![Shenglei Zhang ﹡ , Jiancheng Shi, Youjun Dou Xiaojun Yin, Liying Li, Chenzhou Liu ﹡ [email_address] Experiments of satellite data simulation based on the Community Land Model and SCE-UA algorithm IGARSS 2011, Vancouver, Canada, 24-29 July, 2011 Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing, 100101, China](https://image.slidesharecdn.com/2-shengleizhang-igarss2011-mo3-110725162004-phpapp02/85/2-ShengleiZhang_IGARSS2011_MO3-T04-2-ppt-1-320.jpg)
2 ShengleiZhang_IGARSS2011_MO3.T04.2.ppt
- 1. Shenglei Zhang ﹡ , Jiancheng Shi, Youjun Dou Xiaojun Yin, Liying Li, Chenzhou Liu ﹡ [email_address] Experiments of satellite data simulation based on the Community Land Model and SCE-UA algorithm IGARSS 2011, Vancouver, Canada, 24-29 July, 2011 Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing, 100101, China
- 2. The gridded AMSE-E BT data is the mean state of the whole grid cell and can be regarded as a mixed pixel problem, it is equal to area weighted sum of BT in each sub-pixel: Introduction Land data assimilation provides a framework for taking full advantage of land surface model estimation and various observations to obtain the optimal estimation of land surface variables; It is very important to simulate satellite data （ brightness temperature, BT ） for directly assimilating microwave remote sensing data.
- 3. Land radiative transfer model (RTM) is an important component of land data assimilation system . There are two main problems for RTM : Problems The uncertainties from RTM parameters The uncertainties influence the accuracy of satellite data simulation and land data assimilation There’s not a land RTM to calculate the microwave wetland surface emissivity If we take the wetland patch in a model grid cell as a water surface in the process of the BT simulation, the difference between the simulated and observed BT is very obvious, which will have some uncertain effects on the data assimilation result. How to calculate the microwave wetland surface emissivity ?
- 4. To develop a dual-phase satellite data simulation system to simulate the gridded AMSR-E BT data and calibrate the microwave wetland surface emissivity b ased on the Community Land Model (CLM) , microwave land emissivity model (LandEM) , Shuffled Complex Evolution (SCE-UA) algorithm and AMSR-E BT data, which is an important component of soil moisture data assimilation s ystem. Objective
- 5. Methodology: Satellite data simulation system Flowchart of the satellite data simulation system
- 6. Methodology: Satellite data simulation system The satellite data simulation system uses the outputs of CLM as the inputs of LandEM to simulate the AMSR-E BT; It is implemented in two phases: the parameter optimization and calibration phase and the AMSR-E BT simulation phase; Finally, use the optimal LandEM parameters and the calibrated microwave wetland surface emissivity to simulate the AMSR-E BT.
- 7. Provide inputs ( near-surface soil moisture, ground temperature, canopy temperature and snow depth) for the LandEM; The CLM has been developed by combining the best features of three commonly used land surface models (NCAR LSM, BATS and IAP94). Although the CLM is a single-column model, it considers the sub-grid scale heterogeneity by subdividing each grid cell into a number of sub-grid fractions (Bonan et al . ,2002; Dai et al . ,2003 and Oleson et al . ,2004) . Methodology ： Community Land Model
- 8. Use the LandEM to simulate BT; The LandEM only considers a three-layer medium. The top and bottom layers are considered spatially homogeneous and are represented by uniform dielectric constants. Conversely, the middle layer is snow grains, sand particles, and vegetation canopy. For bare soil surface, the three-layer model may be regarded as a two-layer model (Weng et al . 2001). Methodology ： Microwave land emissivity model
- 9. The SCE-UA algorithm is used to search for the optimal values of the LandEM parameters (surface roughness, radius of dense medium scatterers, fraction volume of dense medium scatterers, leaf thickness) and microwave wetland surface emissivity in their feasible space by minimizing the objective function; The SCE-UA algorithm does not require an explicit expression or the partial derivative for the objective function and can automatically calibrate the model parameters (Duan et al . ,1993, 1994) . Methodology ： SCE-UA algorithm
- 10. Methodology ： Parameters calibration scheme Objective function ： If there is wetland in grid, the BT of grid denotes as following: : microwave wetland surface emissivity : effective temperature : area fraction of wetland and ： simulated BT and ： observed BT ： the number of satellite observations during calibration using SCE-UA algorithm
- 11. The AMSR-E/Aqua daily quarter-degree gridded BT data used in this study was downloaded from the National Snow and Ice Data Center (NSIDC) (Knowles et al . , 2006) (http://nsidc.org/data/docs/daac/nsidc0301_amsre_gridded _tb.gd.html). Experiment - Data
- 12. Experiment: Reference stations information 7% wetland 13% C 4 grass 13% C 3 non-arctic grass 19.5% needleleaf evergreen temperate tree 47.5% corn (24.80ºN, 113.58ºE) ShaoGuan 11% wetland 0.9% C 3 non-arctic grass 0.9% needleleaf deciduous boreal tree 0.9% needleleaf evergreen temperate tree 86.3% corn (44.42ºN, 122.87ºE) TongYu 86% wetland 0.3% broadleaf deciduous temperate shrub 13.7% corn (31.87ºN, 117.23ºE) HeFei Area Fraction Sub-grid Patch Type Location Station
- 13. Time series of the BT simulated by the LandEM in each sub-grid patch and observed by AMSR-E sensor based on the model grid cell at HeFei station. The difference between two sub-grid vegetation patch BT and the wetland patch is extremely evident, the main cause is that there is more water surface in the wetland patch. Results - Sub-grid patch BT
- 14. Time series of the emissivities simulated by the landEM in two sub-grid vegetation patch and calibrated by the SCE-UA algorithm in the sub-grid wetland patch (monthly mean) at HeFei station Results - Calibrated wetland surface emissivity
- 15. Scatterplots of the AMSR-E BT simulated by the LandEM (left) and simulated by the parameters transfer (right) versus that observed by AMSR-E sensor in 2003 at TongYu Results - Parameters transfer validation The monthly mean microwave wetland emissivities calibrated at Hefei in 2003 were transferred to TongYu.
- 16. Application: Soil moisture assimilation Develop a soil moisture data assimilation system to directly assimilate the gridded AMSR-E BT data, which consists of the CLM, LandEM and ensemble Kalman filter (EnKF) ; The monthly mean microwave wetland emissivities calibrated at Hefei in 2003 were transferred to Shaoguan.
- 17. Application: Soil moisture assimilation result Comparisons of the daily volumetric soil moisture content among the simulation, assimilation with the AMSR-E BT data and observation in different soil layers (0-50 cm) at ShaoGuan from 19 June to 31 December 2002
- 18. Develop a dual-phase satellite data simulation system, which was implemented in two phases: the parameter optimization and calibration phase and the AMSR-E BT simulation phase; The SCE-UA algorithm can effectively calibrate the LandEM parameters and microwave wetland surface emissivity, and which possess excellent transportability; The s oil moisture assimilation experiment shows that the dual-phase satellite data simulation scheme is reasonable . Conclusions
- 19. Future perspectives Perform region validation experiments; Develop a wetland surface emissivity model with physical mechanism.
- 20. Thank You!