Soil Moisture Prediction Research Articles

Daily multistep soil moisture forecasting by combining linear and nonlinear causality and attention-based encoder-decoder model

Daily multistep soil moisture forecasting by combining linear and nonlinear causality and attention-based encoder-decoder model

Study of Machine Learning Techniques for the Estimation of Soil Moisture in Agriculture

Soil moisture is crucial in various fields and monitoring it to guide irrigation is challenging. Machine learning has emerged as a promising tool to predict soil moisture levels accurately. This study evaluates machine learning techniques for this task, training models with meteorological variables and direct soil moisture measurements. Four machine learning algorithms were implemented, highlighting the Gradient Boosting Regressor as the most effective. In addition, a processed data set that combines meteorological and soil moisture measurements is presented, hoping it will be helpful for future research. This approach seeks to improve the compression and predictability of soil moisture, which is crucial for agricultural planning and water management in agriculture

Improving subseasonal forecast skill in the Norwegian Climate Prediction Model using soil moisture data assimilation

AbstractThis study shows the importance of soil moisture (SM) in subseasonal-to-seasonal (S2S) predictions at mid-latitudes. We do this through introducing the Norwegian Climate Prediction Model Land (NorCPM-Land), a land reanalysis framework tailored for integration with the Norwegian Climate Prediction Model (NorCPM). NorCPM-Land assimilates blended SM data from the European Space Agency’s Climate Change Initiative into a 30-member offline simulation of the Community Land Model with fluxes from the coupled model. The assimilation of SM data reduces error in SM by 10.5 % when validated against independent SM observations. It also improves latent heat flux estimates, illustrating that the adjustment of underlying SM significantly augments the capacity to model land surface dynamics. We evaluate the added value of land initialisation for subseasonal predictions, by comparing the performance of hindcasts (retrospective prediction) using the standard NorCPM with a version where the land initial condition is taken from NorCPM-Land reanalysis. The hindcast covers the period 2000 to 2019 with four start dates per year. Land initialisation enhances SM predictions, reducing error by up to 2.5-month lead time. Likewise, the error for precipitation and temperature shows improvement up to a lead time of 1.5-month. The largest improvements are observed in regions with significant land-atmospheric coupling, such as the Central United States, the Sahel, and Central India. This method further enhances the prediction of extreme temperature variations, both high and low, with the most notable improvements seen in regions at mid and high latitudes, including parts of Europe, the United States, and Asia. Overall, our study provides further evidence for the significant role of SM content in enhancing the accuracy of subseasonal predictions. This study introduces a technique for improved land initialisation, utilising the same model employed in climate predictions.

Global Prediction of Flash Drought Using Machine Learning

AbstractFlash droughts are rapidly developing extreme weather events with sudden onset and quick intensification. Global prediction of flash droughts at sub‐seasonal time scales remains a great challenge. Current state‐of‐the‐art dynamic models subject to large errors and demonstrate low skills in global flash drought prediction. Here, we develop a machine learning‐based framework that uses meteorological forecasts as inputs to predict global root‐zone soil moisture and flash droughts from 1 day to 2 week lead times. The results indicate that 33% and 24% global flash drought onset and termination events can be correctly predicted by machine learning at 7 day lead time, versus 19% and 11% fractions by state‐of‐the‐art dynamic model. The developed machine learning model demonstrates substantial improvements over dynamic model in global soil moisture prediction, and thus enhances global flash drought forecasting skills in space and time. The presented framework may benefit global flash drought prediction and early warning at sub‐seasonal scales.

Applying minirhizotrons to observe spatiotemporal variations in rooting depth and distribution in agroecosystems to improve the performance of hydrological models

AbstractTo understand and explain soil moisture dynamics, the role of the vegetation is crucial. Hydrological processes in agricultural soils are strongly affected by rooting depth and root distribution. We present a new approach to monitor and model root dynamics and its influence on soil moisture using minirhizotrons combined with phenological observations. Field setups consist of a portable root scanner and acrylic glass tubes installed in the soil at the start of the growing season. 360° scans of soil and roots are taken regularly at different depths in the tubes. Root traits are identified automatically for each soil layer and complemented by observations of aboveground plant phenology. Results from minirhizotron data collected at an agrometeorological observatory in Germany show for both cereal and rapeseed (Brassica napus L.) crops a higher root density in deeper soil layers and a lower density near the soil surface when compared to literature data. An opposite picture emerged for maize (Zea mays L.) and potato (Solanum tuberosum L.), whereas vertical root distribution in sugar beet (Beta vulgaris subsp. vulgaris) had a different seasonal course than expected from literature. Applying the new root distribution data in calculations of the soil water balance resulted in differences of more than 3% in absolute volumetric soil water content. A comparison with in situ measurements of volumetric soil moisture at 20‐ and 50‐cm depth revealed a significant improvement of the model results due to the new parameterization. Thus, we argue that minirhizotrons constitute a useful supplement to hydrological observatories and can help understand and predict soil moisture dynamics in the critical zone.

Integrating multi-source remote sensing and machine learning for root-zone soil moisture and yield prediction of winter oilseed rape (Brassica napus L.): A new perspective from the temperature-vegetation index feature space

Accurately assessing root-zone soil moisture is crucial for precision irrigation, as it directly influences crop yield. The Temperature-Vegetation Index (Ts-VI) Feature Space, which combines land surface temperature (Ts) and vegetation index (VI), is widely used to evaluate root-zone soil moisture in vegetated areas. However, its effectiveness in estimating crop yield remains unclear. Therefore, the objectives of this study are: (1) to collect multispectral and thermal infrared remote sensing data from a two-year (2021–2023) field experiment on winter oilseed rape (Brassica napus L.), and to optimize and evaluate the fitting methods of the dry and wet edges of the Ts-VI feature space based on the selected vegetation indices; (2) to analyze the spatiotemporal patterns of the Temperature Vegetation Dryness Index (TVDI) derived from the optimized Ts-VI feature space and estimate root-zone soil moisture (SM) and crop yield; and (3) to precisely invert the SM and yield of winter oilseed rape in the 0–60 cm root-zone using three machine learning algorithms—Support Vector Regression (SVR), Extreme Gradient Boosting Regression (XGBR), and Random Forest Regression (RFR)—based on the optimized TVDI. Results indicate that, among the various fitting methods, the polynomial fitting method shows the best performance. The performance of the root-zone soil moisture prediction models across different growth stages follows the order of budding stage > seedling stage > flowering stage, and with the increase of soil depth, the performance of the model gradually deteriorates.In the yield inversion of winter oilseed rape, TVDI effectively predicts yield, with the coefficient of determination (R2) ranging from 0.430 to 0.480 and RMSE ranging from 213.399 to 267.212 kg ha−1 during the seedling stage, R2 ranging from 0.640 to 0.747 and RMSE ranging from 110.712 to 178.133 kg ha−1 during the budding stage, and R2 ranging from 0.680 to 0.773 and RMSE ranging from 83.815 to 147.301 kg ha−1 during the flowering stage. The flowering stage effectively reflects crop yield trends and allows for accurate yield prediction of winter oilseed rape up to two months in advance. A comparison of the modeling results from XGBR, SVR, and RFR shows that XGBR provides the best fit for both root-zone soil moisture and yield predictions. Compared to linear regression models, the three machine learning models significantly improve accuracy and fit, providing more precise evaluations of root-zone soil moisture and yield. In addition, through the comparison and verification of this method in other regions, it shows that the results also have certain reference value. The combination of the Ts-VI feature space and machine learning algorithms not only enables precise monitoring of root-zone soil moisture conditions but also predicts future crop yield trends, offering valuable insights for water resource management and irrigation decision-making in precision agriculture.

Comparison of transformer, LSTM and coupled algorithms for soil moisture prediction in shallow-groundwater-level areas with interpretability analysis

Accurate quantification of soil moisture is essential for understanding water and energy exchanges between the atmosphere and the Earth’s surface, as well as for agricultural applications. Predicting soil moisture content is vital for efficient water management, irrigation scheduling, and drought monitoring. Traditional forecasting methods, such as numerical regression models, often struggle due to various influencing factors and poor observation data quality. In contrast, deep learning algorithms, particularly recurrent and convolutional neural networks, show promise in predicting nonlinear data like soil moisture. This study focuses on shallow groundwater regions, using groundwater levels and meteorological data as features while coupling the Transformer model with other neural network structures. We investigate the potential of attention-based neural networks for soil moisture time series prediction. Our findings demonstrate that the Transformer model achieves an average R2 of 0.523 across different time lags, outperforming the LSTM model with an R2 of 0.485. The introduction of LSTM enhances the Transformer’s stability in handling temporal changes. Additionally, we verified the importance of groundwater for soil moisture prediction. This study introduces new methods for soil moisture prediction and offers new insights and recommendations for the development of artificial intelligence technology for soil moisture prediction.

Smart Irrigation System Using Soil Moisture Prediction with Deep CNN for Various Soil Types

Soil moisture sensing plays a crucial role in agriculture as it directly impacts plant growth and can significantly enhance crop productivity. With the advent of technology, agriculture applications have undergone a revolution, enabling more advanced and efficient practices. One such advancement is the use of soil moisture sensors, which provide valuable information about the current water level of the soil, including whether it is dry, wet, or excessively saturated. These sensors have become indispensable tools for farmers and growers, empowering them to make informed decisions regarding irrigation schedules, water management strategies, and overall crop health. By accurately assessing soil moisture levels, farmers can optimize water usage, prevent water stress or overwatering, and promote healthier plant development, ultimately leading to improved yields and sustainability in agriculture. The objective of the proposed study is to investigate the effective soil moisture sensors by considering three sensors and an automated system for watering the soil for agriculture. A comparative analysis is performed for different commercial off-the-shelf soil moisture sensors in cost, accuracy, durability, and corrosion resistance. Secondly, this study further gives soil moisture reading as data input to the Convolutional Neural Network (CNN) to classify whether water is required or not for the soil at a particular temperature which would help to conserve water and develop agriculture. An automated system which can help the farmer for irrigation and cultivation of the soil obtained accuracy as 95.23%.

SOIL MOISTURE PREDICTION MODEL IN PEATLAND USING RANDOM FOREST REGRESSOR

Soil moisture is one of the factors that has recently become the focus of research because it is strongly correlated with forest and land fires, where low soil moisture will increase drought and the incidence of forest and land fires. For this reason, this study aims to create a prediction model for soil moisture as an early prevention of fires in peatlands using the Random Forest Regressor (RFR) algorithm. RFR is used because of its ability to predict values and its resistance to overfitting and outliers. A dataset covering soil moisture, precipitation, temperature, maturity, and peat thickness was collected from August 2019 to December 2023. The data includes soil moisture, precipitation, temperature, maturity, and peat thickness. The data were divided into 80% for modeling and 20% for testing. Model performance was optimized through random search CV, resulting in significant prediction accuracy R-squared: 0.914, MAE: 0.0081, MSE: 0.0007, RMSE: 0 .0271, and MAPE: 0.969. These findings demonstrate the effectiveness of RFR in soil moisture prediction and pave the way for more appropriate and timelier implementation of fire mitigation strategies.

Forecasting Soil Moisture in Caragana Shrubland Using Wavelet Analysis and NARX Neural Network

It is important for sustainable use of soil water resource and high-quality development to forecast the soil moisture in forestland of water-limited regions. There are some soil water models. However, there is not the best model to forecast the change of soil moisture in the caragana shrubland. In this paper, the plant water relationship has been investigated at the same time for a long term in the caragana shrubland of semiarid region of the Loess Plateau of China. The data of soil moisture was divided and then NARX neural network was used to build model I and model II. For model I, low frequency component was the input variable, and for model II, low frequency component and high frequency component were predicted. The results showed the average relative error for model I is 3.5% and for model II is 0.3%. The average relative error of predicted soil moisture in 100 cm layer using model II is 0.8%, then soil water content in the 40 cm and 200 cm soil depth is selected and the forecast errors are 4.9% and 0.4%. The results showed that using model II to predict soil water is well Predicting soil water using model II will be important for sustainable use of soil water resource and high-quality development.

Development of an online prediction system for soil organic matter and soil moisture content based on multi-modal fusion

Realizing accurate determination of in situ soil organic matter (SOM) content and soil moisture (SM) content in the field is of great importance for improving agricultural production efficiency. However, the feature information of a single sensor is limited, and there is still a certain gap in modeling accuracy compared to traditional laboratory methods. The variations of SM in the field also interfere with data collection, limiting the application of single sensor. In order to realize the efficient detection of in-situ SOM and SM content, this study developed an online detection system for SOM and SM content based on the fusion of characteristic wavelengths, visible images and thermal imaging image features. Firstly, based on a vehicle-mounted platform, a visible-thermal imaging camera and a characteristic wavelength integration device were integrated in a deep pine plow to realize the simultaneous acquisition of in-situ soil multi-sensor data. Then, a lightweight multimodal network was constructed to obtain thermal visible light image features and characteristic wavelength features through branch networks, achieving deep fusion of different modal data and predicting SOM and SM content. Finally, output the forecasting results of SOM and SM content, and transmit the predictive information to the cloud platform for storage. It was verified that the Multi-modal proposed in this study worked best in the laboratory environment, with a predicted R2 of 0.91 and RMSE of 2.9 g/kg for SOM, and a predicted R2 of 0.92 and RMSE of 0.77 % for SM. Compared with single image or characteristic spectral data, the fusion of visible image and characteristic wavelength effectively improved the prediction accuracy of SOM. The real-time prediction of SM was realized by fusing thermal imaging data, and the elimination of the moisture effect of visible images and characteristic wavelengths was also realized with the help of deep fusion of Multi-modal network. After field validation, the R2 of the Multi-modal system was 0.84 and the RMSE was 5.0 g/kg for SOM, and the R2 of the SM was 0.88 and the RMSE was 1.03 %. Despite the differences in soil types between the validation field and the sampling field, the system still demonstrated strong generalization and achieved high accuracy prediction of in-situ SOM and SM in the field, which effectively improves the efficiency and applicability of field. The efficiency and applicability of soil testing effectively improve the efficiency and provide technical guidance for precision management in the field.

Spatial-temporal simulation and prediction of root zone soil moisture based on Hydrus-1D and CNN-LSTM-Attention in the Yutian Oasis, Southern Xinjiang, China

Root zone soil moisture (RZSM) plays a critical role in land-atmosphere hydrological cycles and serves as the primary water source for vegetation growth. However, the correlation between RZSM and its associated variables, including surface soil moisture (SSM), often exhibit nonlinearities that are challenging to identify and quantify using conventional statistical techniques. Therefore, this study presents a CNN-LSTM-Attention (CLA) model for predicting RZSM. Owing to the scarcity of soil moisture observation data, the physical model Hydrus-1D was employed to simulate a comprehensive dataset of spatiotemporal vertical soil moisture. Meteorological data and MODIS vegetation characterization parameters were used as predictor variables for the training and validation of the CLA model. The results of the CLA model for soil moisture prediction within the root zone were significantly enhanced compared with those of the traditional LSTM and CNN-LSTM models. This was particularly notable at depth of 80–100 cm, where the coefficient of determination (R²) reached nearly 0.9298. Moreover, the RMSE was reduced by 49% and 57% compared with those of the LSTM and CNN-LSTM models, respectively. This study demonstrates that the integration of physical modeling and deep learning techniques provides a more comprehensive and accurate understanding of spatial and temporal soil moisture variations in the root zone.

Machine‐Learning Based Multi‐Layer Soil Moisture Forecasts—An Application Case Study of the Montana 2017 Flash Drought

AbstractSoil moisture (SM) is an essential climate variable, governing land‐atmosphere interactions, runoff generation, and vegetation growth and productivity. Timely forecasts of SM spatial distribution and vertical profiles are needed for early detection and prediction of potential droughts. However, previous studies have primarily concentrated on historical or near real‐time soil moisture mapping, with less effort devoted to the development and integration of soil moisture forecast components within drought assessment systems. A satellite‐driven machine‐learning approach was developed in this study to build complex relationships between diversified predictor data sets and in situ multi‐layer SM measurements from the Montana Mesonet, a regionally dense environmental station network in the US upper Missouri and Columbia basins. The resulting 30‐m daily SM predictions showed strong performance against in situ SM measurements from 4‐, 8‐ and 20‐inch soil layers, and with 1‐ to 2‐week forecast lead times (R > 0.91; RMSE ≤ 0.047 cm3/cm3). The machine‐learning model was subsequently applied to the entire Montana region, and the SM deficit forecasts with both 1‐ and 2‐week lead times successfully depicted onset, progression, and termination phases of the 2017 Montana flash drought, which was not effectively identified from prevailing operational systems. The resulting system is capable of delineating local scale SM heterogeneity, and could be extended to predict other critical water cycle variables, potentially enhancing future drought forecasts through multivariate assessments and benefiting water resource management, agricultural practices, and the provision of ecosystem services.

Implicit EXP-RBF techniques for modeling unsaturated flow through soils with water uptake by plant roots

Modeling unsaturated flow through soils with water uptake by plan root has many applications in agriculture and water resources management. In this study, our aim is to develop efficient numerical techniques for solving the Richards equation with a sink term due to plant root water uptake. The Feddes model is used for water absorption by plant roots, and the van-Genuchten model is employed for capillary pressure. We introduce a numerical approach that combines the localized exponential radial basis function (EXP-RBF) method for space and the second-order backward differentiation formula (BDF2) for temporal discretization. The localized RBF methods eliminate the need for mesh generation and avoid ill-conditioning problems. This approach yields a sparse matrix for the global system, optimizing memory usage and computational time. The proposed implicit EXP-RBF techniques have advantages in terms of accuracy and computational efficiency thanks to the use of BDF2 and the localized RBF method. Modified Picards iteration method for the mixed form of the Richards equation is employed to linearize the system. Various numerical experiments are conducted to validate the proposed numerical model of infiltration with plant root water absorption. The obtained results conclusively demonstrate the effectiveness of the proposed numerical model in accurately predicting soil moisture dynamics under water uptake by plant roots. The proposed numerical techniques can be incorporated in the numerical models where unsaturated flows and water uptake by plant roots are involved such as in hydrology, agriculture, and water management.

Testing the reliability of soil moisture forecast for its use in agriculture

The increased frequency and intensity of drought events are among the major impacts of climate change in various regions worldwide, including Central Europe. These changes have increased the demand for precise drought monitoring and forecasting tools and their validation. The Czech Drought Monitoring System, which is widely utilized across Central Europe, provides daily soil moisture monitoring and medium-range forecasts using the SoilClim model. The main objective of this study was to describe and evaluate the spatiotemporal reliability of these forecasts. The forecasting performance was evaluated for three variables (relative soil moisture content, soil moisture deficit, and drought intensity) and was evaluated using Pearson’s correlation, mean bias error, and mean absolute error. All the statistical analyses were performed on data from the years 2019 to 2021 aggregated at the administrative district level in the Czech Republic. The growing season data were analyzed in detail to assess the forecasting accuracy during spring and summer. Furthermore, the ability to forecast rapid changes in the soil moisture content according to changes in meteorological variables, such as precipitation and air temperature, was evaluated. Our findings demonstrate that the SoilClim model forecasts are accurate and suitable for practical applications in sectors such as agriculture and forestry. The lowest reported correlation between the monitored and forecasted values was +0.68 for nine-day forecasts at a soil depth of 0–40 cm. For shorter forecast periods of one and four days, the correlation values were +0.80 or greater. For drought intensity, the errors did not exceed one category of drought severity. We identified summer as the most dynamic season, with corresponding variations in the soil moisture and meteorological forecasting accuracy. This study validates the ability of the Czech Drought Monitoring System to provide reliable soil moisture forecasts, thus contributing to our ability to manage and mitigate drought impacts effectively.

Improving soil moisture prediction with deep learning and machine learning models

Reliable soil moisture (SM) data is critical for effective water resources management, yet its accurate measurement and prediction remain challenging. This study was conducted to develop a deep learning regression network for sub-hourly SM prediction and compare its performance with traditional machine learning models, including the eXtreme gradient boosting (XGB), light gradient-boosting (LGB), cat boosting (CB), random forest (RF), k-nearest neighbors (kNN), and long short-term memory (LSTM) models. Sub-hourly SM, electrical conductivity (EC), soil temperature (ST), and weather parameters were collected during research experiments conducted for two years (2020–2021 and 2021–2022) at the Tropical Research and Education Center (TREC), University of Florida. A network of SM sensors and a weather station were installed at the experimental site with 24 plots of green beans and sweet corn under full and three deficit irrigation treatments with three replications. Model performance metrics such as coefficient of determination (r2) and global performance indicator (GPI) were used to evaluate the performance of the models. Results showed that all MLs and DL models performed more than satisfactorily in simulating SM of green beans and sweet corn plots. The testing average r2 and GPI of MLs were 0.83 and 0.02 (green beans) and 0.85 and 0.02 (sweet corn). However, XGB and LGB models outperformed the remaining ML and DL models. The testing r2 and GPI of XGB were 0.86 and 0.014 for green beans, whereas 0.88 and 0.015 for sweet corn. The r2 and GPI values for the LGB were 0.85 and 0.014 for green beans, while 0.88 and 0.015 for sweet corn. Even though DL model took longer and resources to be trained, its performance was not as accurate as that of XGB and LGB models. However, the performance of DL was better than the LSTM model. The r2 and RMSE of the LSTM model were 0.68 and 0.02cm 3 cm-3 for green beans and 0.75 and 0.02cm 3 cm-3 for sweet corn, respectively. Whereas the r2 and RMSE of DL were 0.84 and 0.015cm 3 cm-3 (green beans) and 0.85 and 0.02 cm 3 cm-3 (sweet corn). The ML and DL models performed better in simulating SM of sweet corn plots than green beans. Overall, these results confirmed that the ML and DL models could be alternative tools for SM prediction for agricultural fields, with potential applications for irrigation scheduling and water resources management.

Improving Water-Efficient Irrigation in Terrain Durio Zibethinus Farming Using Hybrid Ant Colony Optimization-Based Soil Moisture Prediction Model

The vegetation stage of Durio Zibethinus trees is characterized by active root development, leaf expansion, and the initiation of reproductive structures. During this crucial phase, adequate irrigation is necessary to satisfy the trees' water requirements. A well-irrigated durian plantation encourages effective nutrient absorption, resulting in healthier trees with increased pest and disease resistance. Understanding the water needs of durian trees is essential for irrigation management to optimize water application and prevent water stress and waterlogging. Typically, sensors measure the soil moisture within the root zone. However, installing soil moisture sensors at each tree is laborious and prohibitively expensive. Using climatic data to forecast the value is a viable option in such a scenario. Climate data are used to create soil moisture predictions incorporated into the irrigation model. This research employs Ant Colony Optimization- Support Vector Regression (ACO-SVR) to predict soil moisture levels. The model is compared to other optimization methods, and its accuracy is assessed using statistical methods. Finally, the prediction models' findings determine the irrigation volume and schedule.

Modeling the spatial distribution of soil physical properties in a semiarid tropical region

In semiarid regions, sustainably managing the limited available water resources is critical for food, water, and economic security. An important aspect of water management is monitoring, analyzing, and predicting soil moisture and hydrologic fluxes, especially under the threats of climate change. Unfortunately, since monitoring of soil moisture and water fluxes is often sparse in these regions, one needs to rely on hydrological modeling. The latter requires accurate knowledge of soil properties (e.g., texture, porosity, hydraulic conductivity), which in semiarid regions is also complicated by high level of spatial heterogeneity that is not captured by the low density of soil sampling efforts as well as by globally gridded soil databases. Using 300 monitoring sites across the Semiarid region of NE Brazil and a soil moisture model, this study aims to estimate soil properties at the sites through inverse-modeling and then spatially interpolation these properties via geostatistics. Inverse modeling with the Levenberg–Marquardt algorithm accurately captured soil properties across the sites, while clear spatial patterns in these properties allowed us to confidently interpolation them across the region. Comparing our results with a global soil dataset shows that the latter does not capture spatial variability in the region, underlining the need to leverage small scale information. The derived soil property datasets may then particularly be valuable for regional water and ecosystem management applications.

Improving global soil moisture prediction through cluster-averaged sampling strategy

Understanding and predicting global soil moisture (SM) is crucial for water resource management and agricultural production. While deep learning methods (DL) have shown strong performance in SM prediction, imbalances in training samples with different characteristics pose a significant challenge. We propose that improving the diversity and balance of batch training samples during gradient descent can help address this issue. To test this hypothesis, we developed a Cluster-Averaged Sampling (CAS) strategy utilizing unsupervised learning techniques. This approach involves training the model with evenly sampled data from different clusters, ensuring both sample diversity and numerical consistency within each cluster. This approach prevents the model from overemphasizing specific sample characteristics, leading to more balanced feature learning. Experiments using the LandBench1.0 dataset with five different seeds for 1-day lead-time global predictions reveal that CAS outperforms several Long Short-Term Memory (LSTM)-based models that do not employ this strategy. The median Coefficient of Determination (R2) improved by 2.36 % to 4.31 %, while Kling-Gupta Efficiency (KGE) improved by 1.95 % to 3.16 %. In high-latitude areas, R2 improvements exceeded 40 % in specific regions. To further validate CAS under realistic conditions, we tested it using the Soil Moisture Active and Passive Level 3 (SMAP-L3) satellite data for 1 to 3-day lead-time global predictions, confirming its efficacy. The study substantiates the CAS strategy and introduces a novel training method for enhancing the generalization of DL models.

Soil Temperature and Moisture Forecasting Using Exponential Smoothing Method Based on Mean Absolute Percentage Error

Soil is an essential medium for almost all plants. Temperature and moisture in the soil become one of the factors that affect plant growth. Unbalanced or unfavorable soil conditions will adversely affect the plant. Therefore, soil conditions, including temperature and moisture, must be monitored to maintain plant health. Forecasting can be used to find out the soil condition in the future so that strategies can be prepared to deal with these conditions. This forecasting is carried out based on existing soil temperature and moisture data, which is then processed using the exponential smoothing method and mean absolute percentage error as a reference for the error rate or accuracy of the forecasting. The application of these methods succeeded in producing forecasting with very good overall accuracy.<em> </em>