are being carried out, multisource, multivariate, and multiscale data assimilation has
gained attention in crop growth modeling and yield forecasting research (Montzka
et al. 2012; Zhiwei et al. 2014).
In this chapter, we aim at introducing the recent advances in crop growth and
yield models, remote sensing, and data assimilation on crop growth modeling and
yield forecasting. This chapter is organized into five main sections. Section 9.2
describes traditional statistical modeling approaches. The physiological/physicalbased modeling approaches and commonly used crop growth models are introduced
in detail in Section 9.3. In this section, we will focus on how to use crop growth
models to forecast crop biomass and yield. In Section 9.4, detailed information for
remote sensing in crop growth monitoring is described. Section 9.5 describes various
data assimilation approaches in crop growth models. The conclusions for crop
modeling and yield forecasting are provided in Section 9.6.
11.2 Statistical Modeling
Crop growth is a complex soil-plant-atmosphere system, which depends on large
environmental factors, such as the level of incoming radiation, climate condition,
photosynthetic characteristics of the leaves, soil moisture, and field management
condition. However, most of the time, these parameters are unavailable; usually, one
or few factors were used to estimate the crop growth state variables or yield.
Statistical analyses are the most commonly used method in crop research, and
numerous statistical models were developed. These different types of statistical
models mainly depend on the relationship between crop growth variables or yield
and weather parameters, such as some models that describe the response of crop
yield to accumulated temperature or relative humidity and the relationship between
leaf areas and days after crop emergence. This approach primarily uses the field
observation or statistics data to establish an equation or set of equations by fits them
to data. The usual method used is regression analysis include linear regression
models and nonlinear regression models, and there are other statistical methods
also used in research such as artificial neural networks, but here we focus on
regression analysis as an example in statistical modeling.
Regression analysis approaches are frequently used in statistical modeling for
crop growth monitoring and yield forecasting; several linear and nonlinear regression models were developed. In the linear regression model,
Y i ¼ a þ bX i þ ε i
where Y i and X i are crop growth state variables or yield, meteorological factors,
respectively, a and b represent model parameters to be fit, and ε i is an error term.
208
H. Pan and Z. Chen
gained attention in crop growth modeling and yield forecasting research (Montzka
et al. 2012; Zhiwei et al. 2014).
In this chapter, we aim at introducing the recent advances in crop growth and
yield models, remote sensing, and data assimilation on crop growth modeling and
yield forecasting. This chapter is organized into five main sections. Section 9.2
describes traditional statistical modeling approaches. The physiological/physicalbased modeling approaches and commonly used crop growth models are introduced
in detail in Section 9.3. In this section, we will focus on how to use crop growth
models to forecast crop biomass and yield. In Section 9.4, detailed information for
remote sensing in crop growth monitoring is described. Section 9.5 describes various
data assimilation approaches in crop growth models. The conclusions for crop
modeling and yield forecasting are provided in Section 9.6.
11.2 Statistical Modeling
Crop growth is a complex soil-plant-atmosphere system, which depends on large
environmental factors, such as the level of incoming radiation, climate condition,
photosynthetic characteristics of the leaves, soil moisture, and field management
condition. However, most of the time, these parameters are unavailable; usually, one
or few factors were used to estimate the crop growth state variables or yield.
Statistical analyses are the most commonly used method in crop research, and
numerous statistical models were developed. These different types of statistical
models mainly depend on the relationship between crop growth variables or yield
and weather parameters, such as some models that describe the response of crop
yield to accumulated temperature or relative humidity and the relationship between
leaf areas and days after crop emergence. This approach primarily uses the field
observation or statistics data to establish an equation or set of equations by fits them
to data. The usual method used is regression analysis include linear regression
models and nonlinear regression models, and there are other statistical methods
also used in research such as artificial neural networks, but here we focus on
regression analysis as an example in statistical modeling.
Regression analysis approaches are frequently used in statistical modeling for
crop growth monitoring and yield forecasting; several linear and nonlinear regression models were developed. In the linear regression model,
Y i ¼ a þ bX i þ ε i
where Y i and X i are crop growth state variables or yield, meteorological factors,
respectively, a and b represent model parameters to be fit, and ε i is an error term.
208
H. Pan and Z. Chen