中国农学通报
中國農學通報
중국농학통보
CHINESE AGRICULTURAL SCIENCE BULLETIN
2013年
11期
139-144
,共6页
土壤有机碳(SOC)%空间预测%普通克里格(OK)%结合土壤信息的克里格(STK)%江苏省沛县
土壤有機碳(SOC)%空間預測%普通剋裏格(OK)%結閤土壤信息的剋裏格(STK)%江囌省沛縣
토양유궤탄(SOC)%공간예측%보통극리격(OK)%결합토양신식적극리격(STK)%강소성패현
soil organic carbon (SOC)%spatial predicting%ordinary kriging (OK)%kriging combined with soil type information (STK)%Pei County of Jiangsu
为了探讨土壤类型作为辅助信息提高潮土区土壤有机碳(SOC)空间预测精度的可行性,基于江苏省沛县212个土壤预测样点和80个验证点,比较分析了普通克里格(OK)和结合土壤类型信息的克里格(STK)2种方法对土壤有机碳的空间预测精度。结果表明,各土壤类型间的SOC含量存在较大差异,其中淤土的含量最高(16.35 g/kg),而沙土含量最低(8.58 g/kg),两合土和轻盐碱土介于两者之间;直接利用OK方法对区域SOC预测时的平滑效应较强,导致误差较大,其80个验证点的真实与预测值的散点图相关系数r为0.45,均方根误差(RMSE)为7.36 g/kg;而STK相应的相关系数r为0.75,RMSE为4.90 g/kg,较前者降低了33.5%;同时,STK预测图斑能较OK更好地反映SOC在各土壤类型间的变异特征。说明STK方法消除了土壤类型间SOC含量差异较大的影响,降低了克里格平滑效应,从而较大幅度地提高了预测精度。表明与SOC关系密切的土壤类型信息,可在潮土区作为辅助信息以提高区域SOC的空间预测精度。
為瞭探討土壤類型作為輔助信息提高潮土區土壤有機碳(SOC)空間預測精度的可行性,基于江囌省沛縣212箇土壤預測樣點和80箇驗證點,比較分析瞭普通剋裏格(OK)和結閤土壤類型信息的剋裏格(STK)2種方法對土壤有機碳的空間預測精度。結果錶明,各土壤類型間的SOC含量存在較大差異,其中淤土的含量最高(16.35 g/kg),而沙土含量最低(8.58 g/kg),兩閤土和輕鹽堿土介于兩者之間;直接利用OK方法對區域SOC預測時的平滑效應較彊,導緻誤差較大,其80箇驗證點的真實與預測值的散點圖相關繫數r為0.45,均方根誤差(RMSE)為7.36 g/kg;而STK相應的相關繫數r為0.75,RMSE為4.90 g/kg,較前者降低瞭33.5%;同時,STK預測圖斑能較OK更好地反映SOC在各土壤類型間的變異特徵。說明STK方法消除瞭土壤類型間SOC含量差異較大的影響,降低瞭剋裏格平滑效應,從而較大幅度地提高瞭預測精度。錶明與SOC關繫密切的土壤類型信息,可在潮土區作為輔助信息以提高區域SOC的空間預測精度。
위료탐토토양류형작위보조신식제고조토구토양유궤탄(SOC)공간예측정도적가행성,기우강소성패현212개토양예측양점화80개험증점,비교분석료보통극리격(OK)화결합토양류형신식적극리격(STK)2충방법대토양유궤탄적공간예측정도。결과표명,각토양류형간적SOC함량존재교대차이,기중어토적함량최고(16.35 g/kg),이사토함량최저(8.58 g/kg),량합토화경염감토개우량자지간;직접이용OK방법대구역SOC예측시적평활효응교강,도치오차교대,기80개험증점적진실여예측치적산점도상관계수r위0.45,균방근오차(RMSE)위7.36 g/kg;이STK상응적상관계수r위0.75,RMSE위4.90 g/kg,교전자강저료33.5%;동시,STK예측도반능교OK경호지반영SOC재각토양류형간적변이특정。설명STK방법소제료토양류형간SOC함량차이교대적영향,강저료극리격평활효응,종이교대폭도지제고료예측정도。표명여SOC관계밀절적토양류형신식,가재조토구작위보조신식이제고구역SOC적공간예측정도。
In order to evaluate the feasibility of soil types being used as auxiliary information to improve the spatial prediction accuracy of soil organic carbon in fluvo-aquic soil area, 212 prediction samples and 80 validation samples were collected throughout the Pei County, and the spatial prediction accuracy of the two methods of ordinary kriging (OK) and the kriging combined with soil types information (STK) were compared for soil organic carbon (SOC) content. The results showed that there were great differences for SOC content among the main soil types. The content was highest in silted soil (16.35 g/kg), while the one was lowest in sandy soil (8.58 g/kg). The ones of mixed soil and light saline-alkali soil were between them. The prediction accuracy using OK directly was low due to the great smoothing effect, and the corresponding correlation coefficient and the RMSE were 0.45 and 8.31 g/kg, while the ones of STK was 0.75 and 6.02 g/kg, and the RMSE was 20% lower than the former. Furthermore, the SOC spatial distribution from STK reflected the difference of SOC content among soil types better and was closer to the regional reality than the one from OK. It was indicated that STK could reduce the smoothing effect and improve the prediction accuracy by eliminating the difference of SOC contents among soil types. The information of soil type, which was closely related to SOC content, could be used as auxiliary information to improve prediction accuracy of SOC in the alluvial soil areas of North Jiangsu.