Analysis of classification accuracy for pre-filtered multichannel remote sensing data

Noise is one of the main factors degrading the quality of original multichannel remote sensing data and its presence influences classification efficiency, object detection, etc. Thus, pre-filtering is often used to remove noise and improve the solving of final tasks of multichannel remote sensing. Recent studies indicate that a classical model of additive noise is not adequate enough for images formed by modern multichannel sensors operating in visible and infrared bands. However, this fact is often ignored by researchers designing noise removal methods and algorithms. Because of this, we focus on the classification of multichannel remote sensing images in the case of signal-dependent noise present in component images. Three approaches to filtering of multichannel images for the considered noise model are analysed, all based on discrete cosine transform in blocks. The study is carried out not only in terms of conventional efficiency metrics used in filtering (MSE) but also in terms of multichannel data classification accuracy (probability of correct classification, confusion matrix). The proposed classification system combines the pre-processing stage where a DCT-based filter processes the blocks of the multichannel remote sensing image and the classification stage. Two modern classifiers are employed, radial basis function neural network and support vector machines. Simulations are carried out for three-channel image of Landsat TM sensor. Different cases of learning are considered: using noise-free samples of the test multichannel image, the noisy multichannel image and the pre-filtered one. It is shown that the use of the pre-filtered image for training produces better classification in comparison to the case of learning for the noisy image. It is demonstrated that the best results for both groups of quantitative criteria are provided if a proposed 3D discrete cosine transform filter equipped by variance stabilizing transform is applied. The classification results obtained for data pre-filtered in different ways are in agreement for both considered classifiers. Comparison of classifier performance is carried out as well. The radial basis neural network classifier is less sensitive to noise in original images, but after pre-filtering the performance of both classifiers is approximately the same.

Spatial demography of Calanus finmarchicus in the Irminger Sea

Continuous Plankton Recorder data suggest that the Irminger Sea supports a major proportion of the surface-living population of the copepod Calanus finmarchicus in the northern North Atlantic, but there have been few studies of its population dynamics in the region. In this paper, we document the seasonal changes in the demographic structure of C finmarchicus in the Irminger Sea from a field programme during 2001/2002, and the associations between its developmental stages and various apparent bio-physical zones. Overwintering stages were found widely at depth (>500 m) across the Irminger Sea, and surviving females were widely distributed in the surface waters the following spring. However, recruitment of the subsequent generation was concentrated around the fringes of the Irminger Sea basin, along the edges of the Irminger and East Greenland Currents, and not in the central basin. In late summer animals were found descending back to overwintering depths in the Central Irminger Sea. The key factors dictating this pattern of recruitment appear to be (a) the general circulation regime, (b) predation on eggs in the spring, possibly by the surviving GO stock, and (c) mortality of first feeding naupliar stages in the central basin where food concentrations appear to be low throughout the year. We compared the demographic patterns in 2001/2002 with observations from the only previous major survey in 1963 and with data from the Continuous Plankton Recorder (CPR) surveys. In both previous data sets, the basic structure of GO ascent from the central basin and G1 recruitment around the fringes was a robust feature, suggesting that it is a recurrent phenomenon. The Irminger Sea is a complex mixing zone between polar and Atlantic water masses, and it has also been identified as a site of sporadic deep convection. The physical oceanographic characteristics of the region are therefore potentially sensitive to climate fluctuations. Despite this, the abundance of C finmarchicus in the region, as measured by the CPR surveys, appears not to have responded to climate factors linked to the North Atlantic Oscillation Index, in contrast with the stocks in eastern Atlantic areas. We speculate that this may because biological factors (production and mortality), rather than transport processes are the key factors affecting the population dynamics in the Irminger Sea. (c) 2007 Elsevier Ltd. All rights reserved.

Toward autonomous measurements of photosynthetic electron transport rates: An evaluation of active fluorescence-based measurements of photochemistry

This study presents a methods evaluation and intercalibration of active fluorescence-based measurements of the quantum yield ( inline image) and absorption coefficient ( inline image) of photosystem II (PSII) photochemistry. Measurements of inline image, inline image, and irradiance (E) can be scaled to derive photosynthetic electron transport rates ( inline image), the process that fuels phytoplankton carbon fixation and growth. Bio-optical estimates of inline image and inline image were evaluated using 10 phytoplankton cultures across different pigment groups with varying bio-optical absorption characteristics on six different fast-repetition rate fluorometers that span two different manufacturers and four different models. Culture measurements of inline image and the effective absorption cross section of PSII photochemistry ( inline image, a constituent of inline image) showed a high degree of correspondence across instruments, although some instrument-specific biases are identified. A range of approaches have been used in the literature to estimate inline image and are evaluated here. With the exception of ex situ inline image estimates from paired inline image and PSII reaction center concentration ( inline image) measurements, the accuracy and precision of in situ inline image methodologies are largely determined by the variance of method-specific coefficients. The accuracy and precision of these coefficients are evaluated, compared to literature data, and discussed within a framework of autonomous inline image measurements. This study supports the application of an instrument-specific calibration coefficient ( inline image) that scales minimum fluorescence in the dark ( inline image) to inline image as both the most accurate in situ measurement of inline image, and the methodology best suited for highly resolved autonomous inline image measurements.