2022-11-19 10:45:32

Precision Agric

DOI 10.1007/s11119-009-9123-3


Estimating soil organic carbon from soil reflectance:

a review

Moslem Ladoni AElig; Hosein Ali Bahrami AElig; Sayed Kazem Alavipanah AElig; Ali Akbar Norouzi

Springer Science Business Media, LLC 2009

Abstract Soil organic carbon (SOC) concentration is a useful soil property with which to guide agricultural applications of chemical inputs. To enable this, simple, accurate, rapid and inexpensive methods are needed to produce maps of surface SOC concentrations. Researchers have investigated estimates of soil surface properties from remotely sensed information as a means of rapidly quantifying and monitoring some surface soil properties, such as SOC. The objective of this paper is to review the potential and limitations of remotely sensed data for mapping and evaluating SOC. Several statistical methods including simple regression models, the lsquo;soil linersquo; approach, principal component analysis and geostatistics have been applied to data to investigate the accuracy of such estimates. A review of the literature shows that predictive equations are not universal and require new regression models for every scene. An important benefit of remotely sensed data is to suggest a sampling strategy that can lead to improved representation of spatial heteroge-neity in SOC.

Keywords Soil organic carbon mapping Remote sensing Soil reflectance lsquo;Soil linersquo; Principal component analysis Geostatistics


Knowledge of soil physical and mechanical properties, as well as the spatial variability of these properties, is essential in the concept of precision agriculture (Hanquet et al. 2004). Spatial variation in soil properties leads to differences in concentrations, fertilizer needs,

M. Ladoni (amp;) H. A. Bahrami

College of Agriculture, Tarbiat Modares University, Tehran, Iran e-mail: moslemladoni@yahoo.com

S. K. Alavipanah

College of Geography, University of Tehran, Tehran, Iran

A. A. Norouzi

Soil and Water Research Center, Tehran, Iran


Precision Agric

activity of herbicides and crop yield within a field (Ferguson et al. 1996; Mallarino and Wittry 2004; Ritter et al. 2008). Thus, uniform treatment of the soil will result in zones within a field that are either over- or under-treated (Roy et al. 2006). The quantification of soil heterogeneity is an obstacle to the widespread adoption of precision agriculture (Franzen et al. 2000). Soil organic carbon (SOS) has a considerable effect on the inter-actions between soil and plants. The SOC content is closely related to soil quality (Reeves 1997; Susanne and Michelle 1998; Al-Kaisi et al. 2005; Huang et al. 2007), not only as an indicator of soil erosion and degradation (Dick 1992), but also as a regulating factor of processes such as water holding capacity and permeability (Pepper 1996), bioavailability and fate of many herbicides (Williams and Mortensen 2000; Graff et al. 2000), pesticide behavior in soil (Patzold et al. 2008), plant-available N (Marschner et al. 2003), the soilrsquo;s ability to adsorb plant nutrients (Shatar and McBratney 1999) and crop yield (Fleming et al. 2000; Khakural et al. 1999). Knowing its concentration in the soil could be useful, especially if its spatial distribution could be determined accurately (Blackmer and White 1998) and at a low cost (Wolf and Buttel 1996; Lu et al. 1997).

To determine the within-field variation in SOC and other properties, sampling tech-niques, such as those based on a grid or zones, have been used. The spatial variation in SOC may occur at a finer spatial scale than can be afforded to physically sample the soil followed by laboratory analysis (Hummel et al. 2001). There is a recognized need to develop methods that use the minimum number of soil samples possible to minimize the costs of producing maps of surface SOC concentrations to support precision agriculture (Rossel et al. 2001; Walvoort and McBratney 2001; Wetterlind et al. 2008), quantitative soil-landscape modeling (McKenzie et al. 2000; Lambert et al 2002) and global soil C monitoring (Post et al. 2001).

Recent research has suggested that reflectances in certain spectral bands have been correlated with soil properties and could provide inexpensive predictions of soil physical, chemical and biological properties (Ben-Dor and Banin 1995; Reeves et al. 2000; Dunn et al. 2002; Daniel et al. 2004; Roy et al. 2006; Stamatiadis et al. 2005; Francis and Schepers 1997; Pocknee et al. 1996; Ehsani et al. 1999). As SOC increases, the soil appears darker, and vice versa (Fig. 1). This general observation formed the basis of the concept that electro-optical sensing of SOC might be feasible (Alexander 1969; Steinhardt and Franzmeier 1979; Hummel et al. 2001). Several researchers have tried to identify SOC using soil reflectance in the laboratory, and the result of their resear



Estimating soil organic carbon from soil reflectance:

a review

Moslem Ladoni bull;Hosein Ali Bahrami bull; Sayed Kazem Alavipanah bull; Ali Akbar Norouzi

Springer Science Business Media, LLC 2009



引言:在精确农业的概念中,土壤物理力学属性以及这些属性的空间变异性的知识是必不可少的(Hanquet et al. 2004). 土壤性质的空间变化导致田间浓度、肥料需要、除草剂活性和作物产量的差异(Ferguson et al. 1996; Mallarino and Wittry 2004; Ritter et al. 2008) . 因此,对土壤的统一处理将导致在田地中的一些区域被过多或过少处理(Roy et al. 2006). 土壤差异性的量化是普及精确农业的一个障碍(Franzen et al. 2000). 土壤有机碳对土壤和植物之间的相互作用有很大的影响,其含量与土壤质量密切相关(Reeves 1997; Susanne and Michelle 1998; Al-Kaisi et al. 2005; Huang et al. 2007), 不仅可以作为土壤侵蚀和退化的指标(Dick 1992),而且是保水能力和渗透能力过程(Pepper 1996)、一些除草剂的生物有效性和杀伤性 (Williams and Mortensen 2000; Graff et al. 2000),农药在土壤中的运作(Patzold et al. 2008),植物有效氮(Marschner et al.2003),土壤吸附植物养分的能力(Shatar and McBratney,1999)和作物产量(Fleming et al. 2000; Khakural et al. 1999) 的调节因子。知道它在土壤中的浓度是有用的,尤其是当它的空间分布可以准确(Blackmer and White 1998)并以低成本(Wolf and Buttel 1996; Lu et al. 1997)确定。

为了确定SOC和其他属性的内部变化,已经使用了采样方法,例如基于网格或区域的采样方法。SOC的空间变化可能发生在比实验室分析采样更精细的空间尺度的土壤上(Hummel et al. 2001).有一个公认且仍需开发的方法,即使用尽可能少的土壤样本,减少制作地表土壤有机碳浓度地图的成本,以支持精确农业(Rossel et al. 2001; Walvoort and McBratney 2001; Wetterlind et al. 2008),定量土壤景观模拟(McKenzie et al. 2000; Lambert et al 2002)和全球土壤碳监测(Post et al. 2001). 最近的研究表明,某些光谱波段的反射率与土壤性质有关,可以经济地对土壤物理,化学和生物性质预测(Ben-Dor and Banin 1995; Reeves et al. 2000; Dunn et al. 2002; Daniel et al. 2004; Roy et al. 2006; Stamatiadis et al. 2005; Francis and Schepers 1997; Pocknee et al. 1996; Ehsani et al. 1999). 随着土壤有机碳的增加,土壤变暗,反过来也一样(图. 1).这种显而易见的现象为SOC的光谱观测奠定了基础(Alexander 1969; Steinhardt and Franzmeier 1979; Hummel et al. 2001). 一些研究人员试图在实验室中利用土壤反射率来识别SOC,使高分辨率光谱传感器得以发展。这些传感器产生遥感数据,可以提供有关土壤的补充信息。遥感图像获得土壤属性估计的能力,可以提高我们了解土表界限的具体位置变化。在过去的30年,已经测试了几种使用反射率来量化SOC的方法。本文的目的是检验用于确定从遥感数据的SOC含量的方法,并探讨它们的可能性和局限性。首先,概述SOC的光谱特征,其次用航空航天遥感获得的反射率中进行评估。

图.1 三种土壤样品,母质相同,但有机碳含量不同; 有机碳含量从左到右增加


物质的光谱特征是由它们的反射或吸收率决定的,是电磁波谱中波长的函数。在受控条件下,这些特征是由原子的电子跃迁和形成分子或晶体的原子结构群的振动拉伸和弯曲引起的。反射光谱中的基本特征出现在允许分子上升到更高振动状态的能级上。土壤中的吸收特征是不同矿物成分和有机物重叠的结果。(Bowers and Hanks 1965; Hunt and Salisbury 1970;

Krishnan et al. 1980; Stoner and Baumgardner 1981; Clark and Roush 1984; Dalal and

Henry 1986; Clark et al. 1990; Henderson et al. 1992).


土壤的反射光谱一般在1100-2500nm范围内,包括在1400,1900和2200nm附近的三个不同的吸收峰,在2200-2500nm之间有一些小的吸收峰(Chang and Laird 2002). 有机质通过降低总体反射率来影响光谱,从而降低光谱对比度,使得光谱和理化性质之间的关系更难以检测。与土壤有机质各种组分相关的基本特征一般出现在中-热红外(2500-25000nm),但是由于NH,OH和CH基团的弯曲,它们的弱泛音和这些基本振动的组合主导了近红外(700–2500 nm)和可见光(400–700 nm)的频谱的一部分(Shepherd and Walsh 2002; Rossel et al. 2006). 在可见光范围内,预测SOC的重要波段在410,570,660和520,540和550 nm附近。在可见光范围内,预测SOC的重要波段在410,570,660(Rossel et al. 2006)和520,540和550 nm附近(Brown et al. 2006; Daniel et al. 2004).有机物在550-700 nm范围内降低了反射率,或者在在500-1300nm范围内OM含量较大时造成凹曲线和对于较小OM量的产生凸曲线(Huete and Escadafal 1991) . OM与近红外范围内的反射率之间也有很强的相关性。反射率在960,1100 (Daniel et al. 2004), 1400 和 1900 nm (Palacio-Orueta and Ustin 1998) 1720, 2180和 2309 nm (Sudduth and Hummel 1991; Shephered and Walsh 2002) 1744, 1870 和2052nm(Dalal and Henry 1986)为OM水平最敏感的波长。与NIR相比,MIR可以更好地提供关于土壤中OC的信息(McCarty and Reeves 2006; McCarty et al. 2002). Henderson et al. (1992)根据土壤反射率对OC、母质和其他土壤性质的光谱波段进行分离。他们发现检测SOM的最有效的波长在2200-2500nm。然而,应避免2225-2255和2275 nm之间的波长,因为其他土壤性质的影响掩盖了土壤的反应。Henderson et al. (1992) 还观察到OC在光谱的可见光谱和近红外光谱(400-1100)占主导地位,通过短波红外区域的某一部分(1100–2500)可能可以评估在大的地理区域不同母质土壤的SOC采样等级。

表.1 使用多变量分析和在电磁波谱的紫外(UV),可见(VIS),近红外(NIR)和中红外(MIR)区域中的光谱响应来比较土壤OC或OM的定量预测

土壤有机质的化学性质复杂,有机质的近红外光谱响应同样复杂。因此,指定特定的土壤有机质官能团的吸收特征是困难的。有机质含量对土壤反射率的影响远大于有机质组成的影响(Henderson et al. 1992). 然而,考虑OM分解在土壤中的阶段(因此正确的OM组成)可以提供更准确的结果(Ben-Dor and Banin 1995). 800-600nm之间的光谱斜率在很大程度上取决于有机物的成熟度(Ben-Dor et al. 1997). 由于腐殖酸是黑色的,在400和2500nm内有强烈的吸收能力(弱反射), 在可见光范围有一个很大的强吸收峰。在近红外范围,腐殖酸的吸收最小值在1850 nm左右。除了与水(1400和1900 nm)相关的吸收峰,腐殖酸在2310和2350 nm处有两个明显的吸收峰,在1700和2150 nm处有小的吸收峰。由于腐殖酸的复杂化学性质和近红外光谱性质,指定腐殖酸的特定有机官能团的吸收峰是不可能的(Chang and Laird 2002).




Baumgardner et al. (1970) 和Al-Abbas et al. (1972)首先进行了航空实验,研究可见光和近红外波段OM和反射率之间的关系。随后,研究调查了遥感数据是否可用于评估农业和环境活动(Nanni 和Dematte 2006a)的SOC状况(表2),直接取样(Varvel et al. 1999)和有效区分土壤类型和植被,以及不同土壤性质之间的区别(Palacio-Orueta et al. 1999). 土壤有机碳与光谱反射率之间的强相关性意味着它可以用土壤反射率来模拟和表示。

也已经通过研究谱的不同部分,以确定评估SOC的最适的波长。Abgu et al. (1990)研究表明土壤有机碳含量只与红绿波段显着相关。Sullivan et al. (2005)的研究结果与McCarty (2002)的相似;他们的结论表明,热红外(TIR)数据可以解释93%的SOC的总变异量,可见光谱和近红外光谱则较小。在所研究的一个区域中,VIS、MIR和TIR比率解释了38%的SOC变化。同样,在另一个区域中,TIR 和 VIS的比值可以解释42%的SOC变化。Bajwa et al (2001) 发现最大的相关性出现在红光区域。在蓝绿区域,波长越小相关性越弱。近红外与其余的光谱范围相比相关性最小。光谱混合会影响光谱斜率,从而降低相关性。例如,由于土壤表面存在非土壤残留物,反射率与有机物(Galvao et al. 2001)在可见光范围内的相关系数很小。这些相关性随波长的增加而增大,最大值在1200~2000 nm之间,然后随波长的增加而减小。

表.2 裸土图像光谱的蓝(B),绿(G),红(R)和NIR波段的反射率与土壤有机质的实验室值之间的测定系数

在最近的综合研究中,Chen et al. (2000 and 2005)用两种不同的方法研究了土壤剖面上15cm处的OC含量和光谱中选定的部分之间的关系。在第一种方法中,用方程计算每个像元的表面SOC浓度,并将所得值分组为八类。在所有图像中,两种方法的测量值和预测值之间具有良好的一致性(表2)。Chen et al. (2000) 从大约115公顷的土地上采集了28个土壤样本来测定这些土壤有机碳和地图的数据。如果用网格采样制作这样的地图,需要10 - 40%的样本数。对于一个0.405公顷的栅格样本, 115公顷的区域至少需要284个样本。与栅格采样相比,SRM方法的主要优点是成本低,可以对土壤有机碳的空间变化进行详细和准确的描述。对于栅格采样,通常采用8-10个核心作为复合样本,代表0.405公顷(1英亩)或更大的面积。即使单个核心样本代表了充分取样的区域,复合样本也不包括关于样本覆盖区变化的信息。但是SRM方法可以将表层SOC浓度以图像像素尺寸的分辨率反映出来。

SRM方法也有一些缺点,例如当从具有不同母质或不同景观的大地理区域采集土壤样品时(Fernandez et al. 1988; Henderson et al. 1992; Schulz



原文和译文剩余内容已隐藏,您需要先支付 30元 才能查看原文和译文全部内容!立即支付