荒漠-绿洲交错带土壤N、P、K含量的高光谱反演模型

利用新疆奇台县荒漠-绿洲交错带的75个土壤样本,选取土壤可见光-近红外光谱的反射率（R）、光谱反射率倒数之对数（lg（1/R））、光谱反射率一阶导数（FDR）和光谱波段深度（Depth）4个指标,分析了其与土壤N、P、K元素含量的关系,分别建立了反演模型并对其精度进行了检验。结果表明：可见光-近红外反射光谱快速估算荒漠-绿洲交错带土壤N、P、K元素含量的潜力大,其预测精度由高到低的排列顺序为：N>P>K。不同光谱指标反演模型的精度各异,指标Depth和FDR的预测效果明显优于lg（1/R）和R,对N和P元素的拟合效果为：lg（1/R）相似文献   

黑土区土壤有机质和全氮含量遥感反演研究

以东北典型黑土区耕地为研究区,以Sentinel-2A（全球环境与安全监测计划的第二颗卫星,于2015年6月23日发射）影像作为数据源,构建光谱指数,分别采用多元逐步线性回归（Multiple Stepwise Linear Regression, MSLR）和随机森林（Random Forest, RF）算法建立土壤有机质（SOM）和土壤全氮（STN）预测模型,并采用十折交叉验证方法评估模型的性能。研究对比分析了不同气候、土壤类型和地形下土壤有机质和全氮的空间分布差异。研究表明：① 海伦示范区的SOM和STN含量最高,其年均温最低,高程最高,年降水量多,SOM含量升高,其年均温最低,年降水量多,STN含量升高;② 与基于多元逐步线性回归算法建立的SOM和STN预测模型相比,随机森林算法建立的SOM和STN预测模型,有着更高的精度和稳定性;③ 运用RF算法建立的SOM反演模型的R²为0.96,均方根误差为5.49 g/kg,STN反演模型的R²为0.95,均方根误差为0.27 g/kg; ④ 不同示范区统一建立SOM和STN预测模型,有助于提高预测精度,实现跨区域建模与制图。     

基于不同卫星光谱模拟的土壤电导率估算研究

土壤电导率 (Electrical conductivity, EC)是评价土壤盐渍化的重要指标。通过实测新疆艾比湖湿地自然保护区土壤EC及可见光—近红外光谱数据,利用波谱响应技术模拟Landsat 8 OLI、Sentinel 2、Sentinel 3卫星的宽波段数据。构建宽波段模拟数据及其5种预处理后的三维光谱指数 (Three-dimensional spectral index, TDSI),采用梯度提升回归树算法 (Gradient boosting regression tree, GBRT) 建立3种卫星土壤EC估算模型,并比对加入TDSI后模型精度的变化。结果表明：在不同土壤EC条件下,3种卫星具有相似的光谱趋势,均在红、近红外波段附近反射率较高;TDSI与土壤EC相关性基本均在0.4以上,最大程度保留了与土壤EC敏感度高的红、绿、蓝、近红外、短波红外波段信息;GBRT对于土壤EC估算能力表现突出,3种卫星对土壤EC的最佳预测精度R²分别为0.831、0.847、0.903,在加入TDSI后,R²分别提高至0.835、0.857、0.935,综合分析发现,Sentinel 3对土壤EC估算效果最佳 (R²=0.935,均方根误差RMSE=2.986 mS·cm^-1,赤池信息准则AIC=57.500)。通过利用波谱响应技术结合TDSI深度挖掘波段间的协同信息,采用GBRT验证了不同卫星对土壤R²的估算效果,二者相结合可以有效提升模型预测精度,为干旱区土壤盐渍化定量监测与防控提供有利指导。     

开垦对科尔沁沙地土壤团聚体分布及稳定性的影响

选取科尔沁地区不同开垦年限玉米田,以沙质草地为对照,采用湿筛法测定了0—40 cm土壤各粒径团聚体分布状况,以大于0.25 mm水稳性团聚体的含量（W_0.25）、平均重量直径（MWD）、几何平均直径（GMD）、分形维数（D）和土壤可蚀性因子（K）为稳定性评价指标,研究随着开垦年限和土层深度变化下土壤水稳性团聚体的含量变化及稳定性差异。结果表明：（1）开垦对土壤水稳性团聚体含量产生了显著的影响,在不同土层呈现出一致性规律,整体上表现为随开垦年限增加大,团聚体含量呈现“N”型态势;（2）开垦15—20 a土壤团聚体的W_0.25、MWD和GMD值均高于其他开垦年限,D、K则相反,且上层团聚体稳定性优于下层;（3）相关分析表明,大于1 mm粒径的土壤团聚体含量与W_0.25、MWD、GMD极显著正相关（P<0.01）,而与0.5—0.25 mm、<0.25 mm粒径团聚体含量、D和K极显著负相关（P<0.01）。开垦活动提高了科尔沁地区土壤团聚体稳定性,开垦15—20 a的土壤抗侵蚀力较强,大于20 a的开垦可能会导致土地退化,应考虑通过采取保护性耕作措施等方式实现该地区土壤可持续健康发展;大于1 mm的水稳性团聚体含量可以作为评价该区域土壤质量的重要参数。     

基于PLSR的典型沼泽湿地植物叶片性状与光谱模型构建——以三江国家级自然保护区为例

通过植物光学特性测量叶片性状是一种非破坏性的、长期的湿地动态监测方法。选择三江国家级自然保护区多种典型湿地植物为研究对象,探究植物性状与叶片光谱之间的联系。研究表明：叶片氮含量与光谱的模型构建效果最好,模型R²为0.61,均方根误差（RMSE）为2.3862;叶片含水量、叶片磷、可溶性糖、纤维素和木质素含量之和的模型一般,R²在0.38~0.55范围内,RMSE在0.0004~10.7019范围内;淀粉含量拟合效果较差, R²为0.29,RMSE为0.0106。光谱预测重要性的结果表明,可见光与近红外边缘范围内的光谱信息对于叶片含水量、叶片氮含量、叶片磷含量、单位叶面积质量、纤维素和木质素之和、可溶性糖和淀粉的预测具有最高的重要性。     

基于多源遥感和机器学习方法的科尔沁沙地植被覆盖度反演

植被覆盖度是监测生态系统及其功能的关键参数,如何提高大区域植被覆盖度的反演精度,对生态脆弱区环境可持续发展至关重要。本研究基于人工神经网络、支持向量回归和随机森林等机器学习方法,利用无人机、Worldview-2与Landsat 8 OLI遥感数据,对科尔沁沙地植被覆盖度进行多尺度反演。结果表明：随机森林模型比人工神经网络、支持向量回归模型表现佳,可在单元（试验区）、区域（研究区）尺度上较高精度地反演沙地的植被覆盖度,反演值与无人机实测值均在线性水平上呈显著相关（P<0.01）;在单元、区域尺度上,构建的植被覆盖度反演模型测试集R²分别为0.84、0.80,MSE分别为0.0145、0.0370,一致性指数d分别为0.9576、0.8991。利用多源遥感数据和机器学习方法,通过局部区域的高精度反演逐步实现低空间分辨率遥感影像的大区域植被覆盖度反演,不仅可有效提高沙地植被覆盖度的反演精度（R²=0.78,大于0.63）,也为区域生态环境监测与生态系统健康评价提供支持。     

基于多光谱遥感的典型绿洲棉田春季土壤盐分反演及验证

为探索快速提取典型绿洲棉田土壤盐分的有效方法,获取区域尺度的土壤盐渍化特征及空间分布,进而为土壤盐渍化防治提供参考。以新疆兵团农二师31团为研究区域,2019、2021年春季Landsat 8 OLI多光谱影像和野外实测土壤含盐量为数据源,将波段组、光谱指数组和全变量组作为模型输入变量组,采用多元逐步回归（Multiple stepwise regression, MSR）、偏最小二乘回归（Partial least squares regression, PLSR）、极限学习机（Extreme learning machine, ELM）、支持向量机（Support vector machine, SVM）和BP神经网络（Back propagation neural network, BPNN）构建基于3个输入变量组的土壤盐分遥感反演模型,探究输入变量和建模方法对模型精度的影响效果,通过对比确定春季土壤盐分最优反演模型,定量反演地表土壤含盐量。结果表明：（1） 研究区主要为非盐化土和轻度盐化土,总样本变异系数为0.67,呈中等变异性;光谱反射率与土壤盐渍化程度的关系表现为土壤盐渍化越重,光谱反射率越高。（2） 海岸波段（b1）、蓝波段（b2）、绿波段（b3）、红波段（b4）和盐分指数（SI1、SI2、SI3、SI4、S3、S4、S5）均通过显著性检验P<0.01,相关系数均达到0.4以上。（3） 所有模型中,基于全变量组建立的BPNN反演模型精度最高,建模集R²为0.705;验证集R²为0.556。（4） 由反演结果可知,2019、2021年春季耕作区土壤主要为非盐化土,分别占耕作区总面积的55.55%和64.62%,其次为轻度盐化土,分别占44.31%和35.17%;2021年土壤盐渍化程度较2019年有所减轻。     

基于高光谱数据的戈壁地表砾石粒径反演研究

戈壁地表砾石粒径组成特征反映戈壁形成过程信息,且在很大程度上决定戈壁改造利用的难易,是开展戈壁研究的基础和前提。结合高光谱数据的微分变换,遴选出砾石粒径的敏感波段与反演方程,进行戈壁地表砾石粒径反演研究。结果表明：微分变换后的砾石光谱反射率与粒径有较好相关性,相关性最好的波段为908nm、983nm和985nm。其中,对数倒数微分变换之后的反射率与粒径成正相关（R² =0.61）,而一阶微分、平方根微分、对数微分3种变换形式之后的反射率与粒径呈负相关,相关系数分别为-0.633、-0.646、-0.649。将一阶微分变换后的光谱数据与粒径进行回归分析,发现一元三次回归模型具有较好的拟合精度,其中对数微分在回归分析中表现最好（R² =0.851）,经过验证得出对数微分预测精度（75.27%）高于其他4种微分形式的精度,表明砾石光谱的对数微分变换之后的908nm波段可应用于戈壁地表砾石粒径的反演。     

新疆焉耆盆地LAI反演及空间分布特征

叶面积指数（LAI）是生态系统物质和能量循环过程的重要结构参数。研究植被LAI对监测估算生态环境极其脆弱的内陆干旱区具有理论和实践价值。使用LAI-2000植物冠层分析仪观测焉耆盆地植被LAI,并结合Landsat OLI的NDVI、SAVI、NDMI、NBR、NBR2、MSAVI和EVI等植被指数数据建立了LAI的反演模型,探求焉耆盆地LAI的空间分布特征及规律。结果表明：（1）各植被指数与观测的LAI均有明显的相关性。其中,SAVI与LAI的对数函数模型更好地模拟出研究区的LAI（R²=0.84）;通过LAI-SAVI模型反演研究区LAI的空间分布,估算精度可达89%（R²=0.82,RMSE=0.23）;（2）博斯腾湖小湖湿地、博斯腾湖大湖西北端的黄水沟入湖沼泽区以及焉耆盆地北部冲洪积扇平原绿洲区的LAI值较高,达3.85;焉耆盆地山区和平原区之间的过渡带和博斯腾湖南部的沙漠区LAI较低;（3）焉耆盆地LAI以博斯腾湖为中心环状分布,LAI随着研究区域与湖、河流距离的增加而逐渐降低的趋势。     

MODIS和Landsat时空融合影像在土壤盐渍化监测中的适用性研究——以渭干河—库车河三角洲绿洲为例

土壤盐渍化的遥感监测依赖于高时空分辨率影像,但受经费预算、卫星回访周期及天气的影响,高时空分辨率的遥感影像较难获取,这就限制了根据采样时间来获取对应时期遥感影像进行土壤盐渍化监测反演的应用。为此,提出融合MODIS和Landsat影像生成高时空分辨率影像来提取土壤盐渍化信息,为时空影像进行土壤盐渍化监测研究提供数据参考。以渭干河—库车河绿洲为研究区,利用增强型时空自适应融合率反射模型（Enhanced spatial and temporal adaptive reflectance fusion model,ESTARFM）和灵活的时空融合模型（Flexible spatiotemporal data fusion,FSDAF）,对MODIS和Landsat影像进行时空融合,并基于融合影像数据构建了关于土壤电导率（EC）的随机森林（RF）预测模型,对比分析时空融合影像应用于土壤盐渍化遥感监测的适用性。结果表明：ESTARFM融合影像的特征波段反射率与Landsat验证影像对应波段反射率一致性决定系数R²（Red）=0.8066、R²（SWIR2）=0.8444;FSDAF融合影像的特征波段与Landsat验证影像对应波段反射率一致性决定系数R²（Red）=0.6999、R²（SWIR2）=0.7493;基于ESTARFM融合影像构建的EC值预测模型精度最高,R²=0.9268,基于FSDAF融合影像构建的EC值预测模型精度良好,R²=0.8987,基于验证影像构建的EC值预测模型R²=0.9103; ESTARFM模型的融合精度高于FSDAF模型,并且基于融合影像构建的EC值预测模型效果良好。     

植被恢复方式对黄土丘陵区土壤理化性质及微生物特性的影响

对甘肃省定西市黄土丘陵区不同植被恢复方式下的土壤水分、pH、有机碳（SOC）、全氮（TN）、碱解氮（AN）、全磷（TP）、速效磷（AP）、速效钾（AK）及微生物生物量、呼吸速率、呼吸熵（qCO₂）以及微生物熵（q_MB）进行了测定与分析。结果表明：不同植被恢复方式下,土壤pH值变化差异不显著（p>0.05）;0~20 cm土SOC、TN、AN、TP、AP、微生物量碳（SMBC）、微生物量氮（SMBN）含量及微生物呼吸速率（MR）均为天然植被（CL）下最高,AK为自然恢复（Q）方式下最高,qCO₂与q_MB在人工重建+自然恢复（RZ）方式下最高;20~40 cm土壤的SOC、TN、AN、TP、AP含量变化同0~20 cm,AK、SMBC、SMBN、 q_MB及MR在RZ方式下最高;40~60 cm土壤的SOC、TN、AN、AP及TP含量在CL方式下最高,SMBC、SMBN、 q_MB及MR在RZ方式下最高,AK、 qCO₂在Q方式下最高。相同植被恢复方式下,随土层深度增加,SOC、TN、AN、TP、AP、AK、SMBC、SMBN含量及q_MB、MR均逐渐降低,但qCO₂逐渐升高（除RZ方式）。不同植被恢复方式,SOC、TN、AN、SMBC、SMBN、q_MB、MR及qCO₂之间存在不同程度的相关性。     

Changes in soil organic carbon and other physical soil properties along adjacent Mediterranean forest, grassland, and cropland ecosystems in Turkey

Cultivation, overgrazing, and overharvesting are seriously degrading forest and grassland ecosystems in the Taurus Mountains of the southern Mediterranean region of Turkey. This study investigated the effects of changes on soil organic carbon (SOC) content and other physical soil properties over a 12-year period in three adjacent ecosystems in a Mediterranean plateau. The ecosystems were cropland (converted from grasslands in 1990), open forest, and grassland. Soil samples from two depths, 0–10 and 10–20 cm, were collected for chemical and physical analyses at each of cropland, open forest, and grassland ecosystems. SOC pools at the 0–20 cm depth of cropland, forest, and grassland ecosystems were estimated at 32,636, 56,480, and 57,317 kg ha⁻¹, respectively. Conversion of grassland into cropland during the 12-year period increased the bulk density by 10.5% and soil erodibility by 46.2%; it decreased SOM by 48.8%, SOC content by 43%, available water capacity (AWC) by 30.5%, and total porosity by 9.1% for the 0–20 cm soil depth (p<0.001). The correlation matrix revealed that SOC content was positively correlated with AWC, total porosity, mean weight diameter (MWD), forest, and grassland, and negatively with bulk density, pH, soil erodibility factor, and cropland. The multiple regression (MLR) models indicated that any two of the three ecosystems and one of the two soil depths accounted for 86.5% of variation in mean SOC values ((p<0.001).     

基于小波分析的土壤速效K含量高光谱反演

选取新疆奇台县的134个土壤样本,利用土壤反射率对数的一阶导数光谱分别对4 种小波函数进行多层离散分解,采用PLSR方法分别建立了土壤速效钾含量的反演模型,并对其精度值进行检验。结果表明：小波分解获得的各层低频系数以1～3层较高,而其余各层则较低。所有函数分解的6层中,均以第2层低频系数建模的精度最高,随着分解层数（>2层）的增加,其精度值和显著性明显降低。相同尺度下,采用4种小波函数的低频系数构建的反演模型的精度差异较小,而Bior1.3为最优函数;基于Bior 1.3分解的ca2低频系数建模的R²达0.964,RMSE仅为8.19 mg·kg^-1,且为极显著水平,为最佳反演模型,经样本检验后发现,此模型可用以快速、准确估算土壤高光谱速效钾含量。     

Spatial and temporal variability of soil properties under Caragana microphylla shrubs in the northwestern Shanxi Loess Plateau, China

Shrub-induced spatial and temporal heterogeneity of soil properties is common in arid and semiarid ecosystems, and it facilitates the development of plant species diversity. We selected 5-, 10-, 20-, 30-, and 40-year-old Caragana microphylla shrubs in the Shanxi Loess Plateau to evaluate the spatial and temporal heterogeneity of soil properties under and outside the shrub canopy. In addition, the presence of adventitious plant species was investigated to assess the development of herbaceous species diversity. Soil samples were collected from two depths (0-5 cm and 5-10 cm). The establishment and development of shrubs promoted temporal variation, improved soil texture, enhanced soil organic matter (SOM), total nitrogen (TN), and cation exchange capacity (CEC), and decreased pH, bulk density (BD), and soil water content (SWC). The results further confirmed that SOM, TN, and CEC were significantly higher at the center than at the outside of the shrub canopies (P < 0.05) and were higher at the 0-5 cm depth than at the 5-10 cm depth. Moreover, the differences in SOM, TN, and CEC from the center to the outside of shrub canopies were greater under 30- and 40-year-old shrubs than under 10- and 5-year-old shrubs. Furthermore, the spatiotemporal heterogeneity of the soil properties facilitated the development of herbaceous species diversity.     

Fractal dimension and fractal growth of urbanized areas

Based on a box-accounting fractal dimension algorithm (BCFD) and a unique procedure of data processing, this paper computes planar fractal dimensions of 20 large US cities along with their surrounding urbanized areas. The results show that the value range of planar urban fractal dimension (D) is 1< D <2, with D for the largest city, New York City, and the smallest city, Omaha being 1.7014 and 1.2778 respectively. The estimated urban fractal dimensions are then regressed to the total urbanized areas, Log (C), and total urban population, Log (POP), with log-linear functions. In general, the linear functions can produce good-fits for Log (C) vs. D and Log (POP) vs. in terms of R₂ values. The observation that cities may have virtually the same D or Log(C) value but quite disparate population sizes indicates that D itself says little about the specific orientation and configuration of an urban form and is not a good measure of urban population density. This paper also explores fractal dimension and fractal growth of Baltimore, MD for the 200-year span from 1792–1992. The results show that Baltimore's D also satisfies the inequality 1< D <2, with D =1.0157 in 1822 and D =1.7221 in 1992. D =0.6641 for Baltimore in 1792 is an exception due mainly to its relatively small urban image with respect to pixel size. While D always increases with Log (C) over the years, it is not always positively correlated to urban population, Log(POP).     

内蒙古锡林河流域羊草草原植物生长旺季呼吸的影响因素分析和区分

Based on the static opaque chamber method,the respiration rates of soil microbial respiration,soil respiration,and ecosystem respiration were measured through continuous in-situ experiments during rapid growth season in semiarid Leymus chinensis steppe in the Xilin River Basin of Inner Mongolia,China. Soil temperature and moisture were the main factor affecting respiration rates. Soil temperature can explain most CO2 efflux variations (R2=0.376-0.655) excluding data of low soil water conditions. Soil moisture can also effectively explain most of the variations of soil and ecosystem respiration (R2=0.314-0.583),but it can not explain much of the variation of microbial respiration (R2=0.063). Low soil water content (≤5%) inhibited CO2 efflux though the soil temperature was high. Rewetting the soil after a long drought resulted in substantial increases in CO2 flux at high temperature. Bi-variable models based on soil temperature at 5 cm depth and soil moisture at 0-10 cm depth can explain about 70% of the variations of CO2 effluxes. The contribution of soil respiration to ecosystem respiration averaged 59.4%,ranging from 47.3% to 72.4%; the contribution of root respiration to soil respiration averaged 20.5%,ranging from 11.7% to 51.7%. The contribution of soil to ecosystem respiration was a little overestimated and root to soil respiration little underestimated because of the increased soil water content that occurred as a result of plant removal.     

基于机器学习和多光谱遥感的银川平原土壤盐分预测

快速获取区域土壤盐渍化程度信息,对于盐渍化治理与生态环境保护具有重要意义。以银川平原为研究区,以盐分影响因子和盐分指数分别作为输入参数,建立支持向量机(SVM),BP神经网络(BPNN)和贝叶斯神经网络(BNN)3种土壤盐分预测模型,选取最佳模型进行研究区不同深度的土壤盐渍化预测。结果表明：(1)0～20 cm土壤盐分预测模型中基于影响因子变量组的BNN模型效果最佳,决定系数(R²)为0.618,均方根误差(RMSE)为2.986;20～40 cm土壤盐分预测模型中基于盐分指数变量组的BNN模型效果最佳,R²为0.651,RMSE为1.947;综合对比下,BNN模型的预测效果最好,可用于研究区土壤盐渍化预测。(2)银川平原主要是以非盐渍化和轻度盐渍化为主,0～20 cm土壤重度盐渍化及盐土共占总面积的11.59%,20～40 cm土壤重度盐渍化及盐土共占总面积的7.04%,20～40 cm土壤盐渍化程度较0～20 cm土壤盐渍化轻。     

Pedotransfer functions to estimate soil bulk density for Northern Africa: Tunisia case

Countries should provide regularly national inventories of greenhouse gas emissions and sinks and, and for the agriculture and forestry sectors this comprise national estimates of soil organic carbon (C) stocks. Estimation of soil C stock requires soil bulk density (D_b) values. However, direct measurement of D_b is often lacking mainly for soils in arid and semi-arid conditions. Much effort has been made in finding alternative solution to predict D_b, either improving in situ determinations, either improving estimation procedures based on other soil properties. Regression models or pedotransfer functions (PTFs) based on easily measured soil properties constitute an adequate tool to assess D_b, since it needs a minimum data set of indicators. A forward stepwise multiple linear regression routine was used to predict D_b from physico-chemical soil properties. In this study, a soil database was organised from published and unpublished data from Tunisia. The database consisted of 238 soil profiles corresponding to 707 soil horizons from Tunisia. A general regression model fitted with all the data showed that OC, Clay, coarse-Sand and pH were the principal contributors to D_b prediction (R² = 0.55, standard error of prediction = 0.14). Additional models based on different set of variables are also provided providing alternative solutions for different levels of soil information. Predictions of the models were often improved when the data were partitioned into groups by soil depth (0–40 and 40–100 cm) and soil orders. This study also showed that CaCO₃ might be an important predictor for deeper soil horizon. The proposed PTFs for Tunisia might be useful for a larger range of soil from arid and sub arid regions.