Effect of soil spectral properties on remote sensing of crop residue cover

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

Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/3/2009
Publication Date: 9/1/2009
Citation: Serbin, G., Daughtry, C.S., Hunt, E.R., Brown, D.J. 2009. Effect of soil spectral properties on remote sensing of crop residue cover. Soil Science Society of America Journal. 73:1545-1558.

Interpretive Summary: Modern agricultural trends utilize conservation (no-till) methods, in which fields are not intensively plowed. Conservation tillage methods typically leave a significant portion of the soil covered with crop residues, which are the remnants (stalks, leaves, and unharvested cobs) of the previous season’s crops. These crop residues are important as they help protect the soil from erosion by wind and water, act as a mulch to reduce evaporation of water from the soil, and they decay to add carbon and nutrients to the soil and decrease the need for fertilizers. The addition of carbon to the soil also helps remove carbon dioxide from the atmosphere, and thus, reduce greenhouse gases. Since farmers can benefit from using conservation tillage by receiving subsidies and selling carbon credits, an efficient verification method becomes necessary. Remote sensing methods from aircraft and spacecraft can allow for rapid measurement of crop residue cover over many fields. These methods utilize the differences that crop residues have with soils in levels of reflected light (reflectance) in the shortwave infrared (SWIR) spectrum between 1900-2400 nm. However, soils have differing mineral and chemical compositions, and this affects reflectance, which could potentially complicate efforts to remotely measure crop residue cover. We tested six spectral indices for efficacy in crop residue estimation. Two of these indices, the Cellulose Absorption Index (CAI), the Lignin-Cellulose Absorption (LCA) Index, utilize spectral bands that detect absorptions in spectra that are related to cellulose and lignin, two chemical compounds found in crop residues. We also tested four spectral indices developed for the Landsat Thematic Mapper (TM), namely the Normalized Difference Tillage Index (NDTI), the Normalized Difference Senescent Vegetation Index (NDSVI), and the Normalized Difference Indices NDI5 and NDI7. NDSVI and NDI5 are also applicable for use with the Indian Resource Satellite’s Advanced Wide-Field Sensor (AWiFS). We analyzed over 4000 soil spectra in the SWIR region, 83 crop residue samples, and 40 spectra of green corn canopy for their spectral index values. We show that soil taxonomic order was not a good predictor of soil index values with significant overlap in values. However, soil index values agreed best within Land Resource Regions (LRRs) and specifically within Major Land Resource Areas (MLRAs). CAI separated crop residue the best from soil and green vegetation, with no overlap in values. LCA perfomed the next best, but degraded residues and green vegetation had overlapping LCA values; certain soils also had overlapping CAI values with crop residue. Soil CAI and LCA values were biased by soil mineralogy. All of the Landsat TM-based indices were strongly affected by green vegetation. Of these, NDTI performed the best; the remaining indices are only usable under very limiting circumstances. Future hyperspectral and advanced multispectral scanners should include CAI spectral bands for crop residue and non-photosynthetic vegetation detection.

Technical Abstract: Conservation tillage practices have been shown to improve soil structure, enhance soil organic carbon content (SOC), and reduce soil erosion. Conservation tillage practices include reduced- and no-till methods, which often leave appreciable amounts of crop residues over the soil surfaces after harvesting. Remote sensing methods have shown great promise in efficiently estimating crop residue cover, and thus, tillage practice. Furthermore, these data can be used in soil carbon models. Six remote sensing spectral indices which have been used for residue cover estimation include the Cellulose Absorption Index (CAI), the Lignin-Cellulose Absorption Index (LCA), the Normalized Difference Tillage Index (NDTI), Normalized Difference Senescent Vegetation Index (NDSVI), and the Normalized Difference Index 5 and 7 (NDI5 and NDI7, respectively). Soil spectral properties were shown to affect spectral indices for crop residue cover estimation. These indices were more dependent upon SOC and soil mineralogy. Soil taxonomic order generally had little effect on spectral reflectance. Soil crop residue index values were similar within Land Resource Regions (LRR) and specifically for Major Land Resource Areas (MLRA). CAI showed the best separation between soils and residues, followed by LCA and NDTI. NDSVI, NDI5, and NDI7 showed significant overlaps between soil and residue index values, and should only be used under very limited circumstances and with great care.