Winter wheat LAI inversion considering morphological characteristics at different growth stages coupled with microwave scattering model and canopy simulation model

https://doi.org/10.1016/j.rse.2020.111681Get rights and content

Highlights

  • A microwave scattering model for wheat at different growth stages was proposed.

  • A winter wheat LAI inversion model coupled with MSM and CSSM was constructed.

  • The calibration, application and validation of model were finished effectively.

  • The proposed LAI inversion model had high accuracy in the regional application.

Abstract

To better eliminate the adverse effects of the ground surface on winter wheat Leaf area index (LAI) inversions and to further improve the accuracy of regional winter wheat LAI inversion using SAR remote sensing data, considering the morphological characteristics at different wheat growth stages, a winter wheat LAI inversion model coupled with the microwave scattering model (MSM) for winter wheat at different growth stages (MSMDGS) and the canopy scattering simulation model (CSSM) was proposed. In this research, taking Hengshui City of Huanghuaihai Plain of North China as the study region, using RADARSAT-2 data as image sources and based on parameter sensitivity analysis and model calibration, the proposed model was applied and validated. The LAI inversion results of winter wheat showed that the proposed model had good performance in the regional application and that LAI inversion results with high accuracy could be obtained. Among the three key growth stages (jointing stage, booting stage and heading stage) of winter wheat, the R2, adjusted R2 and RMSE between the LAI inversion value and the ground-measured data were 0.918, 0.917 and 0.675, respectively, which indicated that the winter wheat LAI inversion model coupled with MSMDGS and CSSM had certain feasibility and applicability.

Introduction

Leaf area index (LAI) is one of the major parameters for describing the structure and characteristics of vegetation canopies (Liang et al., 2015), and it is widely used in research fields such as agriculture, ecology and climate change (Sun et al., 2017; Zeng et al., 2017). Additionally, many studies have shown that crop LAIs are closely related to the yields at crop-specific growth stages; thus, crop LAI is also an important quantitative indicator reflecting crop growth and yield (Huang J. et al., 2015; Li et al., 2018; Lunagaria and Patel, 2019). Therefore, the accurate acquisition of crop LAI may significantly improve the accuracy of crop growth monitoring and yield estimation in agricultural remote sensing. Presently, remote-sensing-based crop LAI inversion mainly uses optical remote sensing data (Hu et al., 2007; Ganguly et al., 2008; Liu et al., 2012; Baret et al., 2007, Baret et al., 2013; Xiao et al., 2013; Liang et al., 2015; Verrelst et al., 2015); however, optical remote sensing data are susceptible to meteorological conditions such as clouds, rain and fog, which not only reduces the inversion accuracy of crop LAI but also increases the uncertainty of the inversion results. With the development of active microwave remote sensing technology, synthetic aperture radar (SAR) has strong application potential to crop parameter inversions and growth monitoring due to its ability to obtain observations all day without the influence of clouds, rain, fog, etc. Furthermore, SAR is sensitive to changes in vegetation characteristics. Currently, SAR has been widely used in the fields of agricultural resource investigation, land resource utilization, crop growth monitoring and yield estimation, and agricultural disaster warnings (Abdikan et al., 2016; Erten et al., 2016; Kumar et al., 2018).

The use of SAR is advantageous because microwaves are sensitive to the structure and dielectric constant of ground objects (Ulaby et al., 1982), and shortwave multipolarization SAR is widely used in the parameter inversion fields of ground surface (soil moisture) or vegetation (LAI, biomass, etc.) (Bériaux et al., 2011; Inoue et al., 2014; Wiseman et al., 2014; Hosseini et al., 2015; Tao et al., 2016). The SAR backscattering coefficient and vegetation parameters are often used to establish parameter inversion models based on SAR data, and the inversion models can be divided into empirical models and physical models. The empirical model uses statistical methods to establish a linear or nonlinear relationship between the vegetation LAI and the SAR backscattering coefficients. The empirical modeling methods are simple and easy to operate, but the methods lack theoretical support and depend largely on modeling data, vegetation type and characteristics (Chakraborty et al., 2005; Asilo et al., 2019; Mandal et al., 2019). The physical model is established by simulating the physical processes of scattering, emission and absorption of microwaves in the vegetation canopy or underlying surface. Although the vegetation LAI inversion process by physical modeling is complex, it has strong theoretical support and can be used to analyze the interaction mechanism between the SAR backscattering microwaves and vegetation canopy (Liang et al., 2005; Park et al., 2012; Alemohammad et al., 2018; Urban et al., 2018).

To study the interaction between the SAR backscattering microwaves and vegetation canopy, some scholars have constructed vegetation microwave scattering models (MSM), and these models can be divided into continuous medium models and discrete medium models. The continuous medium model assumes that the vegetation canopy is a continuous medium of dielectric constant random variation, and the average scattering coefficient of vegetation can be calculated by the variance and correlation function of the dielectric constant (Ulaby et al., 1990). However, the input data of the continuous medium model are not easily related to the actual physical parameters of the vegetation scatterer, and the model cannot describe the vegetation scattering process using microwaves. Therefore, by studying the vegetation MSM, the continuous medium model was gradually replaced by the discrete medium model. In the discrete medium model, vegetation is regarded as a collection of discrete scatterers with certain dielectric constants, dimensions and spatial orientation (McDonald and Ulaby, 1993; Toure et al., 1994; Sun and Ranson, 1995). The theoretical basis of the discrete medium model can be divided into two categories: analytical wave theory and radiative transfer theory. Analytic wave theory is based on Maxwell's equation for establishing differential or integral equations of statistical parameters such as the variance or correlation function to achieve the goal of analyzing the microwave scattering process. Although the analytic wave theory mathematically considers the effects of multiple scattering, diffraction, and interference on the propagation of microwaves within vegetation, it is difficult to include in the actual solution (Lang and Sighu, 1983). Radiative transfer theory explores the electromagnetic radiative intensity during several transmissions, scatterings, and absorptions in heterogeneous and random media by analyzing the accumulation of the electromagnetic field strengths; radiative transfer theory is more quantitative and theoretical than analytical wave theory (Du et al., 2006; Thirion et al., 2006; Wang et al., 2009; Liao et al., 2013; Inoue et al., 2014; Zhang et al., 2014).

When the theory is applied to the study of vegetation electromagnetic scattering, for tall vegetation (such as trees), the microwave scattering characteristics of the canopy, trunk and ground surface layer should be considered in the model (Ulaby et al., 1990). However, for short crops such as wheat and cotton, except for the trunk layer, only the microwave scattering characteristics of the crop and ground surface layer should be considered. Although many scholars have conducted research on the short crop MSM and have obtained some results (Toure et al., 1994; Picard et al., 2003; Jia et al., 2014; Liu et al., 2016; Yuzugullu et al., 2016; Yuzugullu et al., 2017), there are still many problems in the research of crop MSMs, including incomplete crop coverage and variability in the crop data at different growth stages. Therefore, how to analyze the actual microwave scattering of crops more accurately and establish a crop MSM are key technical issues to be addressed in the crop field or farmland parameter inversion based on SAR remote sensing data.

In addition, H polarized waves are scattered one or more times in the vegetation canopy to form HH and HV polarized waves, and the HH and HV polarized waves are sensitive to the vegetation LAI (Ulaby et al., 1984; Durden et al., 1995). Therefore, from the interaction mechanism between the H polarized wave and vegetation canopy, combined with the MSM, the relationship between HH, HV polarized waves and LAI can be established, and LAI can be retrieved by SAR data, which are the main processes of the canopy scattering simulation model (CSSM).

Because of insufficient consideration of crop coverage and the crop growth stage in the existing MSM, to improve the analysis of the winter wheat microwave scattering mechanism at different growth stages, taking Hengshui City of Hebei Province in North China as the study region, a MSM for winter wheat at different growth stages (MSMDGS) for winter wheat coverage was proposed based on the improvement of the original MSM. Based on the MSMDGS, the LAI inversion model of winter wheat coupled with CSSM was constructed, and the parameters of the wheat backscattering coefficient were introduced as input data to achieve the high-accuracy inversion of winter wheat LAI at the regional scale. The proposed LAI inversion method is intended to further improve the accuracy of regional winter wheat LAI based on SAR data and provide new ideas and technical methods for agricultural, ecological and climate change research.

Section snippets

Study area

The study area is in Hengshui City, located on the Huanghuaihai Plain in northern China, which is a typical winter wheat growing area and important food producing region. The study region covers an area of 25 km × 25 km and has a warm temperate continental monsoon climate, where the terrain is relatively flat. The annual double-crop rotation system is the most popular planting pattern in the region. Among the crops, the main summer-harvest crop is winter wheat, which accounts for >90% of the

Methodology

In this study, the winter wheat LAI inversion physical model based on SAR remote sensing data was innovatively established by coupling MSMDGS and CSSM in consideration of the morphological characteristics at different growth stages. First, to improve the analysis that insufficiently considered crop coverage and crop growth stage in the existing MSM, and according to the actual growth and morphological characteristics of winter wheat, the MSMDGS was proposed. The proposed model was used to

Parameter sensitivity analysis

The winter wheat LAI inversion model coupled with MSMDGS and CSSM contained some input parameters, such as SAR parameters, wheat growth data and soil data. Among them, the SAR parameters included the radar frequency and incident angle, which could be obtained from the SAR image. The wheat growth data mainly included wheat leaf and ear data, and soil data mainly included soil moisture, soil correlation length and root mean square height. Wheat and soil data varied within a given region and were

Mechanism of the proposed model and the rationality of its parameters

In this research, the winter wheat microwave scattering ratio r was the junction point of MSM and CSSM, that is, r was the output of MSM and the input of CSSM. Because the microwave backscattering coefficient of the wheat canopy was difficult to measure accurately, the value of r was also not measured accurately. Because r was the input of CSSM and could not be obtained via ground measurement, r must be simulated by MSM. Due to the above reasons, to obtain the r value of the whole study region,

Conclusions

To better eliminate the adverse effects of the ground surface on winter wheat LAI inversion using SAR remote sensing data and to further improve the accuracy of regional LAI inversion, considering the morphological characteristics at different wheat growth stages, a LAI inversion model coupled with MSM and CSSM was constructed for winter wheat. Taking Hengshui City of Huanghuaihai Plain of North China as the study region and using RADARSAT-2 imaging, based on parameter sensitivity analysis and

CRediT authorship contribution statement

Shangrong Wu:Conceptualization, Methodology, Validation, Writing - original draft, Writing - review & editing.Peng Yang:Methodology, Validation, Writing - review & editing.Jianqiang Ren:Validation, Investigation, Writing - review & editing, Supervision.Zhongxin Chen:Validation, Writing - review & editing, Funding acquisition.Changan Liu:Investigation, Writing - review & editing.He Li:Investigation, Writing - review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (41801286, 41871353, 41921001 and 41871358), the Young Elite Scientists Sponsorship Program by CAST (2018CAASS04), the Open Project Fund for Key Laboratory of Agricultural Remote Sensing, Ministry of Agriculture and Rural Affairs (201708), and the Fundamental Research Funds for Central Non-profit Scientific Institution (1610132018016). The authors would like to thank Dr. Philip Lewis from University College London (UCL)

References (57)

  • L. Liang et al.

    Estimation of crop LAI using hyperspectral vegetation indices and a hybrid inversion method

    Remote Sens. Environ.

    (2015)
  • D. Mandal et al.

    An investigation of inversion methodologies to retrieve the leaf area index of corn from C-band SAR data

    Int. J. Appl. Earth Obs. Geoinf.

    (2019)
  • I.M. Sobol

    Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates

    Math. Comput. Simul.

    (2001)
  • F.T. Ulaby et al.

    Relating the microwave backscattering coefficient to leaf area index

    Remote Sens. Environ.

    (1984)
  • J. Verrelst et al.

    Experimental Sentinel-2 LAI estimation using parametric, non-parametric and physical retrieval methods–a comparison

    ISPRS J. Photogramm. Remote Sens.

    (2015)
  • J. Yang

    Convergence and uncertainty analyses in Monte-Carlo based sensitivity analysis

    Environ. Model Softw.

    (2011)
  • Y. Zhang et al.

    Retrieving canopy height and density of paddy rice from Radarsat-2 images with a canopy scattering model

    Int. J. Appl. Earth Obs. Geoinf.

    (2014)
  • S. Abdikan et al.

    Land cover mapping using sentinel-1 SAR data. The international archives of photogrammetry

    Remote Sensing and Spatial Information Sciences

    (2016)
  • S. Asilo et al.

    Relating X-band SAR backscattering to leaf area index of rice in different phenological phases

    Remote Sens.

    (2019)
  • E. Bériaux et al.

    Estimating surface-soil moisture for retrieving maize leaf-area index from SAR data

    Can. J. Remote. Sens.

    (2011)
  • K.S. Chen et al.

    Emission of rough surfaces calculated by the integral equation method with comparison to three-dimensional moment method simulations

    IEEE Trans. Geosci. Remote Sens.

    (2003)
  • J. Du et al.

    A combined method to model microwave scattering from a forest medium

    IEEE Trans. Geosci. Remote Sens.

    (2006)
  • S.L. Durden et al.

    Microwave backscatter and attenuation dependence on leaf area index for flooded rice fields

    IEEE Trans. Geosci. Remote Sens.

    (1995)
  • E. Erten et al.

    Selection of PolSAR observables for crop biophysical variable estimation with global sensitivity analysis

    IEEE Geosci. Remote Sens. Lett.

    (2019)
  • B. Huang et al.

    Backscattering modeling of wheat using vector radiative transfer theory

    J. Appl. Remote. Sens.

    (2015)
  • J. Huang et al.

    Jointly assimilating MODIS LAI and ET products into the SWAP model for winter wheat yield estimation

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

    (2015)
  • M. Jia et al.

    Multi-temporal radar backscattering measurements and modelling of rice fields using a multi-frequency (L, S, C, and X) scatterometer

    Int. J. Remote Sens.

    (2014)
  • Y. Jin et al.

    Theory and Approach for Polarimetric Scattering and Information Retrieval of SAR Remote Sensing

    (2008)
  • Cited by (21)

    • Total leaf area estimation based on the total grid area measured using mobile laser scanning

      2023, Computers and Electronics in Agriculture
      Citation Excerpt :

      The detection of the leaf amount for a target canopy is the first and most critical step in variable-rate spraying and is the key factor limiting the development of this technology (Li et al. 2018; Nan et al. 2022). A variety of sensors are used for real-time canopy information acquisition (Barač and Petrović 2018), such as hyperspectral instruments (Sun et al. 2019; Wu et al. 2020; Zhang et al. 2022), microwave radar systems (Sun et al. 2019; Wu et al. 2020), ultrasonic sensors (Nan et al. 2019; Hu and Feng 2019; Gil et al. 2007; Sosa et al. 2019), light detection sensors (Baio et al. 2018), light detection and ranging (LiDAR) sensors (Wang and Fang, 2020), infrared sensors (Zhang et al. 2018; Underwood et al. 2016), and stereovision sensors (Underwood et al. 2016; Zhang et al. 2018). One of the most widely used sensors is LiDAR, which offers a fine measurement beam, low dispersion, high measurement accuracy, a fast response time, the ability to operate in low-light environments, high interference immunity, and low susceptibility to environmental influences (Barač and Petrović 2018; Xu et al. 2021; Xue et al. 2022).

    View all citing articles on Scopus
    View full text