论文部分内容阅读
土壤参数的时空变异是实施精细农业时要考虑的重要因素 ,在土壤检测的栅格采样中有必要确定最优样本量。本研究的试验田是一块生长中的玉米地块 ,试验区的面积为 4.2 m× 4.2 m ,该试验区被假定为土壤采样中的一个栅格 ,该栅格又被细分为 49个 0 .6 m× 0 .6 m的子栅格。采样时 ,所分析的土壤参数为土壤硝态氮含量 ,从播种到收获共进行了 7次采样。通过对土壤样本土壤硝态氮时空变异的分析 ,揭示了样本量和土壤硝态氮含量预测误差之间的相关关系。土壤硝态氮含量呈非正态分布 ,通过对玉米各个生长期获得的数据分析表明 :含量水平的预测误差随深度的增加而增大 ;当从一个栅格只采集 1个土壤样本时 (样本量为 1) ,预测误差基本在 5 0 %左右 (显著水平 :α≈ 0 .10 ) ,而当从一个栅格采集 5个土壤样本时 (样本量为 5 ) ,预测误差将降至 2 5 %左右。另一方面 ,当要求预测误差低于 30 %时 ,对于普通生长条件下的土壤需要从 1个栅格至少采集 3个样本 ,而对于追肥后的土壤则至少需采集 15个样本。
Spatial and temporal variability of soil parameters is an important factor to be considered in the implementation of precision agriculture. It is necessary to determine the optimal sample size in raster sampling of soil testing. The experimental field in this study is a growing maize plot with an area of 4.2 m × 4.2 m. This plot is assumed to be a grid of soil samples that is subdivided into 49 0s. Sub-grid of 6 m × 0.6 m. At the time of sampling, the soil parameters analyzed were soil nitrate nitrogen content, and a total of seven samples were taken from sowing to harvesting. Through the analysis of spatio-temporal variations of nitrate-N in soil samples, the correlation between the sample size and prediction error of soil nitrate-N content was revealed. The results showed that the prediction error of content level increased with the increase of soil depth. When only one soil sample was collected from one grid (sample With a prediction error of about 50% (significant level: α≈0.10), whereas when 5 soil samples were collected from one grid (5 samples), the prediction error was reduced to 25 %about. On the other hand, when the prediction error is less than 30%, at least 3 samples need to be collected from a grid for normal growth conditions and 15 samples for topdressing soil.