Imagers DMSP-SSMIs, 40 from operational microwave sound-
ing frequencies National Oceanic and Atmospheric Administration Advanced Microwave Sounding Units NOAA-AMSUs, and 15 infrared measurements from geostationary
satellites Geostationary Operational Environmental Satellite GOES MeteosatMeteosat Second Generation MeteosatMSG Mehta and Yang 2008. According to Huffman and
Bolvin 2007, the TMPA is designed to maximize data quality, so TMPA is strongly recommended for any research work not specifically focused on real-time applications.
Several statistical scores were used to determine Indonesian rainfall variability. The types of analysis were monthly means, total means, maximum and minimum variability,
standard deviation, and trends. To investigate the effect of the Asian–Australian monsoon on Indonesia rainfall, peak-to-peak amplitude phase analysis extracted the annual signal in
each grid point of rainfall. After calculating the mean of the annual signal, the mean was removed to reveal the monthly peak amplitude phase. Furthermore, the analyses were also
carried out for the effect of land area and topography on rainfall quantities. The distribu- tion of island, sea, and topography obtained from the Shuttle Radar Topography Mission
SRTM mission was used to compare values of rainfall with regard to island distribution and elevation.
Monthly, seasonal, and long-term time-series analyses were conducted. Monthly analy- sis compared data from the same months of annual observation. Seasonal analysis is based
on the monsoon activity, described by Wyrtki 1961, over the entire observation period. Likewise, long-term analysis observed all time-series data over the entire observation
period.
Monthly accumulated rainfall data from five rain gauges derived from daily mea- surements over Indonesia see Figure 1, observed by the Indonesian Meteorology,
Climatology, and Geophysics Agency BMKG standard manual observatory and auto- matic Hellmann rain gauges, were used as references to compared satellite estimations.
The rain-gauge data cover the period from 1998 to 2010. These data were checked for consistency where unreasonable values from a climatological viewpoint, for instance zero-
rainfall months during the wet season, were deleted. The locations of rain gauges are Medan, Pontianak, Denpasar, Ternate, and Jayapura. The monthly average rainfall esti-
mated from satellite data is compared with the rain-gauge data, particularly to determine the accuracy level between the rainfall estimated from satellite and the rain-gauge data. Point-
by-point analysis was conducted on the monthly data. Point-by-point analysis consisted of a comparison between gauge data and satellite data.
4. Results and discussion