Dry season mapping of savanna forage quality, using the hyperspectral Carnegie Airborne Observatory sensor

Knox, Nichola M.; Skidmore, Andrew K.; Prins, Herbert H. T.; Asner, Gregory P.; Werff, H.M.A. van der; Boer, Willem F. de; Waal, C. van der; Knegt, Henrik J. de; Kohi, Edward M.; Slotow, Rob; Grant, Rina

doi:10.1016/j.rse.2011.02.007

“…18,19 To date, there are a series of machine learning techniques such as random forest (RF) 20 and artificial neural network (ANN) that are also applicable. 13,17 These latter techniques were found to be robust and they circumvent the overfitting and multicollinearity problem when estimating vegetation parameters. Traditionally, machine learning techniques such as RF and ANN were applied more for classification than for regression analysis.…”

Section: Introductionmentioning

confidence: 99%

Potential of Sentinel-2 spectral configuration to assess rangeland quality

Ramoelo

¹

,

Cho

²

,

Mathieu

³

et al. 2015

J. Appl. Remote Sens

Self Cite

View full text Add to dashboard Cite

Abstract. Sentinel-2 is intended to improve vegetation assessment at local to global scales. Today, estimation of leaf nitrogen (N) as an indicator of rangeland quality is possible using hyperspectral systems. However, few studies based on commercial imageries have shown a potential of the red-edge band to accurately predict leaf N at the broad landscape scale. We intend to investigate the utility of Sentinel-2 for estimating leaf N concentration in the African savanna. Grass canopy reflectance was measured using the analytical spectral device (ASD) in concert with leaf sample collections for leaf N chemical analysis. ASD reflectance data were resampled to the spectral bands of Sentinel-2 using published spectral response functions. Random forest (RF), partial least square regression (PLSR), and stepwise multiple linear regression (SMLR) were used to predict leaf N using all 13 bands. Using leave-one-out cross validation, the RF model explained 90% of leaf N variation, with a root mean square error of 0.04 (6% of the mean), which is higher than that of PLSR and SMLR. Using RF, spectral bands centered at 705 nm (red edge) and two shortwave infrared bands centered at 2190 and 1610 nm were found to be the most important bands in predicting leaf N.

show abstract

“…[10][11][12] Using hyperspectral technologies, leaf N has been estimated with acceptable accuracies using statistical analysis based on (1) vegetation indices, (2) full spectra, (3) absorption features, and (4) integrated modeling (combining indices and environmental variables). [11][12][13][14] The challenge is that the hyperspectral imagery is only available at a local scale and is costly. Regional assessment of leaf N has been lagging behind because of the paucity of satellite data with spectral configurations which are appropriate to detect subtle leaf N variation.…”

Section: Introductionmentioning

confidence: 99%

The potential of Sentinel-2 spectral configuration to assess rangeland quality

Ramoelo

¹

,

Cho²,

Mathieu

³

et al. 2014

Self Cite

View full text Add to dashboard Cite

Abstract. Sentinel-2 is intended to improve vegetation assessment at local to global scales. Today, estimation of leaf nitrogen (N) as an indicator of rangeland quality is possible using hyperspectral systems. However, few studies based on commercial imageries have shown a potential of the red-edge band to accurately predict leaf N at the broad landscape scale. We intend to investigate the utility of Sentinel-2 for estimating leaf N concentration in the African savanna. Grass canopy reflectance was measured using the analytical spectral device (ASD) in concert with leaf sample collections for leaf N chemical analysis. ASD reflectance data were resampled to the spectral bands of Sentinel-2 using published spectral response functions. Random forest (RF), partial least square regression (PLSR), and stepwise multiple linear regression (SMLR) were used to predict leaf N using all 13 bands. Using leave-one-out cross validation, the RF model explained 90% of leaf N variation, with a root mean square error of 0.04 (6% of the mean), which is higher than that of PLSR and SMLR. Using RF, spectral bands centered at 705 nm (red edge) and two shortwave infrared bands centered at 2190 and 1610 nm were found to be the most important bands in predicting leaf N.

show abstract

“…However, large herbivores are known to concentrate in nutrient-rich sites and select feed items of high nutritive value [11,12]. We expected that reintroduced buffalo would select foraging patches with grasses of high quality to obtain sufficient quantities of the most limiting forage nutrients such as nitrogen and phosphorus [13][14][15] and avoid patches with high structural carbohydrates and low nutritive value [13,16,17]. Forage quality measures such as the levels of nitrogen and phosphorus are mainly crucial in the dry season when their concentrations decline, while the structural carbohydrate components increase [13].…”

Section: Introductionmentioning

confidence: 99%

Post-Release Monitoring of Diet Profile and Diet Quality of Reintroduced African Buffalo (Syncerus Caffer) in Umfurudzi Park, Zimbabwe

Muposhi

¹

,

Chanyandura

²

,

Gandiwa

³

et al. 2014

Tropical Conservation Science

View full text Add to dashboard Cite

Post-release monitoring of wildlife is essential to the success of ecological restoration initiatives. Translocation of wildlife to new ecosystems is associated with changes in diet profiles of individuals and ultimately animal performance, since productivity of rangelands varies in time and space. The population decline and local extinction of buffalo and other species in Umfurudzi Park, Zimbabwe, in the late 1980s led to temporary suspension of hunting activities. Recently, efforts have been made to resuscitate Umfurudzi Park through wildlife reintroductions and active ecosystem management. We assessed the diet profile and diet quality of the African buffalo (Syncerus caffer) between 2011 and 2012. A total of 42 grass species constituted the diet spectrum of buffalo. Heteropogon contortus, Eragrostis racemosa, Steroechlaena tenuifolia andThemeda triandra contributed significantly to the buffalo diet during the four seasons of the study period. Although a wide spectrum of grass species was selected in the wet season, the diversity of grass species selected was not significantly different among the seasons. Grass species crude protein deteriorated from 4.5% in the wet season to 3.5% in the dry season. Crude protein and phosphorus levels in buffalo faecal samples were within the recommended nutrient scale for southern Africa large herbivores. We conclude that feed availability and quality may not be limiting the persistence of the reintroduced buffalo. There is a need to closely monitor grass availability, dietary shifts, and forage quality over time, as well as resource partitioning with other feeding guilds.

show abstract

Dry season mapping of savanna forage quality, using the hyperspectral Carnegie Airborne Observatory sensor

Cited by 84 publications

References 55 publications

Potential of Sentinel-2 spectral configuration to assess rangeland quality

Potential of Sentinel-2 spectral configuration to assess rangeland quality

The potential of Sentinel-2 spectral configuration to assess rangeland quality

Post-Release Monitoring of Diet Profile and Diet Quality of Reintroduced African Buffalo (Syncerus Caffer) in Umfurudzi Park, Zimbabwe

Contact Info

Product

Resources

About