中文摘要：

基于电极法的土壤硝态氮检测中，共存氯离子、温度变化、土壤水分是影响检测精度的重要因素。为消除干扰因素影响，进行测量模型的研究，基于电极检测原理与最小二乘法多元线性回归进行建模的理论分析，并研究考虑土壤水分影响后湿土直测的校正方法；开展建模集溶液检测实验，探讨各干扰因素对检测结果的影响，并根据建模集64个观测值分别建立25℃的一元线性回归模型（基础模型）、25℃的多元线性回归模型（硝酸根与氯离子浓度同时测量）和5～30℃的多元温度校正模型（温度变化较大时硝酸根与氯离子浓度同时测量）；开展验证集溶液检测实验，验证并比较3种模型的可适用性；并开展不同含水率的湿土直测实验，验证湿土直测校正公式的准确性。结果表明，在温度变化、氯离子共存的条件下，25％的多元线性回归模型效果最佳（硝酸根离子与氯离子浓度的测量误差分别在-8．37％和-12．03％内），满足多组分现场速测的精度要求；用湿土直测校正公式代替繁琐费时的土壤前处理，可有效减小土壤水分引起的误差。因此，利用多元线性回归模型结合湿土直测校正公式进行电极法土壤硝态氮的检测，可减小干扰因素影响、有效提高现场检测的时效性与准确性。

英文摘要：

In the ion selective electrode （ISE） based soil nitrate-nitrogen （NO3^--N） detection, coexisting chloridion （Cl^-）, changing of temperature and soil moisture are the primary interference factors. Aiming at improving the accuracy and timeliness of NO3^--N detection on site, measurement models for eliminating interference factor＇s effects were studied. First, theoretical analysis of modeling and fresh field soil immediate detection was carried on. Modeling theoretical analysis was based on ISE detection principle and multivariate calibration in chemometrics. And to improve timeliness of soil NO3^--N detection on site, how to detect fresh field soil immediately and calculate NO3 -N content were discussed. Second, two groups of experiments were conducted. One was ISE potential signals collection of mixed solutions with different concentrations （NO3^- and Cl^- ） at various temperatures measured by detection instrument. According to modeling dataset of ISE potential signals, interference factor＇ s effects were discussed and three kinds of measurement models were built, including unitary linear regression model of 25℃ （base model）, multivariate linear regression model of 25℃ （simultaneous detection of NO3^- and Cl^- ） , and multivariate temperature correction model of 5 - 30℃ （ simultaneous detection of NO3^- and Cl^- at changing temperature）. And the accuracy of three models were verified and compared according to validation dataset. The other one was NO3^--N content detection of fresh field soil measured by corrected and uncorrected NO3^--N content conservation equation. The results showed that when Cl^- existed and environment temperature changed, multivariate linear regression model of 25℃ was the best （relative error of NO/ and Cl^- was respectively less than - 8.37% and - 12.03% ）. Besides, corrected soil NO3^--N content conservation equation can reduce the error caused by soil moisture.