Integration Of Remote Sensing And Crop Growth Model For Regional Low Temperature Impact Monitoring. Early Warning, And Yield Estimation

China has a variety of climate resources for food crop production but also risky. To guarantee food security is essential for the country's social and development stability. Under the background of developments in national agricultural production and remote sensing technologies, promoting the application of remote sensing and information technologies in agricultural disasters monitoring for assessment and early warning will be of great significances. These efforts will improve the agricultural information services, scientific planning, making reasonable structure adjustment, and also economic measures suitable for agriculture production.Remote sensing of agricultural disasters monitoring is a systematic project spanning agricultural sciences, catastrophology, remote sensing sciences, geoinformatics et al. The pointcut of this thesis was to investigate low temperature impact on double cropping area based on remote sensing technologies and crop growth model, in Shaanxi Guanzhong Plain. With the efforts made, a disaster evaluation and regional yield estimation framework had been established based on remote sensing and crop growth model. Research achievements are summarized as below.1. Mapping crop phenology using time series remote sensing dataVariability of annual crop phenology is an indicator for year to year climate variation. To accurately extract crop phenology using remote sensing, this thesis had adopted China's HJ-1 A/B satellite data, which is higher spatiotemporal resolution, for the construction of vegetation time series dataset. Methodologies, processing schematic including signal filtering and daily interpolation had been proposed for constructing the high spatiotemporal resolution time series, afterward analysis and phenology extraction can be performed. The research had been focused on winter wheat and summer corn in Yangling district, phenology detection results agreed with local phenological calendar. During the research, time series smoothing methods, comparision of phenology detection with MODIS NDVI were discussed. Although this thesis had tried to use remotely sensed phenology detection to present low temperature impact on crop growth, however, difficulties demonstrated that regional variation of crop phenology might due to climate factors; humanity factor appears more critical.2. Assessment of low temperature impact using crop growth modelTo compensate the dificiencies of field-based experiments, this thesis had adopted crop growth model to investigate low temperature impact early warning, loss assessment, and adaptation practices. The CERES model in DSSAT cropping system had been employed to perform simulation using winter wheat and summer corn experiment materials during 2008-2013, in Xianyang agrometeorological station. During the research, a multi objective optimization method had been proposed for crop model parameters calibration; global sensitivity analysis had been employed to understand the impact of model input variables to simulation output. Afterward quantitative assessments of wheat/corn yield loss due to low temperature scenarios and sudden drop in temperature during growth stages were performed. Simulation result demonstrated that the possibilities of low temperature impact would cause yield loss for double-cropping in such climate zone; the loss should depend on year to year climate condition and sowing/planting activities. Reduction in accumulated temperature throughout growing season would cause remarkable yield loss if sowing/planting remained unchanged; winter wheat was less vulnerable than summer corn in low temperature scenario. Scenarios of consecutive 5-day temperature drop with 5 and 10? were applied to specific growth stages, summer corn would suffer yield loss in all growth stages; winter wheat is less vulnerable to sudden temperature drop and even unaffected in warmer climate condition. Besides, this thesis had investigated in adaptation strategies which will allow farmers to take advantage of the unexpected climate scenarios so as to maintain the stable production of double cropping. Quantitative yield assessment of sowing/harvesting variation for winter wheat and summer corn had been tested. Results demonstrated that harvesting date is critical for winter wheat while sowing date is critical for summer corn in this climate zone. According to this, suitable adaptation for sowing/harvesting date had been proposed under cool climate, for maintaining the yield level for wheat-corn double cropping.3. Regional application of crop model driven by remotely sensed phenology parametersThe phenology parameters derived from HJ-1 A/B time series data were used to drive crop model to perform regional simulation; regional yield of winter wheat and summer corn in 2011-2013 had been estimated driven by sowing/harvesting date. Under the low temperature scenario, spatial difference of crop yield elucidated the relationship of farming activities and yield formation. Results demonstrated the possibility of yield reduction for wheat-corn cropping due to low termperature impact in such climate zone; yield reduction levels were depending on different climate year and arrangement of farming activities; both winter wheat and summer corn would have yield reduction in cool climate, but winter wheat is less vulnerable than summer corn. Results could help to understand how sowing/harvesting date are affecting yield formation in a region, indicating that late harvesting for winter wheat and early sowing for summer corn were critical for maintaining yield level in a cool climate.4. Assimilating remotely sensed crop growth information with crop model for regional yield estimationRemote sensing observation has real-time and large-scale capabilities, providing actual crop area and phenology information; crop growth model physically describes crop growth development and yield formation under a specific environmental zone and farming practices. Remotely sensed time series LAI data provides crop growth process, hence using data assimilation strategy to integrate observed LAI and simulated LAI would improve the realisity of model simulation. This thesis had established a data assimilation framework suitable for DSSAT crop model; MODIS-NDVI time series were used for phenology mapping then drove crop model to perform regional simulation. By establishing a cost function, the GLASS-LAI data had been employed for data assimilation with model simulated LAI; crop model parameters related to LAI simulation were optimized pixel by pixel. Finally, regional yield estimation was obtained by integrating remote sensing observation and crop growth model.