【摘要】 准确的土壤盐分空间变异信息是防治土壤盐碱化和制定科学种植规划的重要依据。本文根据电磁型近地传感器EM38测定的土壤表征电导率(EC_a),研究有限最大似然法(REML)在地统计建模趋势去除中的应用,基于线性混合模型将趋势与残差分离,然后利用泛克立格开展海涂围垦区水稻田土壤盐分的空间变异性研究。结果表明,本研究区土壤盐分存在强烈的空间变异性且呈一定的空间分布趋势;REML方法拟合的二次多项式模型可较好的表征土壤盐分的空间分布趋势,在去除趋势的同时,选用指数模型对残差的半方差函数进行拟合;基于泛克立格插值的交叉检验结果可见,该方法可得到较小的预测误差,且标准离差比平均值参数接近理想值,REML方法大大提高了模型的精度;就地统计空间插值结果可看出,土壤盐分呈出从西北角向东北角、从周边向中心渐变的空间分布特征,且与土壤排水条件表现出较好的耦合性。更多还原

【Abstract】 Accurate map of soil salinity is great important for controlling soil salinization and making rational decisions for management. The apparent electrical conductivity(EC_a) was measured by the Geonics EM38 sensor at 192 points in a paddy rice field in reclaimed coastal saline land in Zhejiang Province, China. We discovered a strong spatial trend in the data, which we treated as a quadratic trend surface with correlated random residuals, evaluated the coefficients of the trend and the parameters of the covariance of the residuals by restricted maximum likelihood(REML), and then kriged the EC_a on to a fine grid for mapping by universal kriging, using the estimated covariance parameters and taking into account the form of the trend. All combinations of models and kriging methods gave small mean errors(ME), as expected kriging was unbiased. The mean squared deviation ratios(MSDR), which ideally should equal 1,varied more. The combination with an MSDR closest to 1 was UK with the exponential variogram estimated by REML;its MSDR was 1.03. We used this combination to map the predicted EC_a and associated errors. The horizontal distribution of salinization was characteristic of higher salinity in the northwestward side than the southeast side, or higher in the central part than the edge part. Besides, there existed coupling effect between the soil salinity and the drainage condition.更多还原