With growth in the world’s population comes an increase in demand for food and a need to develop tools for more efficient farming. By integrating remote sensing data into crop models, Yuxi Zhang is using data assimilation techniques to improve wheat monitoring and yield estimation. View Halo Profile >>
Tell us about your research
Data assimilation is a technique to fuse remote sensing observations of wheat and soil to update the model states to provide more accurate model estimates. I am working on assimilating remotely sensed wheat observations into a wheat model with an advanced data assimilation technique. My work showed that some wheat states are promising in improving wheat yield estimation. For example, the LAI was found to improve other wheat states’ estimation and yield at harvest in my experiment, when the model was well-calibrated with ground-based data. I am further investigating whether the assimilation of remotely sensed LAI can still improve the model performance in worse conditions where the model parameters are unavailable and thus uncalibrated.
I am working on assimilating remotely sensed wheat observations into a wheat model with an advanced data assimilation technique.
Can you explain that to a non-scientist?
Wheat monitoring and grain yield estimation are important for farmers to understand their crop and make wiser farm management decisions. While crop models provide an insight into the wheat growth process over time (accounting for the interaction between the plant, atmosphere and soil), these models require multiple types of parameters, which requires a lot of labor to collect or calibrate as input for accurate estimation. Space-borne remote sensing information, primarily spectral and microwave bands, now provide wheat and soil observations at finer resolutions, in a shorter revisit time and with spatial board coverage widely used for agricultural monitoring. Integrating the advantage of crop models and remote sensing data is the technique known as “data assimilation.” I am working on the assimilation of remote sensing data into a wheat model for improved monitoring and yield estimation. This technique provides an understanding of spatial and temporal variability for the yield so that farmers can know when, where and by how much to irrigate and fertilize their farm.
This technique provides an understanding of spatial and temporal variability for the yield so that farmers can know when, where and by how much to irrigate and fertilize their farm.
Why did you choose this area of research?
With the world’s population growth and increasing food demand, agriculture and food security are positively related to everyone on Earth. I devoted myself to this research with the motivation to help farmers improve field productivity with the same cultivated area and less input.
How could your Grants4Ag project someday impact #healthforall #hungerfornone?
Suppose the assimilation of remote sensing data can reduce the uncertainty and provides a competent yield estimation even when the model is uncalibrated. It allows a wheat model to be used in a new field without a prior requirement for collecting soil properties or calibrating the model. Namely, one can simulate a field using only remotely sensed data and a few essential field and management information. The effort and cost for farmers to understand their farms and for the government to make policies related to food security can be reduced.