Comparison of Three Methods for Soil Fertility Quality Spatial Simulation with Uncertainty Assessment
2013
Qu, Mingkai | Li, Weidong | Zhang, Chuanrong | Zhao, Yongcun | Huang, Biao | Sun, Weixia | Hu, Wenyou
Spatial information of soil fertility quality is crucial for sustainable agriculture management. This study compared three methods based on sequential Gaussian simulation for spatial simulation of soil fertility quality index (SFQI) with uncertainty assessment through a case study in Hanchuan County, China. These methods are different in data utilization and simulation procedure: Method 1 first calculates SFQI sample data from the sample data of soil fertility indicators and then simulates the spatial distribution of SFQI with incorporation of land use information; Method 2 is similar to Method 1 except that land use information is not used; and Method 3 first simulates soil quality indicators using their sample data and then derives each SFQI realization from randomly grouped realizations of different soil quality indicators. Validation showed that Method 1 had the best performance in optimal prediction in terms of correlation coefficients between measured and predicted data, global prediction errors (mean error [ME], mean absolute error [MAEE], and root mean square error [RMSE]) and uncertainty modeling (accuracy plots and goodness statistic), and Method 3 was the worst. Methods 1 and 2 achieved, respectively, relative improvements of 32.27 and 12.72% in optimal prediction over Method 3. Probability maps estimated from simulated realizations using Method l showed that paddy fields generally had high quality grades while the fertility quality grades of dry farmlands were usually low in the study area. It is concluded that Method 1 may be a more effective method for SFQI prediction with uncertainty assessment and Method 2 may be adopted when land use information is unavailable.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por National Agricultural Library