A Comparative Framework of VAR and LSTM Networks for Soil Moisture Prediction Using Atmospheric Data

Article Sidebar

PDF
Published: Mar 31, 2026
Keywords:
Long short-term memory Prediction Soil moisture Vector auto-regression Weather data

Main Article Content

Tawsif Mahmud
Department of Mathematics & Physics, North South University, Dhaka 1229, Bangladesh, Center of Applied & Computational Sciences, North South University, Dhaka 1229, Bangladesh
Preetom Nag
Department of Mathematics & Physics, North South University, Dhaka 1229, Bangladesh, Center of Applied & Computational Sciences, North South University, Dhaka 1229, Bangladesh
Goutam Saha
Department of Mathematics, International University of Business Agriculture and Technology, Dhaka 1230, Bangladesh
Jiaul Haque Saboj
Department of Mathematics & Physics, North South University, Dhaka 1229, Bangladesh, Center of Applied & Computational Sciences, North South University, Dhaka 1229, Bangladesh
Suvash C. Saha
School of Mechanical and Mechatronic Engineering, University of Technology Sydney, NSW 2007, Australia

Abstract

Accurate prediction of soil moisture is vital for optimizing water resource management and agricultural productivity, yet it remains a challenging problem due to the complex, nonlinear relationships between soil dynamics and atmospheric conditions. To address this, this study proposes a comprehensive forecasting framework comparing conventional statistical methods with advanced deep learning techniques. The key contribution of this work is a robust methodological pipeline that integrates a comprehensive set of ten meteorological variables, including temperature, relative humidity, wind speed, radiation, and vapor pressure deficit, following rigorous time-series preprocessing and stationarity checks. Specifically, we formulate and compare a Vector Auto-regression (VAR) model against a Long Short-Term Memory (LSTM) network for one-day ahead forecasting of soil moisture during cropping seasons. To evaluate the efficacy of the proposed techniques, we utilized historical data spanning from 1972 to 2024. The evaluation results demonstrate that the LSTM network significantly outperforms the conventional VAR model, achieving exceptional accuracy with an R2 exceeding 0.99 and notably lower RMSE values. These findings validate the reliability of the proposed LSTM framework and establish its superiority over traditional statistical approaches for precise soil moisture estimation.

Article Details

How to Cite
Tawsif Mahmud, Preetom Nag, Saha, G., Jiaul Haque Saboj, & Suvash C. Saha. (2026). A Comparative Framework of VAR and LSTM Networks for Soil Moisture Prediction Using Atmospheric Data. Science & Technology Asia, 31(1), 260–281. retrieved from https://ph02.tci-thaijo.org/index.php/SciTechAsia/article/view/260602
Issue
Vol.31 No.1 (January-March 2026)
Section
Engineering
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

References

LeCun Y, Haffner P, Bottou L, Bengio Y, van Leeuwen J, Hartmanis J, et al. Object recognition with gradient-based learning. In: Lecture Notes in Computer Science. Germany: Springer Berlin/Heidelberg; 1999. p. 319-45.

Rumelhart DE. Learning representations by backpropagating errors. Nature. 1986;323:533-6.

Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997;9(8):1735-80.

Feng D, Fang K, Shen C. Enhancing streamflow forecast and extracting insights using long short-term memory networks with data integration at continental scales. Water Resour Res. 2020;56(9).

Sheikhi F, Kowsari Z. Time series forecasting of COVID-19 infections and deaths in Alpha and Delta variants using LSTM networks. PLoS One. 2023;18(10).

Mosavi A, Ozturk P, Chau KW. Flood prediction using machine learning models: Literature review. Water. 2018;10(11).

Solomatine DP, Ostfeld A. Data-driven modelling: some past experiences and new approaches. J Hydroinform. 2008;10(1):3-22.

Usmani S, Shamsi JA. LSTM based stock prediction using weighted and categorized financial news. PLoS One. 2023;18(3).

Baruah RD, Roy S, Bhagat RM, Sethi LN. Use of data mining technique for prediction of tea yield in the face of climate change of Assam, India. In: 2016 International Conference on Information Technology (ICIT). IEEE; 2016. p. 265-9.

Han J, Mao K, Xu T, Guo J, Zuo Z, Gao C. A soil moisture estimation framework based on the CART algorithm and its application in China. J Hydrol. 2018;563:65-75.

Zhang H, Liu J, Li H, Meng X, Ablikim A. The impacts of soil moisture initialization on the forecasts of weather research and forecasting model: A case study in Xinjiang, China. Water (Basel). 2020;12(7).

Fereres E, Goldhamer DA, Parsons LR. Irrigation water management of horticultural crops. HortScience. 2003;38(5):1036-42.

Ballester C, Brinkhoff J, Quayle WC, Hornbuckle J. Monitoring the effects of water stress in cotton using the green red vegetation index and red edge ratio. Remote Sens. 2019;11(7).

Datta P, Faroughi SA. A multihead LSTM technique for prognostic prediction of soil moisture. Geoderma. 2023;433.

Cai Y, Zheng W, Zhang X, Zhangzhong L, Xue X. Research on soil moisture prediction model based on deep learning. PLoS One. 2019;14(4).

Moghadas D, Badorreck A. Machine learning to estimate soil moisture from geophysical measurements of electrical conductivity. Near Surf Geophys. 2019;17(2):181-95.

Greifeneder F, Notarnicola C, Wagner W. A machine learning-based approach for surface soil moisture estimations with Google Earth Engine. Remote Sens. 2021;13(11).

Wai KP, Chia MY, Koo CH, Huang YF, Chong WC. Applications of deep learning in water quality management: A state-ofthe-art review. J Hydrol. 2022;613.

Jiang S, Chen G, Chen D, Chen T. Application and evaluation of an improved LSTM model in the soil moisture prediction of southeast Chinese tobaccoproducing areas. J Indian Soc Remote Sens. 2023;51(9):1843-53.

Singh S, Kaur S, Kumar P, Singh SN, Panigrahi BK, Kothari DP, et al. Forecasting soil moisture based on evaluation of time series analysis. In: Advances in Power and Control Engineering. Singapore: Springer; 2019. p. 145-56.

Huang C, Li L, Ren S, Zhou Z, Liu Y, Chen Y, et al. Research of soil moisture content forecast model based on genetic algorithm BP neural network. In: Computer and Computing Technologies in Agriculture IV. Berlin: Springer; 2011. p. 309-16.

Wu Z, Cui N, Zhang W, Liu C, Jin X, Gong D, et al. Estimating soil moisture content in citrus orchards using multi-temporal Sentinel-1A data-based LSTM and PSO-LSTM models. J Hydrol. 2024;637.

Han H, Choi C, Jung J, Kim HS. Deep learning with long short-term memory based sequence-to-sequence model for rainfall-runoff simulation. Water (Basel). 2021;13(4).

Filipović N, Brdar S, Mimić G, Marko O, Crnojević V. Regional soil moisture prediction system based on long shortterm memory network. Biosyst Eng. 2022;213:30-8.

Mok JY, Choi JH, Moon YI. Prediction of multipurpose dam inflow using deep learning. J Korea Water Resour Assoc. 2020;53(2):97-105.

Bai P, Liu X, Xie J. Simulating runoff under changing climatic conditions: A comparison of the long short-term memory network with two conceptual hydrologic models. J Hydrol. 2021;592.

Lu F, Zheng Y, Cleveland H, Burton C, Madigan D. Bayesian hierarchical vector autoregressive models for patientlevel predictive modeling. PLoS One. 2018;13(12).

Basir MS, Noel S, Buckmaster D, Ashik-E-Rabbani M. Enhancing subsurface soil moisture forecasting: A long short-term memory network model using weather data. Agriculture (Basel). 2024;14(3).

Yan J, Hu L, Zhen Z, Wang F, Qiu G, Li Y, et al. Frequency-domain decomposition and deep learning based solar PV power ultra short-term forecasting model. IEEE Trans Ind Appl. 2021;57(4).

Hutson AD. A robust Pearson correlation test for a general point null using a surrogate bootstrap distribution. PLoS One. 2019;14(5).

Ozcicek O, Douglas MW. Lag length selection in vector autoregressive models: Symmetric and asymmetric lags. Appl Econ. 1999;31(4):517-24.

Li H, Yang Z, Yan W. An improved AIC onset-time picking method based on regression convolutional neural network. Mech Syst Signal Process. 2022;171.

Rajakumar G, Du KL, Rocha Á. Intelligent communication technologies and virtual mobile networks: Proceedings of ICICV 2023. Singapore: Springer; 2023.

Holtz-Eakin D, Newey W, Rosen HS. Estimating vector autoregressions with panel data. Econometrica. 1988;56(6):1371-95.

Mohsenipour M, Shahid S, Chung ES, Wang XJ. Changing pattern of droughts during cropping seasons of Bangladesh. Water Resour Manag. 2018;32(5):1555-68.

Pires IM, Hussain F, Garcia NM, Lameski P, Zdravevski E. Homogeneous data normalization and deep learning: A case study in human activity classification. Future Internet. 2020;12(11).

Namin AH, Leboeuf K, Muscedere R, Wu H, Ahmadi M. Efficient hardware implementation of the hyperbolic tangent sigmoid function. In: 2009 IEEE International Symposium on Circuits and Systems (ISCAS). IEEE; 2009.

Agarap AF. Deep learning using rectified linear units (ReLU). arXiv. 2018.

Reyad M, Sarhan AM, Arafa M. A modified Adam algorithm for deep neural network optimization. Neural Comput Appl. 2023;35(23):17095-112.

Singh S, Kumar S, Singh DS, Satapathy SC, Biswal BN, Udgata SK. Modified mean square error algorithm with reduced cost of training and simulation time for character recognition in back propagation neural network. In: FICTA 2013. Cham: Springer; 2014. p. 137-45.

Zhang B, Zhang Y, Jiang X. Feature selection for global tropospheric ozone prediction based on the BO-XGBoost-RFE algorithm. Sci Rep. 2022;12(1).

Biazar SM, Fard AF, Singh VP, Dinpashoh Y, Majnooni-Heris A. Estimation of evaporation from saline-water with more efficient input variables. Pure Appl Geophys. 2020;177(11):5599-619.

Chicco D, Warrens MJ, Jurman G. The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci. 2021;7.

Dufour JM, Dagenais MG. Durbin-Watson tests for serial correlation in regressions with missing observations. J Econometrics. 1985;27(3):371-81.

King ML. The alternative Durbin-Watson test: An assessment of Durbin and Watson’s choice of test statistic. J Econometrics. 1981;17(1):51-66.

Jalil A, Rao NH, Ozcan B, Ozturk I. Time series analysis (stationarity, cointegration, and causality). In: Environmental Kuznets Curve (EKC): A Manual. London: Elsevier; 2019. p. 85-99.

Franses PH. A multivariate approach to modeling univariate seasonal time series. J Econometrics. 1994;63(1):133-51.

Campos J, Ericsson NR, Hendry DF. Cointegration tests in the presence of structural breaks. J Econometrics. 1996;70(1):187-220.

Ozturk I, Al-Mulali U. Investigating the validity of the environmental Kuznets curve hypothesis in Cambodia. Ecol Indic. 2015;57:324-30.

Salaeh N, Ditthakit P, Pinthong S, Hasan MA, Islam S, Mohammadi B, et al. Long short-term memory technique for monthly rainfall prediction in Thale Sap Songkhla River Basin, Thailand. Symmetry (Basel). 2022;14(8).

Behroozi-Khazaei N, Nasirahmadi A. A neural network based model to analyze rice parboiling process with small dataset. J Food Sci Technol. 2017;54(8):2562-9.