Authors : Dr. M. Asha Rani; Mallepalli Thanusha Reddy

Volume/Issue : Volume 11 - 2026, Issue 4 - April

Google Scholar : https://tinyurl.com/432m8akj

Scribd : https://tinyurl.com/5efw3ah7

DOI : https://doi.org/10.38124/ijisrt/26apr991

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Drought is a slow-developing hazard that significantly affects agriculture and water resources in semi-arid regions such as Anantapur, Andhra Pradesh, India. This study presents a multi-scalar drought and desertification prediction framework based on the Standardized Precipitation Evapotranspiration Index (SPEI) at 1-, 3-, 6-, 9-, and 12-month time scales integrated with deep learning models. Climate and vegetation datasets spanning 2000–2024, including CHIRPS rainfall, ERA5 temperature, soil moisture, and MODIS NDVI, were processed using Google Earth Engine. Three recurrent neural network models—Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional LSTM (BiLSTM)—were trained using data from 2000–2022 and evaluated on 2023–2024 observations. Model performance was assessed using RMSE, MAE, R², and MAPE. Results indicate that GRU achieves consistently lower prediction errors for SPEI across multiple time scales, demonstrating strong generalization capability. BiLSTM effectively captures short-term vegetation variability in NDVI prediction, while LSTM shows comparatively higher errors. The proposed framework provides a reliable approach for drought monitoring, early warning, and decision support in agriculture and water resource management.

Keywords : SPEI, Drought Prediction, NDVI, LSTM, GRU, BiLSTM, Multi-Scalar Analysis, Anantapur.

References :

1. U. A. Auliyah and A. Sunyoto, “A multi-scalar SPEI drought index prediction using Long Short-Term Memory,” in Proc. 6th Int. Conf. Information and Communications Technology (ICOIACT), 2023, pp. 343–348, doi: 10.1109/ICOIACT59844.2023.10455932.
2. S. Poornima and M. Pushpalatha, “Drought prediction based on SPI and SPEI with varying timescales using LSTM recurrent neural network,” Soft Computing, vol. 23, no. 18, pp. 8399–8412, 2019, doi: 10.1007/s00500-019-04120-1.
3. B. P. Kumar, K. R. Babu, M. Rajasekhar, and M. Ramachandra, “Assessment of land degradation and desertification due to migration of sand dunes in Beluguppa Mandal of Anantapur district (AP, India) using remote sensing and GIS techniques.”
4. B. P. Kumar, K. R. Babu, M. Rajasekhar, C. Krupavathi, B. N. Swamy, Y. Sreenivasulu, and M. Rajasekhar, “Data on identification of desertified regions in Anantapur district, Southern India by NDVI approach using remote sensing and GIS,” Data in Brief, vol. 30, p. 105560, 2020, doi: 10.1016/j.dib.2020.105560.
5. R. C. Staudemeyer and E. R. Morris, “Understanding LSTM — A tutorial into Long Short-Term Memory recurrent neural networks,” 2019. [Online]. Available: http://arxiv.org/abs/1909.09586
6. Z. Hao and V. P. Singh, “Drought characterization from a multivariate perspective: A review,” Journal of Hydrology, vol. 527, pp. 668–678, May 2015, doi: 10.1016/j.jhydrol.2015.05.031.