Advanced Very High Resolution Radiometer Images Research Articles

Restoration and reconstruction of AVHRR images

Describes the design of small convolution kernels for the restoration and reconstruction of Advanced Very High Resolution Radiometer (AVHRR) images. The kernels are small enough to be implemented efficiently by convolution, yet effectively correct degradations and increase apparent resolution. The kernel derivation is based on a comprehensive, end-to-end system model that accounts for scene statistics, image acquisition blur, sampling effects, sensor noise, and postfilter reconstruction. The design maximizes image fidelity subject to explicit constraints on the spatial support and resolution of the kernel. The kernels can be designed with finer resolution than the image to perform partial reconstruction for geometric correction and other remapping operations. Experiments demonstrate that small kernels yield fidelity comparable to optimal unconstrained filters with less computation.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Volcano detection and monitoring using AVHRR data: the Krafla eruption, 1984

Abstract Abstract. Many volcanic eruptions go essentially unmonitored. Potentially the Advanced Very High Resolution Radiometer (AVHRR), with its global coverage, frequent return period, and sensitivity in the thermal infrared, represents a data source capable of monitoring surface volcanic activity unrecorded by ground observations or other satellite sensors. In this study an attempt is made to demonstrate this potential by extracting information for the 1984 eruption at Krafla, Iceland. Seven cloud-free AVHRR images were available for the 14 day period of eruptive activity. The surface activity was detectable as a major thermal anomaly in all three of the longer wavelength channels and was vigorous enough during one night-time pass to be detectable in the near-infrared channel (0.725-1.1μm). Channel 2 and 4 radiance data were used to calculate the size and temperature of sub-pixel heat sources within the lava flow field, and a heat source at 1050° C was estimated as occupying an area of approximately 240000 m2, which was distributed across 20 pixels. Detection and measurement of volcanic heat sources at such short wavelengths using low spatial resolution data has rarely been reported before. Field reports and maps were used to guide and confirm the analysis. Digital number variations within the anomaly could be related to various known features of the eruption. To monitor the eruption a weighted average method was derived and used to sharpen up the images, and the density sliced sharpened images enabled the development of the eruption to be mapped. Results compared well with field reports, suggesting that AVHRR and similar systems could be a useful source of data for monitoring eruptions where contemporaneous field observations are unavailable or incomplete.

Spacecraft attitude variations of NOAA-11 inferred from AVHRR imagery

Abstract A series of 1 km Advanced Very High Resolution Radiometer (AVHRR) images for the period of 1991-1992 are used to compute the attitude variations of the NOAA-11 spacecraft using ground control points (GCPs or land surface reference points) in conjunction with a recently developed precise AVHRR image navigation software package (Rosborough et al. 1994). The images were first corrected for the NOAA satellite clock drift and then roughly 2-3 images per month were used to compute roll, pitch and yaw to match the GCPs in each image. Higher frequency satellite attitude variations were monitored using a 2 to 3 day sampling interval for two month-long periods in May 1991 and February 1992. The pitch and yaw behaviour of NOAA-11 was found to oscillate about a mean value near 0·0, with an amplitude of ±0·002 radians for pitch and ±0·001 radians for yaw. The variability in the roll axis was characterized by an amplitude of ±0·0015 radians about a mean offset of −0·002 radians. The higher frequency month-long...

Rainfall area identification using satellite data

A technique is presented for determining rainfall areas through the simultaneous use of infrared and near-infrared satellite data. These data are a reduced-resolution version of the original AVHRR (Advanced Very High Resolution Radiometer) images from the satellite NOAA-12. The scheme developed was used in order to differentiate among clear sky (or with very few clouds present), raining clouds and non-raining clouds for a mid-latitude region over Argentina. The scheme was developed by applying clustering and discriminant analysis to the maxima obtained from 2-dimensional histograms in the infrared and near-infrared domain. Warm-season data during daylight hours were used to develop and test the scheme. It was tested on 9 dates, and comparison of the classification results with reports from ground stations was encouraging.

Variability in sea surface temperature around New Zealand from AVHRR images

Abstract Multi‐channel Advanced Very‐High Resolution Radiometer (AVHRR) images of sea surface temperature (SST) in the New Zealand region have been archived since 1989. The first 2 years of data are used to investigate the variability in sea surface temperature (SST) and its gradients around New Zealand. Empirical orthogonal function (EOF) analyses show that the annual cycle accounts for 92% of the variance of SST, but for less of the variance in SST gradient. The Southland and Subantarctic Fronts have an annual modulation in their strengths, accounting for about 20% of their variance; these modulations are out of phase, with the Southland Front being strongest in winter. The Subtropical Convergence shows variability that cannot be easily described by one mode, but appears to migrate meridionally with an annual period, being furthest north in March.

Observed biases in the conterminous U.S. AVHRR satellite data set: A North Carolina case study

Abstract The USGS EROS Data Center has produced a national data set for long‐term ecological monitoring termed the “Conterminous U.S. AVHRR (Advanced Very High Resolution Radiometer) Data Set” that includes biweekly maximum‐value composite (MVC) images of NDVI (Normalized Difference Vegetation Index) values, original five channels of calibrated AVHRR satellite data, image viewing and illumination geometry, date of observation, and ancillary data sets pertaining to landcover and political boundaries (Loveland et al., 1991; Eidenshink, 1992). The basic intent of the study was to evaluate the potential of the data set for broad‐scale, multitemporal landscape mapping by assessing the quality and sensitivity of the data set to support such applications. Potential biases existing in the data set were identified and analytical procedures suggested to deal with such biases. Results from analyses within the State of North Carolina suggest that the time series of the NDVI values is influenced by sensitivities to re...

Time scales of pattern evolution from cross‐spectrum analysis of advanced very high resolution radiometer and coastal zone color scanner imagery

We have selected square subareas (110 km on a side) from coastal zone color scanner (CZCS) and advanced very high resolution radiometer (AVHRR) images for 1981 in the California Current region off northern California for which we could identify sequences of cloud‐free data over periods of days to weeks. We applied a two‐dimensional fast Fourier transform to images after median filtering, (x, y) plane removal, and cosine tapering. We formed autospectra and coherence spectra as functions of a scalar wavenumber. Coherence estimates between pairs of images were plotted against time separation between images for several wide wavenumber bands to provide a temporal lagged coherence function. The temporal rate of loss of correlation (decorrelation time scale) in surface patterns provides a measure of the rate of pattern change or evolution as a function of spatial dimension. We found that patterns evolved (or lost correlation) approximately twice as rapidly in upwelling jets as in the “quieter” regions between jets. The rapid evolution of pigment patterns (lifetime of about 1 week or less for scales of 50–100 km) ought to hinder biomass transfer to Zooplankton predators compared with phytoplankton patches that persist for longer times. We found no significant differences between the statistics of CZCS and AVHRR images (spectral shape or rate of decorrelation). In addition, in two of the three areas studied, the peak correlation between AVHRR and CZCS images from the same area occurred at zero lag, indicating that the patterns evolved simultaneously. In the third area, maximum coherence between thermal and pigment patterns occurred when pigment images lagged thermal images by 1–2 days, mirroring the expected lag of high pigment behind low temperatures (and high nutrients) in recently upwelled water. We conclude that in dynamic areas such as coastal upwelling systems, the phytoplankton cells (identified by pigment color patterns) behave largely as passive scalars at the mesoscale and that growth, death, and sinking of phytoplankton collectively play at most a marginal role in determining the spectral statistics of the pigment patterns.

Correcting infrared satellite estimates of sea surface temperature for atmospheric water vapor attenuation

A new satellite sea surface temperature (SST) algorithm is developed that uses nearly coincident measurements from the microwave special sensor microwave imager (SSM/I) to correct for atmospheric moisture attenuation of the infrared signal from the advanced very high resolution radiometer (AVHRR). This new SST algorithm is applied to AVHRR imagery from the South Pacific and Norwegian seas, which are then compared with simultaneous in situ (ship based) measurements of both skin and bulk SST. In addition, an SST algorithm using a quadratic product of the difference between the two AVHRR thermal infrared channels is compared with the in situ measurements. While the quadratic formulation provides a considerable improvement over the older cross product (CPSST) and multichannel (MCSST) algorithms, the SSM/I corrected SST (called the water vapor or WVSST) shows overall smaller errors when compared to both the skin and bulk in situ SST observations. Applied to individual AVHRR images, the WVSST reveals an SST difference pattern (CPSST‐WVSST) similar in shape to the water vapor structure while the CPSST‐quadratic SST difference appears unrelated in pattern to the nearly coincident water vapor pattern. An application of the WVSST to week‐long composites of global area coverage (GAC) AVHRR data demonstrates again the manner in which the WVSST corrects the AVHRR for atmospheric moisture attenuation. By comparison the quadratic SST method underestimates the SST corrections in the lower latitudes and overestimates the SST in the higher latitudes. Correlations between the AVHRR thermal channel differences and the SSM/I water vapor demonstrate the inability of the channel difference to represent water vapor in the midlatitude and high latitudes during summer. Compared against drifting buoy data the WVSST and the quadratic SST both exhibit the same general behavior with relatively small differences with the buoy temperatures.

Clustering, randomness, and regularity in cloud fields: 3. The nature and distribution of clusters

This study applies the K‐nearest‐neighbor (KNN) statistic as well as a morphological filtering approach for the identification and analysis of cloud clusters. These approaches first are evaluated by application to synthetically generated normally distributed clusters within a random background. They are found to provide conservative estimates of cluster size while accurately estimating the percentage of clouds in clusters. In Landsat scenes, typical clusters are found to contain four to five clouds and typically two to three clouds per cluster in advanced very high resolution radiometer (AVHRR) data. In the Landsat imagery the KNN analysis finds that about 15% to 50% of the total cloud field participates in clusters. These clusters are found to be concentrated in less than 2% of the total cloud field area. In AVHRR scenes the KNN method yields from 1.5% to 16% of the cloud field participating in clusters, concentrated within 1% of the total cloud field area. Mean cluster radii are typically 300 m to 600 m in the Landsat scenes and less than 3 km in the AVHRR images. Both the cloud cluster centers and the non‐participating background clouds are randomly distributed within the cloud field. Less than 10% of the cloud clusters form superclusters (i.e., clusters of clusters). When clouds are segregated into classes of uniform size, the clustering signal weakens with increasing cloud size. When the scene as a whole is divided into 30‐km subregions, the clustering signal is relatively homogeneous throughout the scene in spite of significant fluctuations in cloud number density within the various subregions. These results show that the physical mechanism underlying this clustering signal, although producing clusters on scales less than 3 km, is operative over the entire field. A nearest‐neighbor angle statistic is applied to search for regularity at the subkilometer scale. The angle statistic results are consistent with normally distributed point processes and inconsistent with any form of regularity at this scale. Presently, no method is presented which is effective at probing the cloud field structure at scales intermediate to 2 km and 30 km. Therefore cloud field regularity cannot be ruled out at these intermediate scales.

A systematized approach to AVHRR image navigation

Accurate co-location and geo-registration of AVHRR (Advanced Very High Resolution Radiometer) is necessary for most applications involving these data. To obtain the necessary accuracy, most of the current systems available for AVHRR image rectification require corrections to errors arising from several different sources. This paper presents a software package which is able to correct registration errors from all sources using effective roll, pitch, and yaw values determined from ground control point offsets. These effective values are used to establish a baseline registration against which other models are compared. Spacecraft and sampling geometry are revisited in order to systemize and better define the procedure of AVHRR geo-registration. This systematic approach has resulted in a modular software package which has a registration accuracy of 1 satellite pixel, and is easily modified and improved.

A systematized approach to AVHRR image navigation

Accurate co-location and geo-registration of AVHRR (Advanced Very High Resolution Radiometer) is necessary for most applications involving these data. To obtain the necessary accuracy, most of the current systems available for AVHRR image rectification require corrections to errors arising from several different sources. This paper presents a software package which is able to correct registration errors from all sources using effective roll, pitch, and yaw values determined from ground control point offsets. These effective values are used to establish a baseline registration against which other models are compared. Spacecraft and sampling geometry are revisited in order to systemize and better define the procedure of AVHRR geo-registration. This systematic approach has resulted in a modular software package which has a registration accuracy of 1 satellite pixel, and is easily modified and improved.

Upper ocean response to Hurricane Gilbert

The evolving upper ocean response excited by the passage of hurricane Gilbert (September 14–19, 1988) was investigated using current and temperature observations acquired from the deployment of 79 airborne expendable current profilers (AXCPs) and 51 airborne expendable bathythermographs from the National Oceanic and Atmospheric Administration WP‐3D aircraft in the western Gulf of Mexico. The sea surface temperatures (SSTs), mixed layer depths, and bulk Richardson numbers were objectively analyzed to examine the spatial variability of the upper ocean response to Gilbert. Net decreases of the SSTs of 3°–4°C were observed by the profilers as well as by the airborne infrared thermometer (AIRT) along the flight tracks and advanced very high resolution radiometer (AVHRR) imagery. The AXCPs indicated a marked cooling from 29°C to about 25.5°C on September 17, 1988, which was about 1.2 inertial periods (IP) following storm passage. This pool of cooler water (3.5°) was located further downstream in the hurricane wake by September 19 (2.7 IP following the storm) as a result of the near‐inertial currents in the mixed layer. While there was a bias of about 0.6°C and 1.7°C between the in situ and AVHRR‐derived SSTs, respectively, both the AVHRR images and the objectively analyzed fields indicated a rightward bias in the upper ocean cooling that extended from the storm track to about 4Rmax (where Rmax, the radius of maximum winds, is equal to 50 km). The larger SST offset of 1.7°C was due to the difference between the time of the AVHRR image and the time of the aircraft experiment on September 19. The SSTs derived from the AVHRR images and the AIRT also indicated large gradients between the cold wake and the warm eddy in the central Gulf of Mexico. The mixed layer deepened by about 30–35 m on the right side of the track during the storm and 1.2 IP later, with little evidence of continued deepening afterward. The mixed layer current vectors demonstrate that a strong, near‐inertially rotating current was excited by the passage of Gilbert, with maxima of about 1–1.4 m s−1. The currents, observed during and subsequent to (1.2 IP) the storm, diverged from the storm track, whereas the mixed layer current vectors 2.7 IP after storm passage converged toward the track, with relative maxima of 0.8–1 m s−1. This alternating pattern of convergence and divergence of the mixed layer current was associated with the upwelling and downwelling cycles of the baroclinic response. Considerable current shear existed between the mixed layer and the thermocline currents in the cool wake between the storm track and the 4Rmax. Estimates of the bulk Richardson numbers ranged between 0.2 and 1.0 during Gilbert and at 1.2 IP, which suggests that enhanced current shears were responsible for some of the mixed layer deepening.

Sea surface temperature variability of the Iceland‐Faeroe front

Several advanced very high resolution radiometer (AVHRR) images with spatial resolution of 1.1–3.3 km, together with several concurrent aircraft‐deployed expendable bathythermograph (AXBT) surveys and conductivity‐temperature‐depth (CTD) stations, from spring 1989 are used to describe the Iceland‐Faeroe sea surface temperature (SST) front. In the AVHRR images, SST fronts are located by maximizing |∇SST|. Single, large gradient segments of the SST front do exist, with some exceeding 100 km in length, indicating a multiple frontal structure. These single frontal lines are also segments where |∇2SST| is small, and they can be followed uniquely by a single isotherm eastward from Iceland for a distance of 300 km. With a 35‐km sampled AXBT survey, two small subsurface cold eddies were located south of the surface front in an area 170 km × 270 km east of Iceland. From a May 1987 AVHRR image on 1.1‐km resolution, a population of seven such cold eddies are found between Iceland and the Faroes. They appear to be generated along the surface expression of the Iceland Faroes front and populate the northern slope of the Iceland‐Faroes Ridge. Historical data from towed high‐resolution instruments suggest that the cold eddies are ∼30–50 km in size and uplift the main thermocline by 150 m.

Automatic Adjustment of AVHRR Navigation

Abstract An automatic adjustment method of the Advanced Very High-Resolution Radiometer (AVHRR) images navigation has been developed for operational use at the Centre de Meteorologic Spatiale (CMS) in Lannion, France. Selected coastal landmarks are correlated with small windows of an actual AVHRR image using visible and near-infrared channels 1 and 2 for daytime images, and infrared channels 4 and 5 for nighttime images. The various stages of the method, generation of the reference landmarks, cloud discrimination, land-sea mask, and correlation between actual image and reference landmark are described. The CMS method has proven to be successful on 86% of the images of NOAA-11 day and night orbits from May to November 1990. The precision of the navigation after the adjustment is estimated to 0.8 pixel and 1.0 line when the adjustment is performed on the current orbit (58% of the cases), and to 1.0 pixel and 1.7 lines when the adjustment is performed on the preceding orbit (28% of the cams). The line and pi...

Temporal and Spatial Distribution of Savanna Fires in Senegal and the Gambia, West Africa, 1989-90, Derived From Multi-Temporal AVHRR Night Images

Quantitative data on fire regimes from the world's savannas have previously been difficult to obtain due to logistical and scientific reasons. This study addresses the issue for Senegal and The Gambia, West Africa. A bispectral satellite image based method for fire detection has recently been improved and made semi-automatic to provide information on fire regimes of savannas. The method uses thermal night time data from the Advanced Very High Resolution Radiometer (AVHRR) onboard the National Oceanic and Atrnospheric Administration (NOAA) series of weather satellites. Twenty-six near cloud-free AVHRR images have been classified using this method. Though they constitute a small temporal sample, some features of the seasonal periodicity and spatial variations of the region are suggested. Variations in firectivity at the four daily NOAA satellite passing times were recorded from a small field survey in The Gambia, and compared to data from the satellite images. Established relationships between reiative fire activity and data from the various NOAA passes have been used to estimate the maximum daily fire activity at the afternoon passing time. The seasonal variations for the whole study area and regions within are suggested by the estimated maximum values at the twenty-two dates. The locations of fire events in relation to locations of human activities and fuel loadings are examined and a brief summary of the main factors that regulate burning and fire regimes is given.

A comparison of sea ice parameters computed from advanced very high resolution radiometer and Landsat satellite imagery and from airborne passive microwave radiometry

The advanced very high resolution radiometer (AVHRR), flying on the NOAA polar‐orbiting weather satellites, can be used to map both sea ice concentration and ice edge, under cloud‐free conditions. In order to evaluate the accuracy and reliability of these AVHRR ice parameters, we have compared AVHRR‐derived sea ice parameters from the Bering Sea with those computed from nearly coincident (within 6 hours) Landsat multispectral scanner (MSS) imagery and from the aircraft multichannel microwave radiometer (AMMR) flown on the NASA DC‐8. The higher spatial and spectral resolutions of the Landsat data make them appropriate to use as a reference for the AVHRR inferred sea ice parameters while the lower‐altitude aircraft microwave sensor also provides a good source of comparison data. Both the AVHRR near‐infrared (channel 2) and thermal infrared (channel 4) images were used to compute ice concentration. Mean ice concentration differences between the channel 2 AVHRR and Landsat MSS data (Landsat‐AVHRR) ranged from −0.8 to 1.8% with a mean value of 0.5%. Root‐mean‐square (rms) differences ranged from 6.8 to 17.7%. Mean differences were larger for the channel 4 AVHRR ice concentrations ranging from −2.2 to 8.4% with rms differences from 8.6 to 26.8%. Mean differences between AVHRR channel 2 concentrations and the AMMR data (AMMR‐AVHRR) ranged from −19.7 to 18.9% while rms values went from 17.0 to 44.8%. Mean ice edge differences between the Landsat and AVHRR images were 1.8 km for channel 2 and 2.9 km for channel 4 with rms differences of 3.2 km and 4.4 km, respectively. The AVHRR ice edge was south and west of the Landsat edge.

Classification of Antarctic surfaces using AVHRR data: a multispectral approach

The mapping of sea, landice, sea ice and clouds, from Advanced Very High Resolution Radiometer (AVHRR) images taken over Antarctica in daylight is investigated and a classification scheme is proposed on the basis of thresholds retrieved from multispectral patterns of representative data. The scheme, which can be used for real time analysis of AVHRR images in scientific and logistic activities, gives satisfactory separation of different categories. Major misclassification occurs between ice clouds and land ice because of their very similar spectral signature in the AVHRR channels. Comparison of classified samples, obtained from visual inspection of images and from application of the scheme, exhibits a confusion matrix with accuracy A=92% over areas almost free from ice clouds.

Subpixel measurement of tropical forest cover using AVHRR data

Abstract Tropical forest assessment using data from the Advanced Very High Resolution Radiometer (AVHRR) may lead to inaccurate estimates of forest cover in regions of small subpixel forest or non-forest patches and in regions where the pattern of clearance is particularly convoluted. Test sites typifying these two patterns were chosen in Ghana and Rondonia, respectively. To capture the subpixel proportions of forest cover, a linear mixture model was applied to two AVHRR test images over the test sites. The model produced image outputs in which pixel intensities indicated the proporton of forest cover per km2. For comparison, supervised maximum likelihood classifications were also performed. The outputs were assessed against classified Landsat TM scenes, converted to proportions maps and coregistered to the AVHRR images. An empirical method was applied for determining the critical forest cover per km2 needed for an AVHRR pixel to be classified as forest. The critical values exceeded 50 per cent, indicating a tendency for AVHRR classification to underestimate forest cover. This was confirmed by comparing estimates of total forest cover obtained from the AVHRR and TM classifications. In the case of Ghana, a more accurate estimate of forest cover was obtained from the AVHRR mixture model than from the classification. Both mixture model outputs were found to be well correlated with those from Landsat TM. Further work should test the robustness of the approach adopted here when applied to much larger areas.

A study of a three-dimensional histogram using the NOAA AVHRR images

The development of the NOAA satellite Advanced Very High Resolution Radiometer (AVHRR) three-dimensional analysis system, in which the problem of drawing a three-dimensional histogram on a two-dimensional CRT was encountered, is discussed. In this case there are two factors: one is the coordinate position given by the brightness level of the three image channels of NOAA AVHRR, the other is the frequency of the points within the images. To decide the semi-optimum value for displaying the processed three-dimensional histogram, the authors tried to relate the frequency and the total number of pixels. It was found that the standard deviation is reasonable. In the analysis using this method, the differences between major components of the images are displayed in the three-dimensional histogram, allowing the identification of features like sea, land, and clouds by a clustering process.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Identification of fog with NOAA AVHRR images

The authors attempted to identify fog using image-processing data from the US National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR). AVHRR sensors measure radiation at visible, near-infrared, and infrared wavelengths. Attention is confined to daytime data from channel 1 (CH1), channel 3 (CH3), and channel 4 (CH4) for a detection range of lambda =0.58 approximately 0.68 mu m, lambda =3.55 approximately 3.93 mu m, and lambda =10.5 approximately 11.5 mu m, respectively. The authors have analyzed by image processing the data obtained from CH1, CH3, and CH4 by forming two- and three-dimensional histograms and enhancing them in color.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>