基于多重网格求解的土壤属性高精度曲面建模

High accuracy surface modeling of soil properties based on multi-grid

高精度曲面建模(HASM)是近几年发展起来的可用于地理信息系统和生态建模的一种较高精度的曲面建模方法。本研究选择南方红壤丘陵区江西省吉安市辖区、吉安县和泰和县为研究区,采集150个表层土壤(0～20cm)样品,分别随机选取60、90和120个点作为模拟数据集,90、60和30个点作为验证数据集,基于多重网格(MG)作为土壤属性高精度曲面建模的求解方法(HASM-MG),比较HASM-MG、Kriging、IDW和Spline四种方法的插值精度,检验HASM-MG在土壤pH的插值效果,并分析土壤pH在典型南方红壤丘陵区的空间分异规律。结果表明:HASM-MG的模拟结果与测定值最为接近;与其他三种方法相比,其模拟结果在三组数据集中均具有最小的平均绝对误差、均方根误差和与测定值之间最高的相关系数;研究区水稻土分布较多的中部和东南部地区的土壤pH明显高于红壤分布较多的地区。多重网格求解的高精度曲面建模(HASM-MG)在提高了求解速度的前提下,保证了土壤属性的插值精度,将土壤属性的高精度曲面建模向实用性方面向前推进了一步。

Kriging,inverse distance weighting（IDW） and spline are three classical methods for interpolating soil properties.High accuracy surface modeling（HASM） has been developed in recent years,which is successfully used in geographic information system and ecological modeling.In order to assess its feasibility of soil property interpolation,HASM based on multi-grid（HASM-MG） was used to interpolate soil pH for a red soil hilly region of Jiangxi Province,China.A total of 150 samples of topsoil（0—20 cm） were collected in Ji＇an municipal district,Ji＇an County and Taihe County in October,2007.Three different data sets were randomly created for simulation and validation to comparatively simulate performance of HASM,kriging,IDW and spline.A total of 30,60 and 90 samples were removed as validation points,thus 120,90 and 60 data points were selected as simulation points,respectively.The results show that HASM simulation is the closest to the measured data among the four methods.It has much smaller mean absolute error,root mean square error and much higher correlation coefficient of simulated value and measured value than the three classical methods.Soil pH values in the central and southeastern parts of the study area dominated by paddy soils are much higher than those in the red soil area.Therefore,HASM-MG can be considered as an accurate method for interpolating soil properties on the premise of improving the calculation speed,and it advances the practicability of HASM applied in soil property interpolation.