Plant Litter and Soil Reflectance

Abstract

The presence of plant litter on the soil surface influences the flow of nutrients, carbon, water, and energy in terrestrial ecosystems. Quantifying plant litter cover is important for interpreting vegetated landscapes and for evaluating the effectiveness of conservation tillage practices. Current methods of measuring litter cover are subjective, requiring considerable visual judgment. Reliable and objective methods are needed. The spectral reflectance (0.4–2.5 μm) of wet and dry soils (six types) and plant litters (2 crops, 14 forest, and 2 grasses) of different ages were measured. Discrimination of plant litters from the soils was ambiguous in the visible and near-infrared (0.4–1.1μm) wavelength region. An absorption feature associated with cellulose and lignin was observed at 2.1 μm in the spectra of dry plant litter, which was not present in the spectra of soils. A new spectral variable, cellulose absorption index (CAI), was defined using the relative depth of the reflectance spectra at 2.1 μm. CAI of dry litter was significantly greater than CAI of soils. CAI generally decreased with age of the litter. Water absorption dominated the spectral properties of both soils and plant litter and significantly reduced the CAI of the plant litters. Nevertheless, the CAI of wet litter was significantly greater then CAI of wet soil. This study provides a new methodology to discriminate plant litter from soils by differences in spectral reflectance produced by their physical and chemical attributes. This remote sensing method should improve quantification of plant litter cover and thus improve estimates of phytomass production, surface energy balance, and the effectiveness of soil conservation practices. Plant litter reflectance is a verifiable component in vegetative landscapes and should be labeled and modeled separately from soils in landscape studies.

Introduction

Litter is senescent (or dead) plant material that gradually decomposes into soil. It is difficult to classify litter because there is no particular point in time where it shifts from one state of organic matter to another. In this study, litter is considered to be both senesced tree leaves and the portion of annual crops left in the field after harvest.

The decay of litter adds nutrients to the soil, improves soil structure and reduces soil erosion (Aase and Tanaka, 1991). The annual loss of 1.25 billion tons of soil in the United States could be reduced by leaving litter on bare soil (USDA, 1991). Litter also affects water infiltration, evaporation, porosity, and soil temperatures (Reicosky, 1994). Thus, the presence of plant litter on the soil surface influences the flow of nutrients, carbon, water, and energy in terrestrial ecosystems. Quantifying litter is important not only to improve surface energy balance, but also to improve estimates of net primary productivity and nutrient turnover rates. In agriculture systems, quantifying crop residue cover is necessary to evaluate the effectiveness of conservation tillage practices.

Crop residue can be identified and quantified using manual residue cover measurement techniques. Morrison et al. (1993) noted that the most widely used procedure to measure crop residue cover in the field (line-transect method) is tedious and prone to human judgment errors. These methods need to be replaced by more objective, faster, and more accurate spectral measurement techniques Daughtry et al. 1996, McMurtrey et al. 1993.

Remote sensing techniques have had only limited success in quantifying litter cover because the spectral reflectance curves of plant litter and soils have similar, generally featureless shapes in the visible and near-infrared (VIS-NIR, 0.4–1.1 μm) wavelength ranges Aase and Tanaka 1991, Daughtry et al. 1996. The problem is that there are no unique spectral features that can be used to discriminate the similar VIS-NIR curves of plant litter and soils (Wiegand and Richardson, 1992). The slope of the reflectance spectra at the VIS-NIR transition (i.e., 680–780 nm) is generally greater for litter than for soils. However, litter may be brighter or darker than a particular soil depending on moisture conditions and litter decomposition (age), which affects the slope Ahn et al. 1996, Daughtry et al. 1996, Goward et al. 1994.

Several studies have noted that the spectral features of dried litter and soils that are unique to each component in the shortwave infrared (SWIR, 1.1–2.5 μm) region Elvidge 1990, Stoner and Baumgardner 1981. Common spectral features in both plant litter and soils are two broad water absorption bands at 1.4 μm and 1.9 μm. Elvidge (1990) observed diagnostic lignin and cellulose features at 2.09 and 2.3 μm in the reflectance spectra of dried plant materials. Lignin and cellulose absorptions have also been observed in Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data and have been used to calculated a ligno-cellulose index based on the difference between reflectance in the 2.18 μm to 2.22 μm band and 2.31 to 2.38 band (Elvidge, 1988). Daughtry et al. (1996) developed a three-band spectral index for discriminating plant litter from soil that was based on the depth of the ligno-cellulose absorption feature at 2.1 relative to the shoulders at 2.0 μm and 2.2 μm.

Murray and Williams (1988) associated the absorption feature at 2.1 μm with compounds possessing alcoholic -OH groups, such as sugars, starch, and cellulose. In plant litter, the absorption at 2.1 μm is most likely due to cellulose, hemicellulose, lignin, and other structural compounds, since sugars, starches, and other nonstructural compounds are readily degraded by microorganisms. The spectra of most soils show no absorption at 2.1 μm, but rather a mineral absorption at 2.2 μm associated with the crystal lattice of clay minerals Stoner and Baumgardner 1981, Ben-Dor and Banin 1995.

The spectral properties of plant litter in the VIS-NIR wavelength range affects vegetation indices, including the normalized difference vegetation index (NDVI) (van Leeuwen and Huete, 1996). NDVI values typically range from 0.08 to 0.16 for soils and 0.14 to 0.45 for litter (McMurtrey et al., 1993). The fraction of photosynthetically active radiation (f_APAR, 0.4–0.7 μm) absorbed by vegetation is frequently estimated as a function of NDVI. For example, Goward and Huemmrich (1992) calculated daily total (DT) f_APAR as shown in Eq. (1):

$$\text{DT}\text{f}{}_{\mathrm{<mtext>APAR</mtext>}}\text{=107.5·}\text{NDVI}\text{−8.0}$$

If NDVI values for soils, corn residue, and forest litter from McMurtrey et al. (1993) are substituted in Eq. (1), then DT f_APAR ranges from 3.8% to 40.4%, even when there no green vegetation is present. Clearly, canopy models will overestimate phytomass production unless initial surface conditions are known.

van Leeuwen and Huete (1996) found the effect of different vegetation components on the soil adjusted vegetation index (SAVI) was higher for litter and bark canopies than for bare soil, as shown when computed from the SAIL model. They cite the litter/bark scattering properties at the leaf scale as a potential cause of the error in the VI response to green vegetation cover. Distinction between soils can be further demonstrated by differences in slope when cross-plots of their NIR-red reflectance are employed (Huete et al., 1985). Furthermore, variations in VI can be seen if these cross-plots are shown for both plant litter and soils.

The influence of plant litter reflectance has generally not been recognized in canopy spectral measurements (Goward and Huemmrich, 1992). As a result, the spectral reflectance of dry, bare soil rather than plant litter (e.g., Myneni et al., 1995) is used to monitor landscape processes because soil is a more permanent ground component than litter. Thus, the impact of plant litter is often neglected in spectral models that estimate plant productivity. Until plant models can account for energy that is not used to produce dry matter, such as energy absorbed by litter and by soils, these models cannot be used to accurately predict plant productivity or even the physiological state of plant canopies. Canopy models, which include green vegetation, plant litter, and soil optical properties, are more likely to evaluate the condition and yield of vegetation correctly (Daughtry et al., 1992).

The objectives of this work were to (1) acquire and analyze spectral reflectance data for a wide range of soils and plant litters and (2) develop an algorithm for discriminating litter from soil.