Evaluating MODIS soil fractional cover for arid regions, using albedo from high-spatial resolution satellite imagery

Abstract

Broad-scale high-temporal frequency satellite imagery is increasingly used for environmental monitoring. While the normalized difference vegetation index (NDVI) is the most commonly used index to track changes in vegetation cover, newer spectral mixture approaches aim to quantify sub-pixel fractions of photosynthesizing vegetation, non-photosynthesizing vegetation, and exposed soil. Validation of the unmixing products is essential to enable confident use of the products for management and decision-making. The most frequently used validation method is by field data collection, but this is very time consuming and costly, in particular in remote regions where access is difficult. This study developed and demonstrates an alternative method for quantifying landcover fractions using high-spatial resolution satellite imagery. The research aimed to evaluate the bare soil fraction in a sub-pixel product, MODIS Fract-G, for the natural arid landscapes of the far west of South Australia. Twenty-two sample regions, of 3400 sampling points each, were investigated across several arid land types in the study area. Albedo thresholds were carefully determined in Advanced Land Observing Satellite Panchromatic Remote-sensing Instrument Stereo Mapping (ALOS PRISM) images (2.5 m spatial resolution), which separated predominantly bare soil from predominantly vegetated or covered soil, and created classified images. Correlation analysis was carried out between MODIS Fract-G bare soil fractional cover and ALOS PRISM bare soil proportions for the same areas. Results showed much lower correlations than expected, though limited agreement was found in some specific areas. It is posited that the Moderate Resolution Imaging Spectroradiometer (MODIS) fractional cover product, which is based on unmixing using the NDVI and a cellulose absorption index (CAI) proxy, may be generally unable to separate soil from vegetation in situations where both indices are low. In addition, separation is hampered by the lack of 'pure pixels' in this heterogeneous landscape. This suggests that the MODIS fractional cover product, at least in its present form, is unsuited to monitor sparsely vegetated arid landscapes.

An assessment of geographical distribution of different plant functional types over North America simulated using the CLASS-CTEM modelling framework

“…The first relates to the method by which the fractional vegetation cover is calculated for the land cover types in the original remotely sensed land cover productsthat is, for the 22 land cover types in the GLC2000 data set upon which the WANG06 data are based and the 17 land cover types in the MODIS data set. An example of such an error for arid regions is illustrated by Lawley et al (2014), who suggest that the MODIS soil fractional cover product, at least in its present form, is unsuited to monitoring sparsely vegetated arid landscapes and generally unable to separate soil from vegetation in situations where NDVI is low. The second way in which errors are introduced is through the mapping of the remotely sensed land cover types to the CTEM PFTs following Table 2 of WANG06 for the GLC2000 land cover types, and following Table 2 in this paper for the MODIS land cover types.…”

Monitoring to Learn, Learning to Monitor: A Critical Analysis of Opportunities for Indigenous Community‐Based Monitoring of Environmental Change in Australian Rangelands

“…Where monitoring of environmental change does occur in remote Australia, it is typically characterised by large-scale remote sensing (e.g. Bastin and the ACRIS-MC, 2008;Lawley et al, 2014) or small-scale and isolated expert field observations (e.g. Masters et al, 2003;White et al, 2012).…”

Comparison of Hyperspectral Versus Traditional Field Measurements of Fractional Ground Cover in the Australian Arid Zone

“…Data used for calibration and validation of fractional cover are derived from a variety of sources and techniques, and there is currently no international standard. Lawley et al [20], Montesano et al [21], Morisette et al [22], and Xiao and Moody [23] all utilised remotely sensed imagery with high spatial resolution to validate FC products with lower spatial resolution. The advantages of this approach are that high-spatial resolution imagery provides an objective record at the time of the region being assessed and may allow for the validation of areas that cannot be easily accessed.…”