Validation of satellite daily rainfall estimates in complex terrain of Bali Island, Indonesia

Abstract

Satellite rainfall products have different performances in different geographic regions under different physical and climatological conditions. In this study, the objective was to select the most reliable and accurate satellite rainfall products for specific, environmental conditions of Bali Island. The performances of four spatio-temporal satellite rainfall products, i.e., CMORPH 25 , CMORPH 8 , TRMM, and PERSIANN, were evaluated at the island, zonation (applying elevation and climatology as constraints), and pixel scales, using (i) descriptive statistics and (ii) categorical statistics, including bias decomposition. The results showed that all the satellite products had low accuracy because of spatial scale effect, daily resolution and the island complexity. That accuracy was relatively lower in (i) dry seasons and dry climatic zones than in wet seasons and wet climatic zones; (ii) pixels jointly covered by sea and mountainous land than in pixels covered by land or by sea only; and (iii) topographically diverse than uniform terrains. CMORPH 25 , CMORPH 8 , and TRMM underestimated and PERSIANN overestimated rainfall when comparing them to gauged rain. The CMORPH 25 had relatively the best performance and the PERSIANN had the worst performance in the Bali Island. The CMORPH 25 had the lowest statistical errors, the lowest miss, and the highest hit rainfall events; it also had the lowest miss rainfall bias and was relatively the most accurate in detecting, frequent in Bali, ≤ 20 mm day −1 rain events. Lastly, the CMORPH 25 coarse grid better represented rainfall events from coastal to inlands areas than other satellite products, including finer grid CMORPH 8 .

Consistency of satellite precipitation estimates in space and over time compared with gauge observations and snow-hydrological modelling in the Lake Titicaca region

“…The comparison of gauge-based assessment studies confirmed the spatial variability of SPE efficiency in reproducing precipitation, so no single SPE can be said to be the most effective one at global scale. For example, when TMPA, CMORPH, and PERSIANN SPE datasets were compared, TMPA was found to be closer to the observed precipitation in India (Prakash et al, 2014), the Guyana shield (Ringard et al, 2015), Africa (Serrat-Capdevila et al, 2016), Chile (Zambrano-Bigiarini et al, 2017) and South America Andean plateau (Satgé et al, 2016), whereas CMORPH was closer to observed precipitation in Bali, Indonesia (Rahmawati and Lubczynski, 2017), Pakistan (Hussain et al, 2017), and China (Su et al, 2017;Zeng et al, 2018). However, these assessments based on comparison with gauge observations did not assess SPE's potential performance over unmonitored regions.…”

Tropical Cyclone–Related Precipitation over the Northwest Tropical Pacific in Met Office Global Operational Forecasts

“…This may be due to poorly resolved land-sea constrasts or orography, even at 25-km grid spacing, as orographic effects have been observed to impact TC precipitation (Tan et al 2012;Racoma et al 2016). However, TRMM also has errors over complex orography (Rahmawati and Lubczynski 2018) so there may also be observation errors over the Philippine islands. Improvements in the system over time are encouraging, but biases remain considerable.…”

Use of remote sensing and long-term in-situ time-series data in an integrated hydrological model of the Central Kalahari Basin, Southern Africa

“…With advancement in RS techniques, various satellite rainfall products at different spatial and temporal resolutions are now readily available and their spatial and temporal resolution increases. However, these rainfall products still need to be validated against in situ data to select the optimal product and eventually to remove the bias (Lekula et al 2018b;Rahmawati and Lubczynski 2017). Also satellite-derived PET data are available as an RS product, although not as widely as rainfall and at much coarser spatial and temporal resolution.…”