Large-Scale Mapping of Soil Quality Index in Different Land Uses Using Airborne Hyperspectral Data
Israr Majeed, Bhabani S. Das
Abstract
Large-scale mapping of soil quality index (SQI) is a challenging task because of the cost and time involved in measuring required soil parameters through conventional wet chemistry-based methods. Hyperspectral remote sensing (HRS) may be used to overcome such a challenge. We hypothesize that soil quality at a specific location may be estimated from remotely-sensed reflectance spectra because both these attributes are composite parameters. We used the HRS data collected with the Airborne Visible-Infrared Imaging Spectrometer - Next Generation (AVIRIS-NG) sensor to estimate SQIs in an agricultural catchment. Soil quality indices were developed from 16 different soil properties measured at 101 locations coinciding AVIRIS-NG flight. Chemometric models were used to estimate SQIs from spectral reflectance data collected under laboratory conditions and those processed from AVIRIS-NG data before and after linear and nonlinear unmixing. Except for the linearly unmixed AVIRIS-NG data, three other spectral data sources yielded coefficient of determination (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) values exceeding 0.7. Specifically, the R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values for the mixed and nonlinearly-unmixed spectra were 0.71 and 0.72, respectively, suggesting that HRS approach may directly be used for estimating SQIs. With high validation statistics, we converted the AVIRIS-NG imagery to SQI map for the entire catchment. Such high spatial resolution maps allowed us to examine the effects of land use/cover on soil quality. Strong linear dependencies between SQI and land uses and terrain structures suggested that HRS-derived SQI maps may be used to prioritize soil management efforts for sustainable development.