Comparing quantile regression forest and mixture density long short-term memory models for probabilistic post-processing of satellite precipitation-driven streamflow simulations

Zhang, Yuhang; Ye, Aizhong; Analui, Bita; Nguyen, Phu; Sorooshian, Soroosh; Hsu, Kuolin; Wang, Yuxuan

doi:https://doi.org/10.5194/hess-27-4529-2023

Articles | Volume 27, issue 24

https://doi.org/10.5194/hess-27-4529-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-27-4529-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 27, issue 24

Research article

|

20 Dec 2023

Research article |

| 20 Dec 2023

Comparing quantile regression forest and mixture density long short-term memory models for probabilistic post-processing of satellite precipitation-driven streamflow simulations

Yuhang Zhang, Aizhong Ye, Bita Analui, Phu Nguyen, Soroosh Sorooshian, Kuolin Hsu, and Yuxuan Wang

Supplement

https://doi.org/10.5194/hess-27-4529-2023-supplement

Data sets

GPM IMERG Final Precipitation L3 1 day 0.1 degree x 0.1 degree V06 G. J. Huffman et al. https://doi.org/10.5067/GPM/IMERGDF/DAY/06

Global Satellite Mapping of Precipitation (GSMaP) products T. Kubota et al. https://sharaku.eorc.jaxa.jp/GSMaP/index.htm

Global Agroecological Zones Assessment for Agriculture (GAEZ 2008) G. Fischer et al. http://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/harmonized-world-soil-database-v12/en/

Landuse dataset in China (1980-2015) Chinese Academy of Sciences Resource and Environmental Science Data Center http://data.tpdc.ac.cn/en/data/a75843b4-6591-4a69-a5e4-6f94099ddc2d

Dataset and results for ``Comparing machine learning and deep learning models for probabilistic post-processing of satellite precipitation-driven streamflow simulation'' Y. Zhang et al. https://doi.org/10.5281/zenodo.7187505

Model code and software

jnelson18/pyquantrf: DOI release (v0.0.3doi) Jnelson18 https://doi.org/10.5281/zenodo.5815105

Short summary

Our study shows that while the quantile regression forest (QRF) and countable mixtures of asymmetric Laplacians long short-term memory (CMAL-LSTM) models demonstrate similar proficiency in multipoint probabilistic predictions, QRF excels in smaller watersheds and CMAL-LSTM in larger ones. CMAL-LSTM performs better in single-point deterministic predictions, whereas QRF model is more efficient overall.