Using multi-date high spectral resolution data to assess the physiological status of macroscopically undamaged foliage on a regional scale

https://doi.org/10.1016/j.jag.2013.09.009Get rights and content

Highlights

  • Two temporal hyperspectral (HS) image data sets were utilized to detecting stress for the Norway spruce forests growing in a region affected by long-term extensive mining.

  • The classifications were validated by the ground truth data and were associated with the geochemical forest stand conditions.

  • Higher carotenoid to chlorophyll ratio (Car/Cab) enabled a detection of the stressed trees using HS image data.

  • Affected sites showed critically low pH and lower macronutrient values.

Abstract

Forests play an important role in regulation of the global climate; moreover, they provide human beings with a whole range of ecosystem services. Forest health and ecosystem functioning have been influenced by anthropogenic activities and their consequences, such as air pollution, surface mining, heavy metal contamination, and other biotic and abiotic stress factors, which had an especially serious effect on central Europe. Many aspects of the physiological state of trees are more or less related to the concentrations of two main groups of leaf photosynthetic pigments: chlorophylls and carotenoids. Therefore, their contents can be used as non-specific indicators of the actual tree physiological status, stress and the pre-visible tree damage. Variations in leaf biochemical composition affect foliar optical properties and can be assessed remotely using high spectral resolution data (hyperspectral data). These data were successfully used in earlier studies to detect vegetation stress and damage. However, only a few approaches have dealt with the use of hyperspectral remote sensing to assess vegetation physiological status on a regional scale. Moreover, little or no research has been done on assessing vegetation health while utilizing multi-date hyperspectral images.

In this study, the method for assessing forest health conditions using optical indices retrieved from hyperspectral data was applied to the two temporal HyMap date sets acquired in 07/2009 and 08/2010 to detect stress for the Norway spruce forests in Sokolov, NW Bohemia, a region affected by long-term extensive mining. The classification results were validated by ground truth data (total chlorophyll – Cab, carotenoids – Car and carotenoid to chlorophyll ratio – Car/Cab) and were associated with the geochemical conditions of the forest stands. Both biochemical analysis of the sampled foliage and classification of 2009 and 2010 hyperspectral image identified the same sites affected by vegetation stress. In addition to higher Car/Cab, which enabled detection of the stressed trees using hyperspectral image data, these sites showed critically low pH and lower values for the macronutrient parameters in both organic horizons and, in addition, both sites exhibit critically low base cation to aluminum ratios (Bc/Al) for lower organic and top mineral (0–20 cm) soil horizons.

The results of this study demonstrate (i) the potential application of hyperspectral remote sensing as a rapid method of identifying tree stress prior to symptom expression, and (ii) the added value of multi-temporal approaches for hyperspectral data and its further potential for monitoring forest ecosystems.

Introduction

Forests play an important role in regulation of the global climate via the global carbon cycle, evapotranspiration, and earth surface albedo (Bonan, 2008, Jackson et al., 2008). Moreover, forests provide humans with a whole range of ecosystem services including provision of food and forest products, regulation of the hydrological cycle, protection of soil resources, etc. Forest health and ecosystem functioning have recently been influenced by anthropogenic activities and their consequences, such as air pollution, surface mining, heavy metal contamination (Aznar et al., 2009), and other biotic and abiotic stress factors such as pest invasions and soil acidification (Sebesta et al., 2011), which had an especially high effect on central Europe. Therefore, large-scale monitoring of forest health and its methodologies are in the forefront of interest for scientists as well as forest managers.

Many aspects of the physiological state of trees are more or less connected with the concentrations of two main groups of leaf photosynthetic pigments: chlorophylls and carotenoids (Ustin et al., 2009). Vegetation with a high concentration of chlorophyll is considered to be healthy, as the chlorophyll content is linked to greater light-use efficiency, photosynthetic activity and carbon dioxide uptake (Blackburn, 2007, Kramer, 1981, Wu et al., 2008). Chlorophyll generally decreases under stress and during senescence (Blackburn, 2007). Carotenoids play the main role in the process of incident light absorption, transportation of energy to the reaction center of the photosystems, and heat dissipation of energy in case of high irradiation (Demmig-Adams and Adams, 1996). Combination of the influences of chlorophylls and carotenoids is thus connected with light-use efficiency (Landsberg et al., 1997). However, higher carotenoid to chlorophyll ratios indicate vegetation stress and senescence (Demmig-Adams and Adams, 1996, Young and Britton, 1990).

Therefore, the content of biochemical compounds such as photosynthetic pigments can be used as non-specific indicators of the actual tree physiological status, stress and the pre-visible tree damage. Moreover, the contents of photosynthetic pigments are closely related to photosynthetic performance and can serve as non-specific stress indicators in a very early stage, when the needles do not yet show any microscopic or macroscopic damage symptoms (Lepedus et al., 2005, Soukupova et al., 2000, Tzvetkova and Hadjiivanova, 2006). When dealing with photosynthetic pigments as vegetation stress indicators, seasonal dynamics in pigment contents in evergreen conifers must be taken into account. Changes in pigment levels reflect the normal physiological responses in the plant as well as the responses to environmental stress (Gamon and Surfus, 1999, Gitelson et al., 2001, Gitelson et al., 2002, Grisham et al., 2010). The chlorophyll content in needles increases in the spring (from May/June) and during the summer (July–September) and then remains relatively constant until October, when it again decreases during the frost hardening process (Oquist and Huner, 2003). Some of the carotenoids (e.g., lutein, β-carotene) exhibit stable contents during the seasons in contrast to several groups of xanthopylls (e.g., antheraxanthin, zeaxanthin; Yatsko et al., 2011), which increase significantly during the winter and serve as effective protection of the photosynthetic apparatus under conditions of high irradiance and low temperature during the winter and spring (Maslova et al., 2009). The total carotenoid to total chlorophyll ratio decreases in parallel with the chlorophyll increase in the spring (Martz et al., 2007).

Conventional laboratory analyses of leaf biochemical parameters can be very precise, although they have a number of disadvantages (e.g., limited number of samples, high labor and cost demands). It has been demonstrated that variations in leaf biochemical composition affect foliar optical properties (Carter and Knapp, 2001, Clevers and Kooistra, 2012, Kokaly et al., 2009, Schlerf et al., 2010, Sims and Gamon, 2002, Ustin et al., 2009, Zhang et al., 2012). The spectral reflectance characteristics of plant canopies are influenced by the chemical composition, internal leaf structure and spatial distribution of the leaves (Asner, 1998, Ollinger, 2011, Zwiggelaar, 1998). Leaf pigments are well positioned to absorb incident light and can be assessed with spectral reflectance. The more important absorption pigments and their characteristic absorption wavelengths/wavebands were reviewed by Zwiggelaar (1998).

Modern remote sensing has become a novel tool not only for detecting target materials but also for monitoring dynamic processes and physical-property induced changes. The use of multispectral imagery has been demonstrated to effectively map the distribution of ecosystem types and vegetation systems (Everitt et al., 2002, Gould, 2000, Knorn et al., 2009, Lamb and Brown, 2001, Vogelmann et al., 2012); however, the low spectral resolution of multispectral imagery is a major limitation. On the other hand, imagery with higher spectral resolution (e.g., hyperspectral) provides sufficient spectral resolution to describe diagnostic absorption signatures and allows sufficiently detailed species discrimination and biochemical differentiation (Aspinall, 2002, Feret and Asner, 2013, Kokaly et al., 2009, Lass and Prather, 2004, Majeke et al., 2008, Odagawa and Okada, 2009, Underwood et al., 2003, Ustin et al., 2004, Zhao et al., 2013).

Data with very high spectral resolution – also referred to as imaging spectroscopy (IS) data, which is also known in the remote sensing community as hyperspectral data – has been successfully used in earlier studies to detect vegetation stress and damage (Campbell et al., 2004, Campbell et al., 2007, Hamzeh et al., 2013, Hernandez-Clemente et al., 2011, Pu et al., 2008, Rathod et al., 2013, Romer et al., 2012, Sanches et al., 2013a, van der Meer et al., 2002, van der Werff et al., 2008). In the forestry context, most published precedents used IS data as a basis for identification of stress-sensitive wavelengths (e.g., Ahern, 1988, Masaitis et al., 2013, Sanches et al., 2013a), for development of stress-sensitive vegetation indices (e.g., Carter and Miller, 1994, Carter and Knapp, 2001, de Jong et al., 2012, Lausch et al., 2013) and for integrating stress-sensitive indices in more complex models (e.g., Fassnacht et al., 2012, Pontius et al., 2008, Shafri et al., 2012, Zarco-Tejada et al., 2004, Zhao et al., 2013).

In contrast, only a few approaches (Asner and Martin, 2009, Kampe et al., 2010) have dealt with the use of hyperspectral remote sensing (image spectroscopy) to assess the physiological status of vegetation on a regional scale, the term “regional” is understood to mean related or limited to a particular region, area or part, as of a country or the body. Moreover, little or no research has been done on assessing vegetation health utilizing multi-date hyperspectral image data, as a time-series of hyperspectral data and reliable methods to extract change/stress information for remotely sensed data analysis are still lacking. We recently proposed a new method for assessing forest health condition via normalization and further statistical integration of optical indices retrieved from HS image data (Misurec et al., 2012). To assess subtle changes in the physiological status of macroscopically undamaged foliage of Norway spruce, this method integrated the following HyMap derived parameters: quantitative retrieval of chlorophyll concentrations (Cab); Red-Edge Position (REP) (Curran et al., 1995) – the inflection point of the spectral curve in the red-edge region, which is shifted to shorter wavelengths under vegetation stress (e.g., the presence of heavy metals in the soil) (Chang and Collins, 1983, Clevers et al., 2002, Curran et al., 1995, Horler et al., 1983, Cho et al., 2012, Rock et al., 1988, Sanches et al., 2013b); and the Structure Insensitive Pigment Index (SIPI) (Penuelas et al., 1995) which is sensitive to the ratio of bulk carotenoids to chlorophyll. This method is easily applicable and, after further testing and confirming its general applicability, it has a potential to be adopted for other coniferous species or as a monitoring tool if using multi-temporal HS data.

Two temporal HS image data sets (HyMap 2009 and 2010 image data) were processed in order to answer whether the suggested method (Misurec et al., 2012) would be valid if using other temporal HS image data and sensitive enough to detect changes between two subsequent growing seasons. To justify this point, the following issues needed to be studied:

  • -

    What are variations in biochemical parameters between two subsequent growing seasons of a Norway spruce forest? Accordingly, the foliar pigment content from the samples collected in two subsequent growing seasons 2009 (28–29 July) and 2010 (2–3 September) were compared and statistically assessed.

  • -

    Whether the detected stress can be linked with or explained by the geochemical conditions of the forest soils? To answer this question vegetation stress has been assessed within the selected Norway spruce sites while putting together information on the forest stand's geochemical conditions, foliar biochemistry (pigment contents) and the temporal differences detected by classifying the two HS image datasets acquired one year apart.

Section snippets

Test site

The study was performed in the Sokolov basin in the Western part of the Czech Republic, in a region affected by long-term extensive lignite mining (Fig. 1). Due to the mining activities and coal burning power plants that were built in the immediate vicinity of the mined area, this region is one of the most contaminated areas of the Czech Republic where high abundances of trace elements have been detected (Suchara et al., 2011). The average altitude of the study region is about 470 m. Because the

Site soil characteristics

The soil chemistry at all the sites was characterized by low exchangeable pH in both organic (Ol + Of, Oh) and mineral soils (A0–10, A10–20) (Fig. 3). However, the mineral soil chemistry reflected the composition of the parent material. Erika, the most acidic site, is underlain by sandstone and quartzite characterized by extremely low base cation contents. As a consequence, the lowest base saturation (2–2.5%) and Bc/Al ratio (0.04–0.06) were measured in the top mineral soil (0–20 cm). Slightly

Discussion

All the sites are located in the vicinity of the Sokolov basin, an area with historically high emissions of acidifying compounds (SO2, NOx). Unfavorable low Bc/Al ratios across all the sites might be connected to the high deposition of S and N in the past and subsequently to the anthropogenic acidification of forest soils in this region (Kram et al., 1997, Oulehle et al., 2006). Soil acidification promotes leaching of base cations and mobilization of aluminum (Kram et al., 2009). Thus,

Conclusions

In this study, the recently proposed method for assessing forest health conditions using optical indices retrieved from HS data (Misurec et al., 2012) was applied to the two temporal HyMap date sets acquired in 07/2009 and 08/2010 for detecting stress in Norway spruce forests on a regional scale. The classification results were validated by ground truth data (foliar biochemistry: Cab, Car and Car/Cab) and were associated with the geochemical conditions of the forest stands (pH, macronutrient

Acknowledgements

The present research is being undertaken mainly within the framework of a grant No. 205/09/1989 (HypSo: grant funded by the Czech Science Foundation). Additionally, the research was also supported by grants No. 244242 (EO-MINERS: FP7 grant funded by the EC) and No. LM2010007 (CzeCOS). Many thanks are due to Dr. Petr Rojík (Sokolovská uhelná a.s.) for his substantial assistance with the field campaign. The colleagues from DLR are thanked for the atmospheric correction of the HyMap 2010 data set.

References (106)

  • S. Hamzeh et al.

    Estimating salinity stress in sugarcane fields with spaceborne hyperspectral vegetation indices

    International Journal of Applied Earth Observation and Geoinformation

    (2013)
  • R. Hernandez-Clemente et al.

    Assessing structural effects on PRI for stress detection in conifer forests

    Remote Sensing of Environment

    (2011)
  • E.B. Knipling

    Physical and physiological basis for the reflectance of visible and near IR radiation from vegetation

    Remote Sensing of Environment

    (1970)
  • J. Knorn et al.

    Land cover mapping of large areas using chain classification of neighboring Landsat satellite images

    Remote Sensing of Environment

    (2009)
  • R.F. Kokaly et al.

    Characterizing canopy biochemistry from imaging spectroscopy and its application to ecosystem studies

    Remote Sensing of Environment

    (2009)
  • P. Kram et al.

    Long-term changes in aluminum fractions of drainage waters in two forest catchments with contrasting lithology

    Journal of Inorganic Biochemistry

    (2009)
  • D.W. Lamb et al.

    Remote-sensing and mapping of weeds in crops

    Journal of Agricultural Engineering Research

    (2001)
  • F. Oulehle et al.

    The effect of reduced atmospheric deposition on soil and soil solution chemistry at a site subjected to long-term acidification, Nacetin, Czech Republic

    Science of the Total Environment

    (2006)
  • J. Pontius et al.

    Ash decline assessment in emerald ash borer-infested regions: A test of tree-level, hyperspectral technologies

    Remote Sensing of Environment

    (2008)
  • R.J. Porra et al.

    Determination of accurate extinction coefficients and simultaneous-equations for assaying chlorophyll-a and chlorophyll-b extracted with 4 different solvents—verification of the concentration of chlorophyll standards by atomic-absorption spectroscopy

    Biochimica et Biophysica Acta

    (1989)
  • B.N. Rock et al.

    Comparison of insitu and airborne spectral measurements of the blue shift associated with forest decline

    Remote Sensing of Environment

    (1988)
  • I.D. Sanches et al.

    Assessing the impact of hydrocarbon leakages on vegetation using reflectance spectroscopy

    ISPRS Journal of Photogrammetry and Remote Sensing

    (2013)
  • I.D. Sanches et al.

    ‘Unravelling remote sensing signatures of plants contaminated with gasoline and diesel: An approach using the red edge spectral feature

    Environmental Pollution

    (2013)
  • C.B. Schaaf et al.

    First operational BRDF, albedo nadir reflectance products from MODIS

    Remote Sensing of Environment

    (2002)
  • G. Schaepman-Strub et al.

    Reflectance quantities in optical remote sensing-definitions and case studies

    Remote Sensing of Environment

    (2006)
  • M. Schlerf et al.

    Retrieval of chlorophyll and nitrogen in Norway spruce (Picea abies L. Karst.) using imaging spectroscopy

    International Journal of Applied Earth Observation and Geoinformation

    (2010)
  • J. Sebesta et al.

    Acidification of primeval forests in the Ukraine Carpathians: vegetation and soil changes over six decades

    Forest Ecology and Management

    (2011)
  • D.A. Sims et al.

    Relationships between leaf pigment content and spectral reflectance across a wide range of species, leaf structures and developmental stages

    Remote Sensing of Environment

    (2002)
  • I. Suchara et al.

    The performance of moss, grass, and 1- and 2-year old spruce needles as bioindicators of contamination: A comparative study at the scale of the Czech Republic

    Science of the Total Environmen

    (2011)
  • V. Thomas et al.

    Spatial modelling of photosynthesis for a boreal mixedwood forest by integrating micrometeorological, lidar and hyperspectral remote sensing data

    Agricultural and Forest Meteorology

    (2009)
  • E. Underwood et al.

    Mapping nonnative plants using hyperspectral imagery

    Remote Sensing of Environment

    (2003)
  • S.L. Ustin et al.

    Retrieval of foliar information about plant pigment systems from high resolution spectroscopy

    Remote Sensing of Environment

    (2009)
  • J. Verrelst et al.

    Angular sensitivity analysis of vegetation indices derived from CHRIS/PROBA data

    Remote Sensing of Environment

    (2008)
  • J.E. Vogelmann et al.

    Monitoring gradual ecosystem change using Landsat time series analyses: case studies in selected forest and rangeland ecosystems

    Remote Sensing of Environment

    (2012)
  • C.Y. Wu et al.

    Estimating chlorophyll content from hyperspectral vegetation indices: modeling and validation

    Agricultural and Forest Meteorology

    (2008)
  • A.R. Weiskittel et al.

    Influence of Swiss needle cast on foliage age-class structure and vertical foliage distribution in Douglas-fir plantations in north coastal Oregon

    Canadian Journal of Forest Research

    (2006)
  • S.M. Adler-Golden et al.

    Atmospheric correction for short-wave spectral imagery based on MODTRAN4

    Imaging Spectrometry V

    (1999)
  • F.J. Ahern

    The effects of Bark Beetle stress on the foliar spectral reflectance of lodgepole pine

    International Journal of Remote Sensing

    (1988)
  • G.P. Asner et al.

    Airborne spectranomics: mapping canopy chemical and taxonomic diversity in tropical forests

    Frontiers in Ecology and the Environment

    (2009)
  • J.C. Aznar et al.

    Lead exclusion and copper translocation in Black spruce needles

    Water, Air, & Soil Pollution

    (2009)
  • M. Bachmann et al.

    Quality Report on HyMap 2009 Sokolov data

    Internal report of the EO-MINERS project, DLR

    (2011)
  • G.A. Blackburn

    Hyperspectral remote sensing of plant pigments

    Journal of Experimental Botany

    (2007)
  • G.B. Bonan

    Forests and climate change: forcings, feedbacks, and the climate benefits of forests

    Science

    (2008)
  • P.K.E. Campbell et al.

    Assessment of vegetation stress using reflectance or fluorescence measurements

    Journal of Environmental Quality

    (2007)
  • P.K.E. Campbell et al.

    Detection of initial damage in Norway spruce canopies using hyperspectral airborne data

    International Journal of Remote Sensing

    (2004)
  • G.A. Carter

    Primary and secondary effects of water-content on the spectral reflectance of leaves

    American Journal of Botany

    (1991)
  • G.A. Carter et al.

    Leaf optical properties in higher plants: Linking spectral characteristics to stress and chlorophyll concentration

    American Journal of Botany

    (2001)
  • S.H. Chang et al.

    Confirmation of the airborne biogeophysical mineral exploration technique using laboratory methods

    Economic Geology

    (1983)
  • J.G.P.W. Clevers et al.

    Derivation of the red edge index using the MERIS standard band setting

    International Journal of Remote Sensing

    (2002)
  • J.G.P.W. Clevers et al.

    Using hyperspectral remote sensing data for retrieving canopy chlorophyll and nitrogen content

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

    (2012)
  • Cited by (28)

    • Per-pixel analysis of ASTER data for porphyry copper hydrothermal alteration mapping: A case study of NE Isfahan, Iran

      2020, Remote Sensing Applications: Society and Environment
      Citation Excerpt :

      The SFF matches the specific absorption features from image spectra to endmembers within a certain wavelength range (Clark et al., 1990). This method allows assessing a whole absorption shape (Kopacková, 2014) and highlights the smallest absorption features (Clark and Swayze, 1995; Chen et al., 2007). On the other hand, the MRSFF is an advanced SFF method designed to define specific and multiple wavelength ranges of almost every target's absorption feature (Clark et al., 1990; Heiden et al., 2007).

    • Heritable variation in needle spectral reflectance of Scots pine (Pinus sylvestris L.) peaks in red edge

      2018, Remote Sensing of Environment
      Citation Excerpt :

      Number of trees analyzed at both sites: NSkelná Huť = 208; NNepomuk = 315. Spectral signatures derived from foliar pigment content are frequently used as non-specific stress markers in forest trees (Pukacki and Kamińska-Rożek, 2005; Kopačková et al., 2014; Mišurec et al., 2016). Knowledge of any underlying genetic control would be highly beneficial for interpretations in many fields of basic and applied studies in plant ecophysiology, and forest genetics.

    • Deep into the mud: ecological and socio-economic impacts of the dam breach in Mariana, Brazil

      2016, Natureza e Conservacao
      Citation Excerpt :

      The Brazilian Environmental Agency (IBAMA) started monitoring the full 663.2 km extension of the Doce River a week after the accident (13–20 November) through aerial photography (IBAMA, 2015b). While aerial photography only allows to map land use change and spread of the flood, spectral analysis are particularly suitable to quantify relative levels of damage to ecosystems such as heavy metal deposition and bioaccumulation (Horler et al., 1980; Clevers et al., 2004; Kopačková, 2014; Kopačková et al., 2014). High spectral resolution sensors can help quantify physiological surrogates as indicators of plant stress to toxic mineral contaminants (Kemper and Sommer, 2003; Choe et al., 2008), including plant pigments (i.e. chlorophyll, carotenoids, xanthophylls, and anthocyanins) (Blackburn, 2007), and leaf water content (Sims and Gamon, 2002).

    View all citing articles on Scopus
    View full text