基于土壤物理基本参数的土壤导热率模型

土壤物理基本参数是影响土壤导热率的重要因素,为了获取土壤的颗粒组成、有机质含量与土壤导热率计算模型中参数之间的关系,该文分析了陕西省9个地区的土壤质地对土壤导热率的影响,对不同土壤导热率估算模型的准确性进行评价,并在C?té-Konrad模型和Lu-Ren模型的基础上,建立了基于土壤物理基本参数的改进模型,结果表明:改进的C?té-Konrad模型与改进的Lu-Ren模型可以用来拟合不同质地的土壤导热率,且具有较好的拟合精度,决定系数R2均在0.92以上,相对误差(relative error,Re)均低于9.6%;对于砂粒含量或粉粒含量较高的土壤导热率,改进的C?té-Konrad模型模拟结果的均方根误差(root-mean-square error,RMSE)≤0.1183、R2≥0.9259以及Re≤9.55%,均优于C?té-Konrad模型、Lu-Ren模型和改进Lu-Ren模型;对于砂粒和粉粒含量均较低的土壤导热率,改进Lu-Ren模型模拟结果的RMSE≤0.0815、R2≥0.9326,Re≤8.21%,均明显优于其他3种模型。两种改进的模型分别建立了模型参数与颗粒组成、有机质含量之间的关系,能够更加详细描述土壤物理基本参数与导热率之间的关系,并且针对不同的土壤质地,选取合适的改进模型能够更加准确地计算土壤导热率。

Soil thermal conductivity model based on soil physical basic parameters

Abstract: Soil physical basic parameters are key factors for impacting the soil thermal conductivity, and they are also closely related to the model parameters used to calculate the soil thermal conductivity. In order to study the relationship between soil physical basic parameters, organic matter content and soil thermal conductivity model parameters, the precision of different soil thermal conductivity models was discussed by analyzing 10 types of soil samples in this paper. There were 9 types of soil textures which were sampled from different areas in Shaanxi Province, and the last one was sampled from Zhangye, Gansu Province, which was used to verify the feasibility of the new models. According to the sand content, these soil samples were divided into 2 types: fine-textured soil and coarse-textured soil. The soil thermal conductivity models were used to fit these 2 types of soils, and the comparison results indicated that the theoretical models such as C?té-Konrad model and Lu-Ren model were more precise than Campbell model and Johansen model. The fitted results of Johansen model were significantly smaller than the measured values, and the ranges of root mean square error (RMSE), coefficient of determination (R2) and relative error (Re) for this model were 0.0848-0.2548, 0.656-0.827 and 10.32%-20.41%, respectively. Moreover, C?té-Konrad model and Lu-Ren model had better fitting results for fine-textured soil, and the ranges of RMSE, R2 and Re were 0.0810-0.1208, 0.842-0.940 and 9.67%-10.57% for C?té-Konrad model and 0.0725-0.1238, 0.874-0.937 and 8.28%-9.91% for Lu-Ren model. However, these 2 models were not suitable for calculating the soil thermal conductivity of coarse-textured soil when the water saturation was larger than 50%. Thus, the improved models, which described the relationship between thermal conductivity and soil physical basic parameters, were developed based on C?té-Konrad model and Lu-Ren model. The results showed that the improved models could be used to fit the thermal conductivity in different soil textures, and the RMSE was less than 0.0964, the R2 was up to 0.92 and the Re was less than 9.6%. For predicting the soil thermal conductivity with higher sand content or higher silt content, the values of RMSE, R2 and Re for the improved C?té-Konrad model were 0.1183, 0.9259 and 9.47%, respectively, which was better than the C?té-Konrad model, Lu-Ren model, and improved Lu-Ren model through the analysis of simulation error. On the other hand, for predicting the soil thermal conductivity with lower sand and silt contents, the values of RMSE, R2 and Re for the improved Lu-Ren model were 0.0815, 0.9326 and 8.11%, respectively, which was better than the other 3 models. Moreover, the improved models were used to calculate the soil thermal conductivity of the other types of soil textures in Zhangye. Because the soil texture in Zhangye is sandy clay loam soil in which the sand content is higher than 60%, the improved C?té-Konrad model has the best effect when calculating the soil thermal conductivity according to the analysis results. The parameters in the improved models contain soil texture and organic matter content, which can be used to describe the relationship between thermal conductivity and soil physical basic parameters in detail. Furthermore, choosing an appropriate improved model based on soil texture can calculate the soil thermal conductivity more accurately.