山东农业科学
山東農業科學
산동농업과학
SHANGDONG AGRICULTURAL SCIENCES
2015年
2期
119-124
,共6页
BP神经网络%土壤盐渍化%Landsat8%黄河三角洲
BP神經網絡%土壤鹽漬化%Landsat8%黃河三角洲
BP신경망락%토양염지화%Landsat8%황하삼각주
BP neural network%Soil stalinization%Landsat8%The Yellow River Delta
土壤盐渍化是最常见的土壤退化过程,垦利县位于黄河三角洲核心地带,是土壤盐渍化比较典型的区域。本研究在盐渍土野外调查采样的基础上,依据土壤理化分析和Landsat8卫星光谱数据,选取了相关性以及诊断指数较好的3个波段的反射率作为盐分反演因子,分别建立数理统计模型与BP神经网络盐分反演模型。研究表明:BP神经网络模型的精度明显优于传统多元回归模型,且反演模型更适合高盐度区域土壤盐渍化反演制图,具有较好的应用前景。
土壤鹽漬化是最常見的土壤退化過程,墾利縣位于黃河三角洲覈心地帶,是土壤鹽漬化比較典型的區域。本研究在鹽漬土野外調查採樣的基礎上,依據土壤理化分析和Landsat8衛星光譜數據,選取瞭相關性以及診斷指數較好的3箇波段的反射率作為鹽分反縯因子,分彆建立數理統計模型與BP神經網絡鹽分反縯模型。研究錶明:BP神經網絡模型的精度明顯優于傳統多元迴歸模型,且反縯模型更適閤高鹽度區域土壤鹽漬化反縯製圖,具有較好的應用前景。
토양염지화시최상견적토양퇴화과정,은리현위우황하삼각주핵심지대,시토양염지화비교전형적구역。본연구재염지토야외조사채양적기출상,의거토양이화분석화Landsat8위성광보수거,선취료상관성이급진단지수교호적3개파단적반사솔작위염분반연인자,분별건립수리통계모형여BP신경망락염분반연모형。연구표명:BP신경망락모형적정도명현우우전통다원회귀모형,차반연모형경괄합고염도구역토양염지화반연제도,구유교호적응용전경。
Soil salinization is one of the most common land degradation processes .Kenli County is loca-ted in the core area of the Yellow River Delta , where the soil salinization is typical .Based on investigating and sampling saline soil in field , and according to the physical and chemical analysis of soil and Landsat 8 satellite spectral data , the reflectivities of 3 wave bands with better correlation and diagnostic indexes were choosed as the salinity inversion factors , and then the statistic model and BP neural network model were constructed .The results showed that the precision of BP neural network model was far better than that of the traditional multiple regression model .The inversion model was more suitable for soil salinization inversion mapping in high salinity region.It had a good application prospect .