A carregar...
Publicação
Enhancing Runoff Estimation in the Pra River Basin, Ghana, using Remote Sensing and LSTM-based Deep Learning Modeling
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.28 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Estimating runoff is vital for managing water resources, especially in areas where water is
essential for agriculture, industry, and households. Traditionally, this involved using data
from rain gauge stations and complex mathematical equations based on physical principles.
However, this study explores advanced modeling techniques integrating satellite
observations, particularly in regions lacking historical hydrological data. By employing insitu
precipitation data and various remote sensing products such as CHIRPS precipitation,
MOD09GA_006 NDVI, MOD11A1 V6.1 LST, and MOD16A2GF.061 ET from 2001 to
2020, a Fully Connected Long Short-Term Memory (FC-LSTM) model was developed to
predict monthly runoffs in the basin. Compared to the Soil and Water Assessment Tool
(SWAT) model, the LSTM model outperformed, achieving higher accuracy with an R2 of
0.94 and Nash-Sutcliffe Efficiency (NSE) of 0.93, while SWAT obtained R2 and NSE values
of 0.78 and 0.75, respectively. Additionally, integrating remote sensing variables enhanced
model performance, underscoring the potential of remote sensing in hydrological studies.
These findings advance our understanding of runoff estimation methods and underscore the
importance of LSTM models and remote sensing technology in hydrology.
Descrição
Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies
Palavras-chave
runoff estimation LSTM remote sensing Pra River Basin