Improved Remotely-Sensed Estimates of Crop Residue Cover by Incorporating Soils Information.

Authors: Serbin, Guy; Daughtry, Craig; Hunt Jr, Earle; McCarty, Gregory; Doraiswamy, Paul; BROWN, DAVID - WASHINGTON STATE UNIV

Submitted to: IEEE Transactions on Geoscience and Remote Sensing
Publication Type: Abstract Only
Publication Acceptance Date: 3/25/2008
Publication Date: 7/6/2008
Citation: Serbin, G., Daughtry, C.S., Hunt, E.R., McCarty, G.W., Doraiswamy, P.C., Brown, D.J. 2008. Improved remotely-sensed estimates of crop residue cover by incorporating soils information [abstract]. IEEE Transactions on Geoscience and Remote Sensing. 2008 CDROM.

Interpretive Summary:

Technical Abstract: Conservation tillage (CT) methods, which include reduced- and no-till methods, leave substantial quantities of crop residues on the soil surface. These crop residues act as a barrier to wind and water to reduce soil erosion and evaporation. Long-term CT also increases soil organic carbon (SOC) content and sequesters atmospheric CO2 which allows farmers to sell carbon credits. The reduction in tillage also decreases fuel consumption which reduces CO2 emissions. However, the increasing demands for biofuels, e.g., grain- and cellulosic-based ethanol, may result in a decrease of both CT methods and an increase in harvesting of crop residues as ethanol feedstock. An efficient and accurate method is needed to monitor management practices. Current methods to access crop residue cover and tillage intensity, e.g., the line-point transect method and windshield surveys, are cumbersome and time-consuming. No program exists for objectively monitoring crop residue cover and tillage intensity over broad areas. Remote sensing has provided efficient and objective methods for assessing crops conditions over large areas. However, the numerous spectral indices that use Landsat TM shortwave infrared bands for assessing crop residue cover have had mixed success [1]. An alternative approach for assessing crop residue cover is based a broad absorption feature near 2100 nm that is associated with cellulose and lignin in crop residues. The Cellulose Absorption Index (CAI) is a continuum-removal spectral index in the shortwave infrared (SWIR) region of the electromagnetic spectrum [1]: (1) where R2.0, R2.1, and R2.2 are 11 nm wide bands centered at 2031, 2101, and 2211 nm, respectively. CAI is based on the depth of the alcoholic C-OH absorption at 2101 nm found in dry cellulose, but not shared by common soil minerals [2], which results in a consistent contrast between dry residues and soils. Crop residues have CAI values >0 (typically ranging between CAI = 1.1 ~ 6.3, depending on residue type and age), whereas soils have CAI values typically ranging from 0 to -10.5 depending on mineralogy and SOC. CAI was linearly related to crop residue cover [3]: (2) where fr denotes residue cover fraction and the mix, res, and soil subscripts of CAI denote spectral mixture, residue, and soil CAI values, respectively. Dry soil CAI values may be strongly affected by soil mineralogy and organic matter content, which in turn, could affect the soil end-member for estimating crop residue cover. 2. METHODOLOGY Shortly after planting in May 2006 and June 2007, airborne hyperspectral imagery (SpecTIR) were acquired over test sites in Fulton and Cass counties in north-central Indiana. The imaging spectrometer provided 5-nm bands of over the 400-2400 nm wavelength region with 4 m spatial resolution. Images from the multiple flight lines were geo-registered and mosaicked in order to cover the test sites. Crop residue cover was measured at 2 representative locations in >50 corn and soybean fields using the line-point transect method. Vertical and oblique photographs plus notes on tillage intensity were acquired at each location. For selected fields, samples of the crop residues and soils were also acquired. A wide area augmented system (WAAS) enabled GPS receiver recorded the position of each sampling location. All soil samples were analyzed for SOC and carbonate contents, the latter of which were not significant in quantity. Soil mineralogy was verified for selected soil samples using X-ray diffractometry. Reflectance spectra of the soils and crop residues at a range of water contents were acquired at 1-nm intervals over the 350-2500 nm range in the lab using a spectroradiometer (ASD FieldSpec Pro). 3. RESULTS Soils with high SOC (>4%) had lower reflectance values across the spectrum and higher CAI values than soils with low SOC. As water conte