< Back to previous page

Project

Assimilation of microwave backscatter data into a regional crop model: estimating soil moisture and biomass over European croplands

Crop production has experienced exponential growths over the past centuries due to scientific innovations and mechanical automation. The more recent developments in technology, related to big data and remote sensing, have led to a growing interest of upscaling process-based crop models, which typically calculate crop production for a field under homogeneous soil, crop and management conditions. By using these models to simulate over multiple adjacent areas, crop production can be assessed from a regional perspective, which is useful for e.g. policy makers. A limitation of these upscaled crop models is generalized information of spatial input data. One way to try to correct for model errors is the use of independent observations to update (state) variables in a data assimilation system.

In my doctoral research, I focused on the potential of using Synthetic Aperture Radar (SAR) observations from the microwave satellite Sentinel-1 to correct for regional AquaCrop soil moisture and biomass simulations over Europe via data assimilation.

Before the benefits of a data assimilation system could be addressed, the model set-up for regional simulations needed to be evaluated. The first phase therefore considered the performance evaluation of a spatially distributed version of the field-scale AquaCrop model. An efficient parallel processing system was implemented to run the model regionally with coarse scale re-analysis forcings as global meteorological input, 1-km soil textural information, and generic crop information, and the model was evaluated with mainly satellite-based information in terms of soil moisture and crop biomass. In the second phase, the regional AquaCrop model simulations of soil moisture and biomass were translated to backscatter, to facilitate evaluation with Sentinel-1 observations in preparation for Sentinel-1 based data assimilation system. Finally, the evaluated regional AquaCrop model was implemented into a data assimilation framework; NASA's Land Information System (LIS). Two experiments were performed: one in which only soil moisture was updated and a second experiment in which both soil moisture and biomass were updated with Sentinel-1 backscatter observations.

Overall, the regional model set-up of AquaCrop has shown potential for large-scale applications. The model performs well in simulating soil moisture and biomass productivity over diverse regions in Europe using generic input data. The data assimilation showed potential for correcting soil moisture and biomass, but finding the optimal methodology and evaluation for updating vegetation with SAR requires further research.

Date:22 Mar 2019 →  28 Aug 2023
Keywords:Data assimilation, remote sensing, crop and water management
Disciplines:Environmental engineering modelling
Project type:PhD project