| Accurate crop growth monitoring and yield prediction is very important to secure food supply and sustain agricultural development. Crop models can be forceful tools for monitoring status of crop growth and predicting yield over homogeneous areas. However, their application to a larger spatial domain is hampered by lack of sufficient spatial information of model inputs, such as parameter values and initial conditions, which may have great difference between regions even between fields. Application of of remote sensing data helps to overcome this problem. By incorporating remote sensing data with the crop model WOFOST (for example through LAI), it is possible to use remote sensing variables (for example vegetation index) for each pixel of the spatial domain, and to reestimate new values of the parameters or initial conditions, to which the model is particularly sensitive.Based on medification of crop model WOFOST, a winter wheat growth model was applied in Yucheng region of Shandong Province in the North China Plain. Combination method of remote sensing information with crop model in water stress production level was studied. Through coupling remote sensing information, crop model was optimized by reestimating its parameters and initial conditions. A new method of regional remote sensing combining crop model was established and its application was studied. This method has highly potential application in crop growth monitoring and yield forecasting. The main outcomes in this study are as follows:(1) In order to improve the simulation accuracy of WOFOST model about the influences of water stress on crop growth, some modules had been modified: (a) Using the Penman-Monteith equation (1998) recommended by the FAO to replace the Penman formula (1948) in original WOFOST; (b) Changing crop coefficients according to the growth stages, which are fixed to 1 in original WOFOST. Based on the biological significance and sensitivity of crop growth parameters of the crop model WOFOST, using the FSEOPT procedures or "trial and error" method, some of crop parameters, such as specific leaf area, leaf senescence index, partitioning coefficients, maximum photosynthetic rate were adjusted. The WOFOST model were calibrated and validated against field experimental data. The simulation result of crop model fitted field data well. The results showed that after the two improvement of the original WOFOST source code and the adjustments of main crop and soil parameters, the evapotranspirati0n of winter wheat simulated by the adjusted WOFOST model has been improved; the LAI and aboveground biomass simulated by the adjusted WOFOST model fitted for the measured data consistently. It was proved that the modified WOFOST model can be used to simulate the winter wheat growth, development, and yield formation under water limited condition in regional scale.(2) The regional daily ET model, based on remotely sensed and weather station data, was established to estimate regional surface moisture condition,regional latent heat flux (λΕ) and regional daily ET. The model applied two methods, "Universal triangle" method and "Actual triangle" method, to estimate the surface moisture index which is called surface Temperature-Vegetation Cover Index (TVCI). The "Universal triangle" method was estimated TVCl according to Carlson et al. (1995). Using a trapezoid (triangle) correlation between surface temperature and fractional vegetation cover, we constructed an improved 'Actual triangle' method to estimate TVCl and coupling the Penman-Monteith equation (1998) to estimate daily ET. Daily ET based on the 'Actual triangle' methods was compared well with that by the 'soil water lost method', while daily ET based on the 'Universal triangle' methods was underestimated. So, it is suitable to use 'Actual triangle' method to estimate TVCl instead of 'Universal triangle' method in the North China Plain even if the method was applied under different climate conditions. This study aims to establish an applicable algorithm of ET calculation. Comparing with other regional estimating ET methods using remote sensing, this method is simple and practical, not requiring synchronous meteorological observation data, and remote sensing data. Thus, this greatly enhances practicality.(3) This study presented the special views to develop a reasonable, practical methods, 'forward deducing + pixel model" through inversion LAI coupling classification to combine remote sensing information and crop simulation model, the integration superiority of remote sensing and advantage of crop models, crop simulation model can be applied in the region using this method, it has considerable scientific significance and high practical value. By study and review of many LAI inversion methods, through validation and comparison, a new LAI inversion method was developed on the basis of the revised coefficient method, the revised LAI can be suitable for the study area LAI distribution. Base on study the methods of remote sensing information combination crop simulation model, there are four combination ways summarized and proposed in this paper, they are 'forward deducing + pixel mode', 'forward deducing + lattice model', 'inversion deducing + pixel mode','inversion deducing + lattice model'. The advantages and disadvantages of the various methods were compared and analysised in this paper. By using the 'forward deducing + pixel model', through inversion LAI coupling classification method to combine remote sensing information and crop simulation model, to simulate winter wheat growth and development on the pixel scale over Yucheng region in 2000-2001, the regional distribution of winter wheat was given out, the results of acreage and yield are very close to the statistics given by the government statistical results.(4) Methodology which crop growth, development and yield formation in regional scale simulated by combining remotely-sensed information with crop model was studied and some good results were approached. However,because of complex interdiscipline involving agronomy, geoscience and remote sensing techniques and lack of data, some problem, such as analysis of accumulation error, further improvement of crop models, adjustment of parameters and match of spatial scale in remote sensing data with crop model, need to be studied in future.
|
PDF Full Text Request