TE-LSTM: A Prediction Model for Temperature Based on Multivariate Time Series Data

Remote Sensing, Год журнала: 2024, Номер 16(19), С. 3666 - 3666

Опубликована: Окт. 1, 2024

In the era of big data, prediction has become a fundamental capability. Current methods primarily focus on sequence elements; however, in multivariate time series forecasting, is critical factor that must not be overlooked. While some consider time, they often neglect temporal distance between elements and predicted target relationship essential for identifying patterns such as periodicity, trends, other dynamics. Moreover, extraction features inadequate, discussions how to comprehensively leverage data are limited. As result, model performance can suffer, particularly tasks with specific requirements. To address these challenges, we propose new model, TE-LSTM, based LSTM, which employs encoding method fully extract features. A weighting strategy also used optimize integration information, capturing each element relative element, integrating it into LSTM. Additionally, this study examines impact different granularities model. Using Beijing International Airport station area, applied our temperature prediction. Compared baseline showed an improvement 0.7552% without granularity, 1.2047% granularity 3, 0.0953% when addressing The final results demonstrate superiority proposed highlight its effectiveness overcoming limitations existing approaches.

Язык: Английский

Probability links between meteorological drought and hydrological drought from a 3D perspective

Journal of Hydrology Regional Studies, Год журнала: 2024, Номер 56, С. 102001 - 102001

Опубликована: Окт. 10, 2024

Язык: Английский

Quantifying the Evolution of Extreme Drought Under Climate Change and Its Impacts on Vegetation Productivity Over the Hai River Basin of China

International Journal of Climatology, Год журнала: 2024, Номер unknown

Опубликована: Дек. 22, 2024

ABSTRACT There has been increasing attention paid to the effects of drought, especially extreme on vegetation productivity under climate change. However, there are still challenges in quantifying variations and adverse effect drought at a regional scale within context historical This study quantified changes characteristics droughts their Hai River Basin (HRB) China, using factual (with trends) counterfactual (detrended) data from ISIMIP3a project. Standardised Precipitation Evapotranspiration Index (SPEI) Run theory were utilised determine characteristic factors (drought frequency, duration, severity, intensity peak) By comparing forcing simulations, detected attributed climatic trends. The negative gross primary (GPP) quantified. Results showed that more serious events occurred HRB 1901 2019 than those climate. difference was exacerbated late stages (1980–2019) over most basin. A deceleration found rising pattern GPP last four decades, exacerbating Compared during 1982–2000, experienced further losses related 2000–2018 rate 2°gC°m −2 °day −1 . Furthermore, drought‐related pronounced summer, indicating sensitive this season. These findings enhance our understanding historically observed trends suggest strategies should be implemented protect drought.

Язык: Английский