Logo 2016 Dragon 3 Final Results Symposium

Conference Agenda

Overview and details of the sessions and sub-session of this conference. Please select a date or session to show only sub-sessions at that day or location. Please select a single sub-session for detailed view (with abstracts and downloads if available).

 
Session Overview
Session
F2: ID.10448 Farmland Drought
Time:
Tuesday, 05/Jul/2016:
10:30am - 11:30am

Session Chair: Stefano Pignatti
Workshop: Land & Environment
Location: Sun Moon Room -2, 5.5 Floor, Junyi Dynasty Hotel

Show help for 'Increase or decrease the abstract text size'
Presentations
Oral presentation

Farmland drought evaluation based on the assimilation of multi-temporal multi-source remote sensing data into Aqua crop model

Guijun Yang1, Hao Yang1,4, Stefano Pignatti2, Raffaele Casa3, Paolo Cosmo Silvestro3

1Beijing Academy of Agriculture and Forestry Sciences, China, People's Republic of; 2CNR IMAA; 3Universita' della Tuscia; 4Institute of Forest Resource Information Techniques, Chinese Academy of Forestry;

In the context of the Dragon-3 Farmland Drought evaluation, our cooperation study focuses on developing methods for the assimilation of biophysical variables, estimated from multi-source remote sensing data, into crop growth model, in order to assess drought-induced crop yield losses at regional scale. As a parallel strategy relative to Italy partners, our research will be crop biophysical variables estimation by multi-source remote sensing data, and the estimated biophysical variables was assimilated in FAO Aqua crop model.

Firstly, the above-ground biomass of winter wheat was estimated by combined using of SAR and optical satellite data. Multi-temporal Radarsat-2 and HJ-1A/B imagery was obtained during the entire growing season in Yangling rural region in 2014. The study investigated: 1) the relationships of biomass with several optical spectral vegetation indices (OSVIs) and radar polarimetric parameters (RPPs), 2) estimate biomass with combined OSVIs and RPPs, 3) to use multiple stepwise regression (MSR) and partial least squares regression (PLSR) to test and compare the estimations of biomass of winter wheat. The results showed the product of OSAVI and DERD are highly correlated with biomass and the estimation accuracy was better than using OSVIs and RPPs alone, which indicated that COSVI-RPPs can be used to robustly estimate biomass. Extended Fourier Amplitude Sensitivity Test (EFAST) method was used for assessing the contribution of different crop parameters to model output.

Secondly, the retrieved biomass was assimilated into crop water response model (FAO Aqua crop), in order to estimate crop yield in regional scale. In order to reduce the used parameters in assimilation procedure, the most important crop parameters of Aqua crop model was determined. Then, the estimated biomass was assimilating to FAO Aqua crop model for improving the winter wheat yield estimation, with the Particle Swarm Optimization (PSO) method, which used to reduce the gap between the remotely sensed and model simulated biomass. These procedures were used in a spatial application with data collected in the rural area of Yangling (Shaanxi Province) in 2014 and were validated for a number of wheat fields for which ground yield data had been recorded and according to statistical yield data for the area. Our study highlights the potential of multi-source remote sensing data assimilation with simple crop water response models in estimating the yield losses due to drought at regional scale.


Oral presentation

Spatialized Application Of Remotely Sensed Data Assimilation Methods For Farmland Drought Monitoring Using Two Different Crop Models

Paolo Cosmo Silvestro1, Raffaele Casa1, Stefano Pignatti2, Simone Pascucci2, Hao Yang3,4, Guijiun Yang3

1Università della Tuscia (Viterbo), Italy; 2CNR IMAA, Tito Scalo, Potenza, Italy; 3National Engineering Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, China; 4Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing, China;

To assess the effect of water stress on yield losses, in the context of farmland drought evaluation, our objective was an improved prediction of crop yield at the regional scale in areas prone to drought, assimilating remotely sensed biophysical variables such as Leaf Area Index and Canopy Cover, into crop growth models. Biophysical variables are retrieved using an algorithm based on the training of artificial neural networks (ANN) on PROSAIL, to convert the reflectance, measured by the optical sensors on board of HJ1A, HJ1B and Landsat8, into Leaf Area Index (LAI) and Canopy Cover (CC).

We used two crop models of differing degree of complexity: the more complex Aquacrop and a modified version of the simpler SAFY (Simple Algorithm For Yield) including a very simple water balance module according to the FAO 56 equations.

To identify the models’ key parameters and variables, to be tackled during the assimilation, a combination of two global sensitivity analysis methods, Morris and EFAST, was applied. These methods were used to obtain a rank of the most influential parameters for both Aquacrop and SAFY, in order to select a limited group of parameters to be further optimized. The use of a reduced number of parameters to vary in the assimilation procedures is appropriate for the correct functioning of the assimilations algorithms based on EnKF, so that the preliminary sensitivity analysis is an essential processing phase. The combined use of sensitivity and assimilation procedures has allowed to assess the impact of water stress on the production.

For Aquacrop we develop an optimization procedure to reduce the gap between the remotely sensed and model simulated Canopy Cover. For the modified version of SAFY, we employed an assimilation procedure based on the Ensemble Kalman Filter, which controls crop model runs, in order to recalibrate the most influential parameters and assimilate the remotely sensed LAI value, improving the estimation of yield and monitoring the crop’s growth status as affected by drought.

These procedures, using both models, were tested in a spatialized application, by using data collected in the rural area of Yangling (Shaanxi Province) in 2013, 2014 and 2015. Results, for a selected number of wheat fields, have been validated by utilizing both ground yield data and statistical yield data for the selected area.

Silvestro-Spatialized Application Of Remotely Sensed Data Assimilation Methods_Cn_version.pdf

Poster

Comparison Between The Use Of SAR And Optical Data For Wheat Yield Estimations Using Crop Model Assimilation

Paolo Cosmo Silvestro1, Hao Yang2,3, Guijun Yang2, Raffaele Casa1, Stefano Pignatti4

1Università della Tuscia (Viterbo), Italy; 2National Engineering Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, China; 3Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing; 4CNR IMAA, Tito Scalo, Potenza, Italy;

The ultimate aim of this work is to develop methods for the assimilation of the biophysical variables estimated by remote sensing in a suitable crop growth model. Two strategies were followed, one based on the use of Leaf Area Index (LAI) estimated by optical data, and the other based on the use of biomass estimated by SAR. The first one estimates LAI from the reflectance measured by the optical sensors on board of HJ1A, HJ1B and Landsat, using a method based on the training of artificial neural networks (ANN) with PROSAIL model simulations. The retrieved LAI is used to improve wheat yield estimation, using assimilation methods based on the Ensemble Kalman Filter, which assimilate the biophysical variables into growth crop model. The second strategy estimates biomass from SAR imagery. Polarimetric decomposition methods were used based on multi-temporal fully polarimetric Radarsat-2 data during the entire growing season. The estimated biomass was assimilating to FAO Aqua crop model for improving the winter wheat yield estimation, with the Particle Swarm Optimization (PSO) method. These procedures were used in a spatial application with data collected in the rural area of Yangling (Shaanxi Province) in 2014 and were validated for a number of wheat fields for which ground yield data had been recorded and according to statistical yield data for the area.

Silvestro-Comparison Between The Use Of SAR And Optical Data_Cn_version.pdf


 
Contact and Legal Notice · Contact Address:
Conference: 2016 Dragon 3 Final Results Symposium
Conference Software - ConfTool Pro 2.6.94
© 2001 - 2016 by H. Weinreich, Hamburg, Germany