Multi-time soil moisture dynamic prediction based on RF-BiLSTM neural network

doi:10.14088/j.cnki.issn0439-8114.2023.04.030

Abstract

Abstract: In order to explore the change characteristics of soil moisture and improve the prediction accuracy of soil moisture, a soil moisture prediction method based on the combination of random forest and two-way long-term and short-term memory network（RF-BiLSTM） was proposed. Using the hourly data of soil volume moisture at the depth of 10 cm from 2016 to 2021 of Sanya National Climate Observatory and the data of 7 meteorological elements （air temperature, ground temperature, 10 cm ground temperature, sunshine hours, relative humidity, precipitation and evaporation） in the same period, the multi-time soil moisture prediction was carried out. The results showed that the average absolute errors （MAE） of RF-BiLSTM model for predicting soil volume water content after 6, 12, 24 and 48 hours were 0.462%, 0.702%, 0.889% and 1.282% respectively, and the determination coefficients （R²） were 0.983, 0.967, 0.951 and 0.913 respectively. The accuracy was higher than that of the long short-term memory neural network model, and BP neural network model.

Key words: bidirectional long short-term memory neural network（BiLSTM）, random forest, soil moisture, multi-time prediction

CLC Number:

S152.7

LI Wei, KUANG Chang-wu, HU Xin-xin. Multi-time soil moisture dynamic prediction based on RF-BiLSTM neural network[J]. HUBEI AGRICULTURAL SCIENCES, 2023, 62(4): 168-174.

References

[1] 李帅,邓婉珍,李健,等. 广西喀斯特地区耕作对土壤水分及降雨响应机制的影响[J]. 节水灌溉, 2021(6):31-36.
[2] 赵思远,贾仰文,牛存稳,等.黄土塬区不同土地利用方式土壤水分对降雨的响应研究[J].中国农村水利水电,2022(3):174-182,188.
[3] 周洋, 赵小敏, 郭熙. 基于多源辅助变量和随机森林模型的表层土壤全氮分布预测[J]. 土壤学报, 2022, 59(2):451-460.
[4] 吕沛恒,孙坚,刘绍.新型遥感卫星土壤水分产品在不同地表覆盖下的精度验证与分析[J].中国农村水利水电,2021(9):96-101.
[5] JAMEI M,KARBASI M,MALIK A,et al.Long-term multi-step ahead forecasting of root zone soil moisture in different climates: Novel ensemble-based complementary data-intelligent paradigms[J]. Agricultural water management,https://doi.org/10.1016/j.agwat.2022.107679.
[6] ZHANG Q D,WEI W,CHEN L D,et al.Corrigendum to “Plant traits in influencing soil moisture in semiarid grasslands of the Loess Plateau,China”[Sci. Total Environ. 718 (2020)137355][J]. Science of the total environment, https://doi.org/10.1016/j.scitotenv.2022.153297.
[7] 高言,沈洪政,杨婷,等.不同生物炭添加量对土壤水分运动过程的影响及模拟研究[J].中国农村水利水电,2021(9):134-140.
[8] MUPPIDI S, OM P P G, KISHORE B. Dragonfly political optimizer algorithm-based rider deep long short-term memory for soil moisture and heat level prediction in IoT[J]. The computer journal, https://doi.org/10.1093/comjnl/bxab215.
[9] GANESHI N G,MUJUMDAR M,YUHEI T, et al.Assessing the impact of soil moisture-temperature coupling on temperature extremes over the Indian region[J]. Atmospheric and oceanic physics, https://doi.org/10.48550/arXiv.2204.04079.
[10] GUTIERREZ E G, MAYTORENA A M, GONG Z, et al.Applying Random Forest Classification to Ultracool Dwarf Discovery in Deep Surveys. II. Color Classification with PanSTARRS, 2MASS, UKIDSS, and WISE Photometry[J]. Research notes of the aas, doi: 10.3847/2515-5172/ac6522.
[11] CHRISTOPHER B,MICAH T,CHANHEE P, et al.Seq Who:Reliable, rapid determination of sequence file identity using k-mer frequencies in Random Forest classifiers[J].Bioinformatics,2022,38(7):1830-1837.
[12] 高言,沈洪政,杨婷, 等. 不同生物炭添加量对土壤水分运动过程的影响及模拟研究[J].中国农村水利水电, 2021(9): 134-140.
[13] BILAL D K,UNEL M,TUNC L T, et al.Development of a vision based pose estimation system for robotic machining and improving its accuracy using LSTM neural networks and sparse regression-ScienceDirect[J]. Robotics and computer-integrated manufacturing, 2022, 74(1846):102262.
[14] FANG L,SHAO D.Application of long short-term memory(LSTM) on the prediction of rainfall-runoff in Karst area[J]. Frontiers in physics, doi: 10.3389/fphy.2021.790687.
[15] XU D, HU X, HONG W, et al.Power quality indices online prediction based on VMD-LSTM residual analysis[J]. IOP Publishing Ltd, doi: 10.1088/1742-6596/2290/1/012009.
[16] ALKHEDER S, ALOMAIR A.Urban traffic prediction using metrological data with fuzzy logic, long short-term memory (LSTM), and decision trees (DTs)[J]. Natural hazards, 2022, 111(2):1685-1719.
[17] 李幔, 马元婧. 基于BiLSTM改进聚类的空气质量监测点位优化[J]. 计算机系统应用, 2022, 31(6):217-223.
[18] 郝林如,郭向红,雷涛,等.气象要素缺失条件下不同机器学习模型计算参考作物蒸散量比较[J].节水灌溉,2022(7):102-108,118.
[19] 王德应, 杨永崇, 王涛, 等. 基于TVDI的河南省土壤湿度时空变化及影响因素分析[J].中国农村水利水电, 2022(6): 138-146.
[20] SANNABE A.How to improve SME performance using iterative random forest in the empirical analysis of institutional complementarity[J]. Palgrave communications, doi: 10.1057/s41599-022-01123-6.
[21] 李海涛, 孙亚男, 付建浩. RF-BiLSTM神经网络在海浪预测中的应用[J]. 计算机系统应用, 2022, 31(6):331-338.
[22] 孙晓聪,付玉慧. 基于RF-双向LSTM的集装箱吞吐量预测[J]. 上海海事大学学报, 2022, 43(1):60-65.
[23] 李柳阳, 朱青, 刘亚,等. 基于气象因子的长三角地区农田站点土壤水分时间序列预测[J]. 水土保持学报,2021(2):131-137.
[24] 杜瑞麒,张智韬,巨娟丽, 等.基于波文比和降水的综合干旱指数的构建及应用[J].节水灌溉,2020(8):63-71.
[25] 路璐, 王振龙, 杜富慧,等. 淮北平原基于水文气象多因子的土壤水分动态预测[J]. 水资源与水工程学报, 2019, 30(4):237-243.
[26] 梁鑫婕, 李卫东, 孟凡谦,等. 基于神经网络的土壤水分动态预测模型研究[J]. 南方农机, 2021, 52(15):14-17.
[27] KONG X, LING X, TANG L, et al.Random forest-based predictors for driving forces of earth pressure balance (EPB) shield tunnel boring machine (TBM)[J]. Tunnelling and underground space technology, 2022, 122:104373.
[28] 李艳,张成才,恒卫冬,等.基于多源遥感数据反演土壤墒情方法研究[J].节水灌溉,2020(8):76-81.
[29] LIU J,RAHMANI F,LAWSON K, et al.A multiscale deep learning model for soil moisture integrating satellite and in situ data[J]. Geophysical research letters,https://doi.org/10.1029/2021GL096847.