Spatial-temporal modeling of soil moisture using optical and thermal remote sensing data and machine learning algorithms

Nouraki, Atefeh; Golabi, Mona; Albaji, Mohammad; Naseri, Abdali; Homayouni, Saeid

doi:10.22059/ijswr.2023.356707.669469

Spatial-temporal modeling of soil moisture using optical and thermal remote sensing data and machine learning algorithms

Document Type : Research Paper

Authors

Atefeh Nouraki ¹

mona golabi ¹

mohammad albaji ¹

abdali naseri ¹

Saeid Homayouni ²

¹ Department of Irrigation and Drainage, Faculty of Water and Environmental Engineering, Shahid Chamran University of Ahvaz, Ahvaz, Iran

² Centre Eau Terre Environnement, Institut National de la Recherche Scientifique (INRS), Quebec, Canada

10.22059/ijswr.2023.356707.669469

Abstract

Spatiotemporal estimation and monitoring of soil moisture based on remote sensing observations (optical and thermal) is challenging due to its physical nature in high vegetation conditions, necessitating improving and increasing the accuracy of soil moisture estimation in these areas. Therefore, this research aimed to develop a new approach to estimating surface soil moisture in agricultural fields with dense vegetation using machine learning algorithms by incorporating optical and thermal remote sensing data and soil physical properties. For this objective, 16 Landsat-8 satellite images and more than 430 control locations were used during the sugarcane crop’s growth period in 2018-2019 at the Hakim Farabi Sugarcane Agro-Industrial company in the Khuzestan province of Iran. A set of 10 scenarios of various unique combinations of the available input variables were developed and then evaluated by five machine learning algorithms, including multiple linear regression (MLR), decision tree-based algorithms (CART and M5P), and ensemble learning-based algorithms (gradient-boosted regression trees (GBRT) and random forest regression (RFR)). According to the results, the highest correlation between input variables and surface soil moisture was observed in Soil Wetness Index (SWI) and Normalized Soil Moisture Index (NSMI) with R values of 0.79 and 0.69, respectively. Also, the highest accuracy of machine learning algorithms based on R2, RMSE, and MAE results was obtained in GBRT (0.99, 0.011, and 0.006) and RFR (0.99, 0.014, and 0.007), respectively. In general, the findings of this research show the importance of using variables based on Landsat-8 remote sensing data in combination with ensemble learning algorithms that can be independent of any ground measurements.

Keywords

High vegetation

machine learning algorithms

remote sensing

soil moisture

soil physical properties

20.1001.1.2008479.1402.54.4.5.1

Subjects

آبیاری و زهکشی

EXTENDED ABSTRACT

Introduction:

Spatiotemporal estimation and monitoring of soil moisture based on remote sensing observations are essential for managing water resources, improving agricultural land productivity, increasing water use efficiency, and assessing crop drought conditions. In this regard, methods based on optical and thermal remote sensing data have successfully estimated surface soil moisture at different scales. However, the physical nature of these data has limited and challenged their application in dense vegetation conditions, necessitating improving and increasing the estimation accuracy in these areas.

Objective:

This research aims to develop a new approach to estimating surface soil moisture in agricultural fields with dense vegetation conditions, such as sugarcane fields, using machine learning algorithms by incorporating optical and thermal remote sensing data and soil physical properties.

Materials and methods:

This study used 16 Landsat-8 images during the sugarcane crop’s growth period in 2018-2019 at Hakim Farabi Sugarcane Agro-Industrial company in the Khuzestan province of Iran. Soil moisture measurements were collected simultaneously as the satellite passed through the study area at more than 430 control locations during the period. A set of 10 scenarios of various unique combinations of the available input variables were developed. Five popular machine learning algorithms evaluated the scenarios, including multiple linear regression (MLR), decision tree-based algorithms (Classification and Regression Trees (CART) and M5-pruned (M5P)), and ensemble learning-based algorithms (gradient-boosted regression trees (GBRT) and random forest regression (RFR)).

Results and discussion:

According to the results, the highest correlation between input variables and surface soil moisture was observed in Soil Wetness Index (SWI) and Normalized Soil Moisture Index (NSMI) with R values of 0.79 and 0.69, respectively. While the NIR band with an R-value of 0.56 showed the lowest correlation. The obtained results showed the high ability of machine learning algorithms to estimate surface soil moisture in the area. The highest accuracy of machine learning algorithms based on R², RMSE, and MAE results was obtained in GBRT (0.99, 0.011, and 0.006) in scenario 9, RFR (0.99, 0.014, and 0.007) in scenario 9, M5P (0.90, 0.054, and 0.042) in scenario 9, CART (0.87, 0.058, and 0.046) and MLR (0.70, 0.07, and 0.056) in scenario 6, respectively. The importance of incorporating soil physical properties, especially clay percentage, with remote sensing data was observed only in the MLR algorithm. While in CART, M5P, GBRT, and RFR algorithms, the use of soil physical properties in combination with optical bands and different vegetation, humidity, and temperature indices did not lead to proper surface soil moisture predictions.

Conclusion:

In general, the findings of this research show the importance of using variables based on Landsat-8 remote sensing data (NTR, NSMI, NDVI, SWIR2, NIR, LST, and SWI) in combination with ensemble learning algorithms (RFR and GBRT) that can be independent of any ground measurements. The proposed method provides valuable results for estimating and monitoring surface soil moisture in high-vegetation areas.

Acharya, U., Daigh, A. L., & Oduor, P. G. (2021). Machine learning for predicting field soil moisture using soil, crop, and nearby weather station data in the Red River Valley of the North. Soil Systems, 5(4), 57.

Adab, H., Morbidelli, R., Saltalippi, C., Moradian, M., & Ghalhari, G. A. F. (2020). Machine learning to estimate surface soil moisture from remote sensing data. Water, 12(11), 3223.

Araya, S. N., Fryjoff-Hung, A., Anderson, A., Viers, J. H., & Ghezzehei, T. A. (2021). Advances in soil moisture retrieval from multispectral remote sensing using unoccupied aircraft systems and machine learning techniques. Hydrology and Earth System Sciences, 25(5), 2739-2758.

Babaeian, E., Paheding, S., Siddique, N., Devabhaktuni, V. K., & Tuller, M. (2021). Estimation of root zone soil moisture from ground and remotely sensed soil information with multisensor data fusion and automated machine learning. Remote sensing of environment, 260, 112434.

Babaeian, E., Sadeghi, M., Franz, T. E., Jones, S., & Tuller, M. (2018). Mapping soil moisture with the OPtical TRApezoid Model (OPTRAM) based on long-term MODIS observations. Remote sensing of environment, 211, 425-440.

Barrett, B., & Petropoulos, G. P. (2013). Satellite remote sensing of surface soil moisture. Remote sensing of energy fluxes and soil moisture content, 85.

Blake, G. (1965). Bulk density. Methods of Soil Analysis: Part 1 Physical and Mineralogical Properties, Including Statistics of Measurement and Sampling, 9, 374-390.

Breiman, L., Friedman, J., Olshen, R., & Stone, C. (1984). Classification and regression trees. Monterey, CA: Wadsworth & Brooks. In: Cole Advanced Books and Software.

Day, P. R. (1965). Particle fractionation and particle‐size analysis. Methods of Soil Analysis: Part 1 Physical and Mineralogical Properties, Including Statistics of Measurement and Sampling, 9, 545-567.

Duethmann, D., Smith, A., Soulsby, C., Kleine, L., Wagner, W., Hahn, S., & Tetzlaff, D. (2022). Evaluating satellite-derived soil moisture data for improving the internal consistency of process-based ecohydrological modelling. Journal of hydrology, 614, 128462.

Eshaghi, A., Motamedvaziri, B., & Feiznia, S. (2010). Landslides Hazard Zonation Using Logistic Regression Method (Case Study: Safaroud Watershed). Territory, 6(24), 67-77.

Fathololoumi, S., Vaezi, A. R., Alavipanah, S. K., Ghorbani, A., & Biswas, A. (2020). Comparison of spectral and spatial-based approaches for mapping the local variation of soil moisture in a semi-arid mountainous area. Science of the Total Environment, 724, 138319.

Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189-1232.

Gates, D. M., Keegan, H. J., Schleter, J. C., & Weidner, V. R. (1965). Spectral properties of plants. Applied optics, 4(1), 11-20.

Ge, L., Hang, R., & Liu, Q. (2019). Retrieving soil moisture over continental us via multi-view multi-task learning. IEEE Geoscience and Remote Sensing Letters, 16(12), 1954-1958.

Goetz, S. (1997). Multi-sensor analysis of NDVI, surface temperature and biophysical variables at a mixed grassland site. International journal of remote sensing, 18(1), 71-94.

Han, J., Mao, K., Xu, T., Guo, J., Zuo, Z., & Gao, C. (2018). A soil moisture estimation framework based on the CART algorithm and its application in China. Journal of hydrology, 563, 65-75.

Haubrock, S. N., Chabrillat, S., Lemmnitz, C., & Kaufmann, H. (2008). Surface soil moisture quantification models from reflectance data under field conditions. International journal of remote sensing, 29(1), 3-29.

Jones, S. B., Blonquist, J., Robinson, D. A., Rasmussen, V. P., & Or, D. (2005). Standardizing Characterization of Electromagnetic Water Content SensorsPart 1. Methodology. Vadose Zone Journal, 4(4), 1048-1058.

Khellouk, R., Barakat, A., Boudhar, A., Hadria, R., Lionboui, H., El Jazouli, A., Rais, J., El Baghdadi, M., & Benabdelouahab, T. (2020). Spatiotemporal monitoring of surface soil moisture using optical remote sensing data: a case study in a semi-arid area. Journal of Spatial Science, 65(3), 481-499.

Korres, W., Reichenau, T., Fiener, P., Koyama, C., Bogena, H. R., Cornelissen, T., Baatz, R., Herbst, M., Diekkrüger, B., & Vereecken, H. (2015). Spatio-temporal soil moisture patterns–A meta-analysis using plot to catchment scale data. Journal of hydrology, 520, 326-341.

Kubelka, P., & Munk, F. (1931). An article on optics of paint layers. Z. Tech. Phys, 12(593-601), 259-274.

Lobell, D. B., & Asner, G. P. (2002). Moisture effects on soil reflectance. Soil Science Society of America Journal, 66(3), 722-727.

Na, L., Na, R., Bao, Y., & Zhang, J. (2021). Time-lagged correlation between soil moisture and intra-annual dynamics of vegetation on the Mongolian plateau. Remote Sensing, 13(8), 1527.

Nguyen, T. T., Pham, T. D., Nguyen, C. T., Delfos, J., Archibald, R., Dang, K. B., Hoang, N. B., Guo, W., & Ngo, H. H. (2022). A novel intelligence approach based active and ensemble learning for agricultural soil organic carbon prediction using multispectral and SAR data fusion. Science of the Total Environment, 804, 150187.

Nolet, C., Poortinga, A., Roosjen, P., Bartholomeus, H., & Ruessink, G. (2014). Measuring and modeling the effect of surface moisture on the spectral reflectance of coastal beach sand. PLoS One, 9(11), e112151.

Potopová, V., Trnka, M., Hamouz, P., Soukup, J., & Castraveț, T. (2020). Statistical modelling of drought-related yield losses using soil moisture-vegetation remote sensing and multiscalar indices in the south-eastern Europe. Agricultural Water Management, 236, 106168.

Robinson, D. A., Campbell, C. S., Hopmans, J. W., Hornbuckle, B. K., Jones, S. B., Knight, R., Ogden, F., Selker, J., & Wendroth, O. (2008). Soil moisture measurement for ecological and hydrological watershed-scale observatories: A review. Vadose Zone Journal, 7(1), 358-389.

Sandholt, I., Rasmussen, K., & Andersen, J. (2002). A simple interpretation of the surface temperature/vegetation index space for assessment of surface moisture status. Remote sensing of environment, 79(2-3), 213-224.

Wang, J., Ding, J., Yu, D., Teng, D., He, B., Chen, X., Ge, X., Zhang, Z., Wang, Y., & Yang, X. (2020). Machine learning-based detection of soil salinity in an arid desert region, Northwest China: A comparison between Landsat-8 OLI and Sentinel-2 MSI. Science of the Total Environment, 707, 136092.

Wang, Y., & Witten, I. H. (1996). Induction of model trees for predicting continuous classes.

Zreda, M., Desilets, D., Ferré, T., & Scott, R. L. (2008). Measuring soil moisture content non‐invasively at intermediate spatial scale using cosmic‐ray neutrons. Geophysical research letters, 35(21)