Decision tree-based reduction of bias in monthly IMERG satellite precipitation dataset over India

Chaudhary, Shushobhit; Dhanya, C. T.

doi:10.2166/h2oj.2020.124

Cited by 7 publications

(2 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Có thể thấy các thuật toán học máy đã trở thành một công cụ mạnh mẽ liên kết dữ liệu vệ tinh với việc ước tính lượng mưa trong những năm gần đây. Hiện nay, nhiều nghiên cứu về các kỹ thuật học máy khác nhau đã được cung cấp rộng rãi, do đó nó mở ra một hướng mới cho việc mô phỏng các mối quan hệ giữa dữ liệu vệ tinh với việc khôi phục trường lượng mưa, ví dụ như kỹ thuật sử dụng mạng thần kinh nhân tạo (ANN) [1,[6][7][8][9][10]; thuật toán học máy có giám sát dạng vector hỗ trợ (SVM) [11][12][13], cây quyết định (DT) [14][15], học sâu (DL) [16][17]. Bên cạnh đó, thuật toán rừng ngẫu nhiên (RF) [18], cũng được sử dụng rộng rãi và rất thích hợp sử dụng trong việc ước tính lượng mưa viễn thám [19][20][21][22][23][24].…”

Section: Giới Thiệuunclassified

Ước tính lượng mưa sử dụng dữ liệu vệ tinh Himawari-8 dựa trên mô hình học máy Random Forest

Nguyễn,

Bùi,

Nguyễn

et al. 2023

VNJHM

View full text Add to dashboard Cite

Bài báo sử dụng phương pháp học máy Random Forest (RF) để nghiên cứu ước tính lượng mưa từ dữ liệu vệ tinh Himawari-8. Dữ liệu được sử dụng là 09 kênh ảnh riêng lẻ và 36 kênh ảnh tổ hợp của vệ tinh Himawari-8 và dữ liệu đo mưa bề mặt để ước tính lượng mưa cho khu vực Việt Nam. Phương pháp RF được áp dụng thử nghiệm cho 04 đợt mưa lớn xảy ra trong các năm 2019, 2020 và 2023. Kết quả nghiên cứu cho thấy rằng phương pháp RF đã mô phỏng được trọng tâm của các đợt mưa lớn. Hệ số tương quan giữa lượng mưa ước tính từ vệ tinh và lượng mưa quan trắc (R) đều đạt từ 0,8 trở lên, giá trị sai số tuyệt đối trung bình (MAE) dưới 1,1 mm và sai số bình phương trung bình gốc (RMSE) dưới 2,17 mm. Tuy nhiên phương pháp RF có xu hướng cho kết quả ước tính lượng mưa thiên thấp hơn so với lượng mưa quan trắc thực tế.

show abstract

Section: Giới Thiệuunclassified

Ước tính lượng mưa sử dụng dữ liệu vệ tinh Himawari-8 dựa trên mô hình học máy Random Forest

Nguyễn,

Bùi,

Nguyễn

et al. 2023

VNJHM

View full text Add to dashboard Cite

show abstract

“…Rainfall varies spatially and temporally (Courty et Traditionally, rainfall observations obtained from a gauge station network are used to drive hydrological models (Zhang and Han 2017; Zeng et al 2018) to obtain runoff data at several points on a river. The observed rainfall from the gauges is considered an accurate measurement of the data source and this serves as the basis for calibrating and assessing precipitation estimates from other sources such as satellite and weather radar (Navarro et (Chaudhary and Dhanya 2020). Hence there is a need to evaluate their performance and applicability using different statistical procedures and hydrologic modeling.…”

Section: Introductionmentioning

confidence: 99%

Evaluation of Remote Sensing Rainfall Products, Bias Correction and Temporal Disaggregation Approaches, for Improved Accuracy in Hydrologic Simulations

yaswanth

Kannan

Bindhu

et al. 2023

Water Resour Manage

View full text Add to dashboard Cite

Flood risk management studies require reliable estimates of extreme precipitation at high spatial-temporal distribution to force hydrologic models. Recently, Remote Sensing Rainfall Products (RRPs) have gained signi cant importance in the eld of hydrometeorology, but their applicability in urban hydrologic predictions remains uncertain. The current study evaluates the accuracy of RRPs in comparison with observed rainfall and the signi cance of space-time representation of rain in simulating single and bimodal ood hydrographs. The current study is conducted for the Adyar river basin, a rapidly developing urban area in Chennai experiencing frequent oods. Sub-daily rainfall retrievals from three different satellites and Doppler Weather Radar (DWR) are the Remote sensing Rainfall products (RRPs) selected in the present study. Continuous and categorical statistical indices are selected to evaluate the performance of satellite rainfall estimates. Then the hydrologic utility of RRPs is conducted using the HEC-HMS model for ve extreme precipitation events. The RRPs simulated the rising and recession portion of ood hydrographs accurately with a bias in peak discharge. Then, two approaches are selected to further improve the ood hydrograph simulations in the current study; 1) Hydrologic model simulations after disaggregating the daily station data to sub-daily scale using time characteristics of RRPs, 2) Hydrologic simulations after bias adjusting the RRPs with station data. We found substantial improvements in model results in the two approaches. The disaggregation approach using satellite rainfall estimates has overcome the insu ciency of sub-daily rainfall observations. The bias adjusted radar rainfall data is found as best performing for the ood hydrograph simulations.

show abstract

Bias correction method of high-resolution satellite-based precipitation product for Peninsular Malaysia

Iqbal

Shahid

Ahmed

et al. 2022

Theor Appl Climatol

View full text Add to dashboard Cite

Satellite-based precipitation (SBP) is emerging as a reliable source for high-resolution rainfall estimates over the globe. However, uncertainty in SBP is still significant, limiting their use without evaluation and often without bias correction. The bias correction of SBP remained a challenge for atmospheric scientists. In this study, the performance of six SBPs, namely, SM2RAIN-ASCAT, IMERG, GsMap, CHIRPS, PERSIANN-CDS and PERSIANN-CSS in replicating observed daily rainfall at 364 stations over Peninsular Malaysia was evaluated. The bias of the most suitable SBP was corrected using a novel machine learning (ML)-based bias-correction method. The proposed bias-correction method consists of an ML classifier to correct the bias in estimating rainfall occurrence and an ML regression model to correct the amount of rainfall during rainfall events. The performance of different widely used ML algorithms for classification and regression were evaluated to select the suitable algorithms. IMERG showed better performance, showing a higher correlation coefficient (R 2 ) of 0.57 and Kling-Gupta Efficiency (KGE) of 0.5 compared to the other products. The performance of random forest (RF) was better than the knearest neighbourhood (KNN) for both classification and regression. RF classified the rainfall events with a skill score of 0.38 and estimated the rainfall amount of a rainfall event with the modified Index of Agreement (md) of 0.56. Comparison of IMERG and bias-corrected IMERG (BIMERG) revealed an average reduction in RMSE by 55% in simulating observed rainfall. The proposed bias correction method performed much better when compared with the conventional bias correction methods such as linear scaling and quantile regression. The BIMERG could reliably replicate the spatial distribution of heavy rainfall events, indicating its potential for hydro-climatic studies like flood and drought monitoring in the study area.

show abstract

Decision tree-based reduction of bias in monthly IMERG satellite precipitation dataset over India

Cited by 7 publications

References 43 publications

Ước tính lượng mưa sử dụng dữ liệu vệ tinh Himawari-8 dựa trên mô hình học máy Random Forest

Ước tính lượng mưa sử dụng dữ liệu vệ tinh Himawari-8 dựa trên mô hình học máy Random Forest

Evaluation of Remote Sensing Rainfall Products, Bias Correction and Temporal Disaggregation Approaches, for Improved Accuracy in Hydrologic Simulations

Bias correction method of high-resolution satellite-based precipitation product for Peninsular Malaysia

Contact Info

Product

Resources

About