top of page
Search

Dr. Yufang Zhang’s recent publication on Earth System Science Data

  • Writer: GEOG HKU
    GEOG HKU
  • Oct 27
  • 2 min read

A seamless global daily 5 km soil moisture product from 1982 to 2021 using AVHRR satellite data and an attention-based deep learning model

 

Congratulations to Dr. Yufang Zhang and colleagues on publishing a research article on Earth System Science Data (Vol. 17, pp. 5181-5207). The title of the article is “A seamless global daily 5 km soil moisture product from 1982 to 2021 using AVHRR satellite data and an attention-based deep learning model” and it is now available at https://doi.org/10.5194/essd-17-5181-2025. The complete product can be freely downloaded from https://glass.hku.hk/archive/SM/AVHRR/, and the annual average dataset is available at https://doi.org/10.5281/zenodo.14198201.

This study presents a deep learning (DL) framework to generate a consistent and seamless global surface soil moisture (SM) product spanning four decades (1982-2021). The proposed approach addresses a key challenge in climate change research: the lack of long-term, high-resolution, and spatiotemporally complete SM datasets. The framework uses an attention-based deep learning model (AtLSTM) to integrate four decades of AVHRR-derived albedo and land surface temperature, ERA5-Land SM, and terrain/soil data. The AtLSTM model was found to outperform other DL models and the benchmark XGBoost model, especially in high moisture conditions. The resulting 5 km GLASS-AVHRR SM product demonstrates high accuracy against in-situ measurements (median correlation coefficient = 0.73, unbiased RMSE = 0.041 m3m-3 against pre-2000 ISMN sites) and corrects for large wet biases present in the input ERA5-Land reanalysis data.



Flowchart of the proposed long-term global GLASS-AVHRR Soil Moisture estimation framework.
Flowchart of the proposed long-term global GLASS-AVHRR Soil Moisture estimation framework.

Mean global Soil Moisture maps of the (a, b) 0.25° ESA CCI combined, (c, d) 0.1° ERA5-Land, (e, f) 1 km GLASS-MODIS, and (g, h) 5 km GLASS-AVHRR SM product in January and July of 2016.
Mean global Soil Moisture maps of the (a, b) 0.25° ESA CCI combined, (c, d) 0.1° ERA5-Land, (e, f) 1 km GLASS-MODIS, and (g, h) 5 km GLASS-AVHRR SM product in January and July of 2016.

Reference: Zhang, Y., Liang, S., Ma, H., He, T., Tian, F., Zhang, G., Xu, J., 2025. A seamless global daily 5 km soil moisture product from 1982 to 2021 using AVHRR satellite data and an attention-based deep learning model. Earth System Science Data 17, 5181–5207. https://doi.org/10.5194/essd-17-5181-2025.

 

Keywords: Soil moisture, Deep learning, Attention-based LSTM, Long-term, Global, AVHRR

Comments

bottom of page