リモートセンシングとGISのジャーナル

概要

Ecogeographically and Non-Ecogeographically Forecasting Discontinuous Canopied Simulium damnosum s.l. Habitats by Interpolating Metrizable Sub-Mixel Mean Solar Exoatmospheric Quantum Scalar Irradiance

Benjamin G. Jacob, Robert J.Novak, Laurent D. Toe, Moussa S. Sanfo, Koala Lassane, Daniel A. Griffith, Thomson L. Lakwo, Peace Habomugisha, Moses N. Katabarwa, Thomas R. Unnasch

We interpolated a geospectrally decomposed 5 meter (m) RapidEye™,Red Edge, Normalized Difference Vegetation Index (NDVI), unmixed, endmember, biosignature of a georeferenced, larval habitat of Similium damnosum s.l.,a black fly vector of onchocerciasis. We did so to identify unknown, unsampled, prolific, habitats in African riverine environments. The S. damnosum s.l.larval habitat was initially geosampled in a riverine village in Burkina Faso and overlaid onto the 5m resolution data. The Band Mathfunction of ENVI 4.8TM was employed to calculate the RedEdge NDVI. Before applying the spectral index to the imagery raw mixel (“mixed pixel”) values, digital numbers(DN)] were converted into physically meaningful units to differentiate absorption reflectance spectra and immature Similium productivity based on habitat size.Linear regression was used to equate and quantitate band data to DN and the reflectance values which in the geospectral,sub-mixel, risk analysis was equivalent to removing the solar irradiance and the atmospheric path radiance in the object-based classifier. A radiometric calibration tool then calibrated the spaceborne sensor data to radiance and top-of-atmosphere (ToA) reflectance.Additionally,Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH®)removed the effects of multiscattering in the scene.We calculated the internal relative reflectance which normalized the image to a scene average spectrum. ENVI’s Log Residuals Correction Tool removed the instrument gain, topographic effects, and albedo effects from the reflectance, transmittance, wavelength emissitivities. The instantaneous fraction of direct beam radiation intercepted by the habitat canopy was calculated and described as fPAR = 1 - exp (-k (leaf area index)/cosθs) where the extinction coefficient k was a function of leaf angle distribution.We employed a successive progressive algorithm, a two stream radiative atmospheric transfer analyses, a geometric-optical model and a bidirectional reflectance distribution function to unmix the S. damnosum s.l., larval habitat,canopied endmembers.The non-parametric, residual, explanatorial, decomposed, sub-mixel estimators derived from the RapidEye™data were then used to construct a Boolean model.Therefater,the imaged larval habitat and its geospatially, ecohydrological, within-canopy pigments (e.g., chorophyll, zeathinins) were defined and a Red Edge,NDVI, endmember biosignature was decomposed in ENVI. An autocorrelation uncertainty matrix was deconvolved into combinations of the unmixed canopied endmembers.Subsequently, the NDVI, endmember biosignature, decomposed, canopied endmembers with its multiple ToA noise-adjusted coefficients were kriged in Geospatial Analyst of ArcGIS 10.3®to identify unknown, unsampled, prolific, S. damnosum s.l., georeferencable, larval habitats along a northern Ugandan riverine ecosystem. Of the forecasted prolific, shaded, larval habitats by the canopy model, 72% were found to contain S. damnosum s.l. larvae when field verified. The sensitivity of the test was 78.26 while the specifity was 100.