Validation and intercomparison of satellite-based rainfall products over Africa with TAHMO in-situ rainfall observations

Abstract

Increasingly, satellite-derived rainfall data is used for climate research and action in Africa. In this study, we use six years of rain gauge data from 596 stations operated by the Trans-African Hydro-Meteorological Observatory (TAHMO) to validate three gauge-calibrated satellite rainfall products – CHIRPS, TAMSAT and GSMaP_wGauge – and one satellite-only rainfall product – GSMaP. Validations are stratified to evaluate performance across the continent and in East Africa, Southern Africa, and West Africa at daily, pentadal, and monthly timescales. For daily mean rainfall over Africa, CHIRPS has the highest bias at 15.5 % (0.5 mm) whereas GSMaP_wGauge has the lowest bias at 0.02 mm (0.7 %). We find higher daily rainfall event detection scores in the GSMaP products than in CHIRPS or TAMSAT. Generally, for every two rainfall events predicted by CHIRPS and TAMSAT, the GSMaP products predict three or more events. The highest mean monthly biases are produced by CHIRPS in East Africa (29 %; 26.3 mm wet bias), TAMSAT in Southern Africa (13 %; 10.4 mm dry bias) and GSMaP in West Africa (23 %; 19.6 mm wet bias). Considerable biases in seasonal rainfall are observed in all sub-regions for every satellite product. There is an increase of 0.6–1.3 mm in satellite rainfall RMSE for a 1 km increase in elevation revealing the influence of elevation on rainfall estimation by satellite models. Overall, satellite-derived rainfall products have notable errors, while GSMaP products produce comparable or better results at multiple timescales relative to CHIRPS and TAMSAT.

On the importance of phenology in the evaporative process of the Miombo Woodland: Could it be why satellite-based evaporation estimates differ?

“…The challenge posed by using satellite precipitation data in African catchments is that most, if not all, satellite precipitation products are geographically biased towards either underestimation or overestimation, despite some of them having good correlation with ground observations (i.e., Macharia et al, 2022;Asadullah et al, 2008;Dinku et al, 2007). The lack of adequate ground precipitation observations makes it difficult to validate, as well as correct, the product(s') biases with a good degree of certainty.…”

“…The lack of adequate ground precipitation observations makes it difficult to validate, as well as correct, the product(s') biases with a good degree of certainty. There is not a single precipitation product that can be said to perform better across African landscapes and southern Africa in particular (i.e., Macharia et al, 2022). There is no guarantee any of the precipitation products are spatially representative of a basin that is about 159,000 square kilometres with varying topographical attributes.…”

Validasi Curah Hujan Harian Chirps Precipitation Satellite Product Di Provinsi Kalimantan Barat

“…Berdasarkan sumbernya, data hujan produk satelit dapat dikelompokkan menjadi tiga, yaitu berbasis satelit, reanalysis, dan gabungan satelit dengan hujan observasi (Degefu, et al, 2022). Data hujan satelit Climate Hazards Group Infrared Precipitation with Stasions (CHIRPS) merupakan jenis data satelit gabungan data multi-satelit dan hujan observasi (Degefu, et al, 2022;Macharia, et al, 2022) yang dikembangkan oleh United States Geological Survey dan University of California, Santa Barbara.…”

“…Akurasi data hujan satelit sangat bervariasi antar wilayah, salah satu penyebabnya adalah dampak faktor lingkungan (Gebremedhin, et al, 2021), seperti faktor iklim lokal, topografi dan musim hujan (Macharia, et al, 2022), kondisi medan, jenis bioma, dan dominasi sistem hujan konvektif (Paredes -Trejo, et al, 2017). Faktor lingkungan tersebut menjadi salah satu penyebab data CHIRPS tidak terlepas dari ketidakpastian akurasi (Nashwan, et al, 2020), sehingga pemanfaatan data CHIRPS dalam bidang hidroklimatologi diperlukan validasi terlebih dahulu.…”

“…Hal ini sejalan dengan Trejo, et al, (2016), bahwa stasiun hujan yang lebih dekat dengan pantai memberikan hasil akurasi numerik yang lebih baik dibandingkan stasiun hujan yang letaknya jauh dari garis pantai. Hasil yang sama juga diperoleh Belay, et al, (2019);Rahmawati, et al, (2021) dan Macharia, et al, (2022) bahwa data CHIRPS overestimate dalam mengestimasi curah hujan ringan. Namun, terdapat tiga stasiun yang menunjukkan bahwa estimasi CHIRPS underestimate pada curah hujan kategori ringan, yaitu stasiun Pangsuma pada rentang curah hujan 0,1 -2,0 mm, Nangapinoh pada rentang 0,1 -1,3 mm, dan stasiun Susilo pada rentang curah hujan 0,1 -0,2 mm.…”