Rice Growth Monitoring Based on ORYZA2000Model and Remote Sensing Data
|School||Nanjing University of Information Engineering|
|Keywords||ORYZA2000 remote sensing MODIS growth monitoring|
China is the world’s largest rice producer and consumer country, nearly half of China’s total food production is rice. Therefore, rice is the first major food crop. Rice growth monitoring can provide rice growth and yield trends and other important information, which is the important basis of State and Government to develop policies, the implementation of the economic plan. Using remote sensing data can be a long time to obtain a wide range information of vegetation canopy and crop model can simulate the characteristics of single-station crop conditions, the Rice model (ORYZA2000) and remote sensing information coupled in order to achieve the purpose of the monitoring of a large area of rice growth and yield estimation. Firstly, the introduction of the ORYZA2000Rice model based on the2010-2011field trial data and single-station weather data, parameter sensitivity analysis and parameter calibration; using gridded weather data driven model to analysis the study area by the distribution of light, temperature, water, heat affected the growth of the seedlings; extracted growth period data from the agrometeorological site of the study area, while filtering remote sensing data(using S-G method) to eliminate noise, and then based on the empirical formula of LAI to calculate the spatial and temporal distribution of LAI; using the performance characteristics of LAI time series to get the spatial and temporal distribution of the growth period (transplanting, heading and maturity) of the study area; finally, using gridded weather data, fertility period data and LAI data drive model to estimate rice yield.Main content of each part and conclusion is introduced as follows:(1) the results of model parameter sensitivity analysis show that the model parameters of ORYZA2000EMD (sowing time) and TOD (Rice optimum temperature) have significant influence on the simulation results.(2) The comprehensive evaluation of the accuracy of ORYZA2000is higher than WOFOST in Jiangxi single-station test, especially in the simulation of spike biomass and leaf area index.(3) The simulation value of the length of the growing period in Anhui by ORYZA2000model is less than the measured value for2-7d underestimation. The NRMSE of the length of the growing period is3.4%to7.5%, the NRMSE of the two sets of data from2010and2011indicators of total aboveground biomass is16%to22%, green leaf biomass and stem biomass is20%to25%,17%to21%, spike biomass is19%to25%, leaf area index is24%to26%. The NRMSE of the total biomass and yield respectively is6%to13%, and5%to14%.(4) The average simulated value of biomass is close to the average measured value in single-station of Jiangsu Province, and the probability of t test values are greater than0.05, no significant difference. The NRMSE of total aboveground biomass, leaf biomass, stem biomass and spike biomass were9%,19%,18%,13%,25%and16%,25%,17%,19%,24%.(5)Using gridded meteorological data to drive model, ORYZA2000model can be reginonalized. Only limited by the weather conditions, the coastal area of Jiangsu can be classified as the first category area, it provides adequate lighting and rice accumulated temperature needed for rice growth and development; the central region of Jiangsu is a transition zone, hydrothermal conditions affecting seedlings, the area changes from north to south by the first category to the secend category; the southern Jiangsu Province performance for the third category area.(6) Using the method of S-G to filter EVI time-series data, the growth period can be extacted from remote sensing information, by compared with the statistics data, most results of the seedling stage, the panicle initiation stage and mature stage within±16d, transplanting stage and heading stage within±8days (7) In the study of yield estimation, the simulated results of the50sites in Jiangsu Province, the minimum error is1.55%, the maximum error is11.56%, average error is5.17%, simulated value is higher than the measured value and the difference does not exceed10%.