assimilation algorithms usually have two steps—forecast and analysis. Remotely
sensed data of the current state are combined with the simulated state from the model
(the forecast) to analysis; then the result becomes the forecast in the next
analysis step.
11.5.1 Sequential Data Assimilation Algorithms
Crop growth model parameter calibration using remotely sensed data is most widely
used in researches of crop data assimilation. The objective is how to minimize the
cost function J. Several good research results were obtained. Fang minimization
between the modeled and observed LAI using optimization algorithm POWELL got
better LAI and yield simulation. In some studies, the global optimized algorithm
SCE-UA used Dente to assimilate LAI from ASAT and MERIS data to improve the
CERES-wheat model simulation capacity. The updating model state variable
approach also has some valuable researches in crop data assimilation. Assimilation
remotely sensed based on EnKF was widely used. Pauwels’s assimilation of
observed soil moisture and LAI into a WOFOST model using EnKF found that
the yield prediction well improved. The soil water index derived from microwave
remotely sensed data assimilated to correct the error of water balance in WOFOST
based on EnKF.
In recent years, data assimilation algorithms and variables tend to diversify.
Assimilation variables increased from LAI to canopy reflectance, soil moisture,
ET, and a combination of multiple variables. Therefore, with the development of
crop data assimilation, it will help to improve the capability of crop modeling and
yield forecasting.
11.6 Conclusion
Various kinds of methods such as statistical models, physiological/physical-based
models, remote sensing, and data assimilation are in use for monitoring and forecasting crop growth and yield. The traditional crop modeling and forecasting
methods are still widely used: statistical modeling and crop growth models.
Although there are many statistical models and crop growth models, the search for
news models is still necessary. Remote sensing models are mainly based on the
spectral indices and quantitative products from remotely sensed data. The quality of
remote sensing data is critical for the performance of crop modeling and yield
forecasting. In recent years, data assimilation of crop modeling is a hot topic and
promising. Described. More researches are needed for the full use of the value of
remote sensing and crop growth model in crop growth monitoring and yield forecasting at the regional scale.
216
H. Pan and Z. Chen
sensed data of the current state are combined with the simulated state from the model
(the forecast) to analysis; then the result becomes the forecast in the next
analysis step.
11.5.1 Sequential Data Assimilation Algorithms
Crop growth model parameter calibration using remotely sensed data is most widely
used in researches of crop data assimilation. The objective is how to minimize the
cost function J. Several good research results were obtained. Fang minimization
between the modeled and observed LAI using optimization algorithm POWELL got
better LAI and yield simulation. In some studies, the global optimized algorithm
SCE-UA used Dente to assimilate LAI from ASAT and MERIS data to improve the
CERES-wheat model simulation capacity. The updating model state variable
approach also has some valuable researches in crop data assimilation. Assimilation
remotely sensed based on EnKF was widely used. Pauwels’s assimilation of
observed soil moisture and LAI into a WOFOST model using EnKF found that
the yield prediction well improved. The soil water index derived from microwave
remotely sensed data assimilated to correct the error of water balance in WOFOST
based on EnKF.
In recent years, data assimilation algorithms and variables tend to diversify.
Assimilation variables increased from LAI to canopy reflectance, soil moisture,
ET, and a combination of multiple variables. Therefore, with the development of
crop data assimilation, it will help to improve the capability of crop modeling and
yield forecasting.
11.6 Conclusion
Various kinds of methods such as statistical models, physiological/physical-based
models, remote sensing, and data assimilation are in use for monitoring and forecasting crop growth and yield. The traditional crop modeling and forecasting
methods are still widely used: statistical modeling and crop growth models.
Although there are many statistical models and crop growth models, the search for
news models is still necessary. Remote sensing models are mainly based on the
spectral indices and quantitative products from remotely sensed data. The quality of
remote sensing data is critical for the performance of crop modeling and yield
forecasting. In recent years, data assimilation of crop modeling is a hot topic and
promising. Described. More researches are needed for the full use of the value of
remote sensing and crop growth model in crop growth monitoring and yield forecasting at the regional scale.
216
H. Pan and Z. Chen