It is of vital interest to understand how cloud particles interact with ambient atmospheric radiation fields. We developed
a comprehensive analytical radiative transfer model for passive infrared remote sensing applicable to ground-based and
airborne sensors. We show the qualitative difference between simple non-scattering aerosols (pseudo vapor cloud) and
an aerosol cloud where scattering, absorption and emission occur. Simulations revealed two interesting observations:
aerosol cloud detection from an airborne platform may be more challenging than for a ground-based sensor, and the
detection of an aerosol cloud in emission mode is different from the detection of an aerosol cloud in absorption mode.
Performance of the matched filter and anomaly detection algorithms relies on the quality of the inverse sample
covariance matrix, which depends on sample size (number of vectors). The "RMB rule" provides the number of vectors
required to achieve a specific average performance loss of the matched filter. In this paper we extend the RMB rule to
provide the number of vectors needed to ensure a minimum performance loss (within a certain confidence). We also
review a general metric for covariance estimation accuracy based on the Wishart distribution and discuss anomaly
detector performance loss.
Hyperspectral imagery is often visualized as a three-dimensional image cube (two spatial dimensions and one spectral). When a hyperspectral sensor is set to stare at a fixed location a fourth dimension (time) is created as each new cube is sampled in time. In a ground-based stare-mode geometry each new cube has near perfect spatial registration with the previous data cubes. The problem with standard spectral-only hyperspectral detection algorithms is that they do not make effective use of temporal information. In this paper we combine temporal-differencing with temporal-spectral detection algorithms. The temporal-differencing allows for removal of most of the background prior to temporal-spectral detection. The temporal-spectral approaches combine temporal information with standard spectral-only statistical methods. By combining temporal-differencing with temporal-spectral information we are able to significantly improve detector performance and reduce the false alarm rate. We demonstrate the performance of these methods using data from the FIRST (Field-Portable Imaging Radiometric Spectrometer Technology). All the computer simulations and field data experiments show that temporal-differencing improves performance, inclusion of temporal-spectral information improves performance, and that the combination of temporal-differencing with temporal-spectral information greatly improves performance.
An algorithm using N-way analysis for the detection of multiple clouds in multi-wavelength lidar data is presented. Nway
analysis is a tool for algebraic manipulation of N-dimensional (ND) data arrays, and it allows for spatial (range),
temporal (time), and spectral (wavelength) information to be extracted simultaneously from 3D lidar data. The algorithm
tracks the spectral signal strength and location of each of the multiple clouds through time within the lidar measurements
via a method that is shown to be similar to multivariate anomaly detection. The method is data driven and can be applied
to arrays of any number of dimensions (e.g., polarization as the 4th dimension). Results of the algorithm for CO2 lidar
simulations of aerosol clouds are shown and discussed.
Recent experimental work has shown that passive systems such as hyperspectral FTIR and frequency-tunable IR cameras have application in detection of biological aerosols. This provided the motivation for a new detection technique, which we call Aerosol Ranging Spectroscopy (ARS), whereby a scattering LIDAR is used to augment passive spectrometer data to determine the location and optical depth of the aerosol plume. When the two systems are co-aligned or boresighted, the hybrid data product provides valuable enhancements for signal exploitation of the passive spectral data. This paper presents the motivation and theoretical basis for the ARS technique. A prototype implementation of an ARS system will also be described, along with preliminary results from recent outdoor field experiments.