A refined classification approach by integrating Landsat Operational Land Imager (OLI) and RADARSAT-2 imagery for land-use and land-cover mapping in a tropical area

Producing accurate land-use and land-cover (LULC) mapping is a long-standing challenge using solely optical remote-sensing data, especially in tropical regions due to the presence of clouds. To supplement this, RADARSAT images can be useful in assisting LULC mapping. The fusion of optical and active remote-sensing data is important for accurate LULC mapping because the data from different parts of the spectrum provide complementary information and often lead to increased classification accuracy. Also, the timeliness of using synthetic aperture radar (SAR) fills information gaps during overcast or hazy periods. Therefore, this research designed a refined classification procedure for LULC mapping for tropical regions. Determining the best method for mapping with a specific data source and study area is a major challenge because of the wide range of classification algorithms and methodologies available. In this study, different combinations and the potential of Landsat Operational Land Imager (OLI) and RADARSAT-2 SAR data were evaluated to select the best procedure for LULC classification. Results showed that the best filter for SAR speckle reduction is the 5 × 5 enhanced Lee. Furthermore, image-sharpening algorithms were employed to fuse Landsat multispectral and panchromatic bands and subsequently these algorithms were analysed in detail. The findings also confirmed that Gram–Schmidt (GS) performed better than the other techniques employed. Fused Landsat data and SAR images were then integrated to produce the LULC map. Different classification algorithms were adopted to classify the integrated Landsat and SAR data, and the maximum likelihood classifier (MLC) was considered the best approach. Finally, a suitable classification procedure was designed and proposed for LULC as mapping in tropical regions based on the results obtained. An overall accuracy of 98.62% was achieved from the proposed methodology. The proposed methodology is a useful tool in industry for mapping purposes. Additionally, it is also useful for researchers, who could extend the method for different data sources and regions.     

Symbolic target detection in SAR imagery via rotationally invariant-weighted feature extraction

This article introduces the application of a physics-based symbolic image partitioning method to detect targets in synthetic aperture radar (SAR) imagery. ‘Targets’ in this case refer to vehicular objects which produce a distinct radar return pattern, and have spatial characteristics that are known a priori. The proposed Rotationally Invariant Symbolic Histogram (RISH) detection method co-analyses both target and speckle statistics, and significantly reduces computational requirements by partitioning the data into a discrete number of state representations. RISH requires only one pass for robust detection, unlike other SAR detection methods which rely on difference metrics calculated using multiple passes. To improve performance in high-resolution data, RISH uses a weighted feature extraction algorithm to avoid the common requirement of processing each pixel of the image equally. The weighted structure extracts geometrically undefined and rotationally invariant target features. This article details the analysis of 24 experimentally obtained very high-frequency (VHF)-band SAR magnitude images using this novel approach to SAR target detection. In localizing small (~8.4 m²) foliage-concealed targets, without the aid of pre-processing, this method results in high performance characteristics (90% true positive) with a low Type-II error rate of 6.4 false alarms per 1 × 10⁶ m². With the addition of change detection, RISH lowers the error rate by 85%.     

Retrieval of phenological stages of onion fields during the first year of growth by means of C-band polarimetric SAR measurements

The phenological stages of onion fields in the first year of growth are estimated using polarimetric observables and single-polarization intensity channels. Experiments are undertaken on a time series of RADARSAT-2 C-band full-polarimetric synthetic aperture radar (SAR) images collected in 2009 over the Barrax region, Spain, where ground truth information about onion growth stages is provided by the European Space Agency (ESA)-funded agricultural bio/geophysical retrieval from frequent repeat pass SAR and optical imaging (AgriSAR) field campaign conducted in that area. The experimental results demonstrate that polarimetric entropy or copolar coherence when used jointly with the cross-polarized intensity allows unambiguously distinguishing three phenological intervals.     

Evaluation of hurricane wind speed retrieval from cross-dual-pol SAR

Understanding and forecasting hurricanes are important components of weather prediction and climate studies. A critical concern is accuracy in hurricane wind speed estimates, especially in areas over the ocean where in situ measurements are sparse. Moreover, for very intense hurricanes, remotely sensed ocean surface winds generally lack accuracy. Recent studies on cross-polarization RADARSAT-2 synthetic aperture radar (SAR) data show very promising capability for high wind speed retrieval. The monotonic increase of cross-pol radar backscattered intensity with wind speed, and less sensitive dependence on wind direction, makes it superior to co-pol SAR measurements for operational application. Before further application of the methodology, it is important to evaluate the capability of hurricane forced wind speed retrieval from the promising, simple cross-pol SAR data. In this study, we apply a newly developed wind retrieval model to hurricanes using VH polarization dual-pol mode (VH dual-pol) RADARSAT-2 ScanSAR images. Validation of SAR wind retrievals is via surface wind analysis data from the Hurricane Research Division (HRD) of the National Oceanic and Atmospheric Administration (NOAA) and airborne stepped frequency microwave radiometer (SFMR) measurements. We found that compared to the co-polarization RADARSAT-2 SAR measurements, retrieved wind speeds from cross-polarization mode SAR have better overall accuracy and are more consistent with expected hurricane structures. Moreover, the cross-polarization model for VH dual-pol-retrieved winds does not appear to exhibit the speed ambiguity problem, which is one of the main obstacles for co-polarization retrieval of hurricane winds. SAR-retrieved winds contain detailed wind structures of the hurricane eyewall which are potentially of value for ongoing improvements in numerical models of hurricanes, air–sea interactions, and climate change. Special attention must be paid to biases caused by precipitation in order to reduce remaining errors in SAR retrieval of hurricane winds; precipitation has not been explicitly considered by current geophysical model functions.     

An approach to evaluate and monitor RISAT-1 SAR from level-0 raw data

Data quality evaluation of synthetic aperture radar (SAR) products is essential with respect to mission performance and reliability of the data products for users. The generation of any value-added SAR product starts with a level-0 raw data product, so it is vital firstly to analyse and monitor the raw data. In other words, any inconsistency or error occur because of data, the SAR processor requires to ‘quality evaluate’ the raw data first to ensure that the higher level of data products meets the user specifications. Many parameters can be evaluated from the raw data per se before SAR processing. Here, an independent approach is formulated to evaluate the long-term stability and performance of the SAR system from raw data. Quality parameters such as echo data statistics in terms of bias in the mean of Inphase (I) and Quadrature in phase (Q), power imbalance, phase imbalance, antenna pattern in elevation for the qualified swath, Doppler centroid estimation, and chirp phase and amplitude stability were evaluated and monitored for various scenes of an orbit and characterized in detail for different beams covering the near to far range. In the Radar Imaging Satellite (RISAT-1), the SAR beam is formed by a number of transmit/receive modules (TRMs) (total number of active TRMs depends on the look angle), and the consolidated effect of TRMs can be observed from the raw data instead of individual TRMs, which enables examination of the overall behaviour of beams, from near to far range in different orbits, for temporal stability. In this article, for the first time, a novel approach is taken to observe and evaluate the antenna pattern for the 3 dB qualified swath from range-uncompressed raw data and to relatively monitor the different beams from level-0 raw data for the qualified swath, for both extended homogeneous and non-homogeneous targets. Furthermore, the approach is justified in that it is not necessary to perform range compression, and also one can use scenes with heterogeneous targets to estimate and monitor the beamwidth parameter of the antenna pattern. This article focuses on the results obtained from RISAT-1 level-0 raw data from various beams covered in strip-map mode. The methodology to evaluate the RISAT-1 SAR system from raw data is discussed. Analysis is carried out with respect to different beams (60 beam data of different orbits either side, covering near to far nadir swath coverage) of fine resolution strip-map-1 (FRS-1) mode in circularly transmitted and linearly received polarization, with a defined swath of 25 km with 3 m × 2 m (azimuth × range) resolution. The results of evaluation and monitoring analysis show consistency in the identified parameters observed over time for different beams, and they are within specifications.     

Mapping growing stock volume and biomass carbon storage of larch plantations in Northeast China with L-band ALOS PALSAR backscatter mosaics

ABSTRACT

Reliable spatial information on growing stock volume (GSV) and biomass is critical for creating management strategies for plantation forests. This study developed empirical models to map the GSV and biomass of larch plantations (LPs) in Northeast China (1.25 million km² total area) by integrating L-band synthetic aperture radar (SAR) data with ground-based survey data. The best correlation model was used to map the GSVs and biomasses of LPs. The total GSV and biomass carbon storage were estimated at 224.3 ± 59.0 million m³ and 113.0 ± 29.7 × 10¹² g C with average densities of 85.1 m³ ha^?1 and 42.9 10⁶ g × C ha^?1, respectively, over a total area of 2.64 million ha. The saturation effect of SAR was determined beyond 260 m³ ha^?1, which was expected to influence the estimations for a small proportion of the study area. The accuracy of the estimations has limitations mainly due to the uncertainties in the GSV inventories, discrimination of natural larch and the SAR dataset. Based on the mapping results of the GSVs of LPs, a planning strategy for multipurpose management was tentatively proposed. This study can inform policies and management practices to assure broader and sustainable benefits from plantation forests in the future.     

Accurate crop-type classification using multi-temporal optical and multi-polarization SAR data in an object-based image analysis framework

Accurate crop-type classification is a challenging task due, primarily, to the high within-class spectral variations of individual crops during the growing season (phenological development) and, second, to the high between-class spectral similarity of crop types. Utilizing within-season multi-temporal optical and multi-polarization synthetic aperture radar (SAR) data, this study introduces a combined object- and pixel-based image classification methodology for accurate crop-type classification. Particularly, the study investigates the improvement of crop-type classification by using the least number of multi-temporal RapidEye (RE) images and multi-polarization Radarsat-2 (RS-2) data utilized in an object- and pixel-based image analysis framework. The method was tested on a study area in Manitoba, Canada, using three different classifiers including the standard Maximum Likelihood (ML), Decision Tree (DT), and Random Forest (RF) classifiers. Using only two RE images of July and August, the proposed method results in overall accuracies (OAs) of about 95%, 78%, and 93% for the ML, DT, and RF classifiers, respectively. Moreover, the use of only two quad-pol images of RS-2 of June and September resulted in OAs of 92%, 75%, and 90% for the ML, DT, and RF classifiers, respectively. The best classification results were achieved by the synergistic use of two RE and two RS-2 images. In this case, the overall classification accuracies were 97% for both ML and RF classifiers. In addition, the average producer’s accuracies of 95% and 96% were achieved by the ML and RF classifiers, respectively, whereas the average user accuracy was 94% for both classifiers. The results indicated promising potentials for rapid and cost-effective local-scale crop-type classification using a limited number of high-resolution optical and multi-polarization SAR images. Very accurate classification results can be considered as a replacement for sampling the agricultural fields at the local scale. The result of this very accurate classification at discrete locations (approximately 25 × 25 km frames) can be applied in a separate procedure to increase the accuracy of crop area estimation at the regional to provincial scale by linking these local very accurate spatially discrete results to national wall-to-wall continuous crop classification maps.     

Applying Spatial Augmented Reality to Anti-Smoking Message: Focusing on Spatial Presence,Negative Emotions,and Threat Appraisal

As smoking has emerged as a health-risk behavior, communication scholars and practitioners have put many efforts into finding out ways to achieve smoking cessation. In this research, participants (N = 57) were randomly assigned to a spatial augmented reality (SAR) condition (3D projection mapping) and 2D flat screen to be exposed to an anti-smoking message. This research provides insightful evidence that the effects of SAR on people’s behavioral intention to spread anti-smoking messages online could be explained by spatial presence and negative emotions. Implications for research on the potential of SAR in terms of emotions and online viral behavioral intentions are discussed.     

The influence of filtration and decomposition window size on the threshold value and accuracy of land-cover classification of polarimetric SAR images

In this study we use ALOS PALSAR satellite data to classify land cover using a decision tree algorithm. We apply polarimetric decomposition methods to coherence and covariance matrices obtained from the data and then use threshold values to classify terrain. We evaluate the influence of speckle filter and decomposition window sizes on the threshold value used in the decision algorithm and on the accuracy of the classification. We also study the sensitivity of the classification to the accuracy of the threshold value.

First, we processed a fully polarimetric Synthetic Aperture Radar (SAR) L-band image using different sizes of speckle filtration and decomposition window (3 × 3 pixels, 5 × 5, 7 × 7, 9 × 9), and the decomposition methods available in PolSARPro software. We evaluated these methods and chose the most efficient. Then we developed a simple hierarchical classification scheme based on threshold values. In the first step we divided the terrain into smooth and rough areas and then separated these into more detailed subclasses (water and agriculture, and forest and urban) which correspond to smooth and rough areas, respectively. A more detailed analysis separated continuous and discontinuous urban fabric and deciduous and coniferous forests. The maximum overall accuracy of the classification was 86.1% for the four main land cover classes, and 80.4% for the six more detailed classes. The accuracy of the classification dropped by about 10% when non-optimal window sizes were used in image filtration or decomposition.     

Two‐scale surface deformation analysis using the SBAS‐DInSAR technique: a case study of the city of Rome,Italy

We have exploited the capability of the differential synthetic aperture radar (SAR) interferometry (DInSAR) technique, referred to as Small BAseline Subset (SBAS) approach, to analyse surface deformation at two distinct spatial scales: a low resolution, large scale, and a fine resolution, local scale. At the large scale, the technique investigates DInSAR data with a ground resolution of the order of 100 m×100 m and leads to generate mean deformation velocity maps and associated time series for areas extending to some thousands of square kilometres. At the local scale, the technique exploits the SAR images at full spatial resolution (typically of the order of 5 m×20 m), detecting and analysing localized deformation phenomena. The study is focused on the city of Rome, Italy, and we used the ERS‐1/2 satellite radar data relevant to the 1995–2000 time period. The presented results demonstrate the capability of the SBAS approach to retrieve, from the low‐resolution DInSAR data, large‐scale deformation information leading to identify several sites affected by significant displacements. Our analysis permitted us to conclude that a major contribution to the detected displacements is due to the consolidation of the alluvial soils present in the area, mostly enforced by the buildings' overload. Furthermore, in a selected area, a detailed analysis was carried out by exploiting the full resolution DInSAR data. In this case we investigated deformation phenomena at the scale of single buildings. As key result we showed that differential displacements of few mm a^?1, affecting single man‐made structures or building complexes, could be detected, thus allowing to identify sites that may potentially be involved in critical situations.     

Growth monitoring and classification of rice fields using multitemporal RADARSAT-2 full-polarimetric data

Spaceborne synthetic aperture radar (SAR) can be used for agricultural monitoring. In this study, three single-polarimetric and four full-polarimetric observation data sets were analysed. A rice paddy field in northern Japan was used as the study site; the data for this site were obtained using RADARSAT-2, which carries a full-polarimetric C-band SAR. Soybean and grass fields were also present within the paddy fields. The temporal change in the backscattering coefficient of the rice paddy fields for the single-polarization data agreed with the temporal change obtained for a rice growth model based on radiative transfer theory. A three-component decomposition approach was applied to the full-polarimetric data. With each rice growth stage, the volume scattering component ratio increased, whereas the surface scattering component ratio generally decreased. The soybean and grass fields showed a smaller double-bounce scattering component than the rice fields for all the acquired data. The results of this study show that multitemporal observation by full-polarimetric SAR has great potential to be utilized for estimating rice-planted areas and monitoring rice growth.     

Lake-area mapping in the Tibetan Plateau: an evaluation of data and methods

Lake area derived from remote-sensing data is a primary data source, because changes in lake number and area are sensitive indicators of climate change. These indicators are especially useful when the climate change is not convoluted with a signal from direct anthropogenic activities. The data used for lake-area mapping is important, to avoid introducing unnecessary uncertainty into long-term trends of lake-area estimates. The methods for identifying waterbodies from satellite data are closely linked to the quality and efficiency of surface-water differentiation. However, few studies have comprehensively considered the factors affecting the selection of data and methods for mapping lake area in the Tibetan Plateau (TP), nor of evaluating their consequences. This study tests the dominant data sets (Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) data) and the methods for automated waterbody mapping on 14 large lakes (>500 km²) distributed across different climate zones of the TP. Seasonal changes in lake area and data availability from Landsat imagery are evaluated. Data obtained in October is optimal because in this month the lake area is relatively stable. The data window can be extended to September and November if insufficient data is available in October. Grouping data into three-year bins decreases the effects of year-to-year seasonal variability and provides a long-term trend that is suitable for time series analysis. The Landsat data (Multispectral Scanner, MSS; Thematic Mapper, TM; Enhanced Thematic Mapper Plus, ETM+; and Operational Land Imager, OLI) and MODIS data (MOD09A1) showed good performance for lake-area mapping. The Otsu method is used to determine the optimal threshold for distinguishing water from non-water features. Several water extraction indices, namely NDWI_McFeeters, NDWI_Xu, and AWEI_non-shadow, yielded high overall classification accuracy (92%), kappa coefficient (0.83), and user’s accuracy (~90%) for lake-water classification using Landsat data. The MODIS data using NDWI_McFeeters and NDWI_Xu showed consistent lake area (r² = 0.99) compared with Landsat data on the corresponding date with root mean square error (RMSE) values of 86.87 and 103.33 km² and mean absolute error (MAE) values of 25.7 and 29.04 km², respectively. The MODIS data is suitable for great lake mapping, which is the case for the large lakes in the TP. Although automated water extraction indices exhibited high accuracy in separating water from non-water, visual examination and manual editing are still necessary. Combined with recent Chinese high-resolution satellites, these remotely sensed imageries will provide a wealth of data for studies of lake dynamics and long-term lake evolution in the TP.     

Detecting a moorland wildfire scar in the Peak District,UK, using synthetic aperture radar from ERS-2 and Envisat ASAR

Wildfires occur annually in UK moorland environments, especially in drought years. They can be severely damaging to the ecosystem when they burn deep into the peat, killing ground-nesting birds and releasing CO₂ into the atmosphere. Synthetic aperture radar (SAR) was evaluated for detecting the 18 April 2003 Bleaklow wildfire scar (7.4 km²). SAR’s ability to penetrate cloud is advantageous in this inherently overcast area. SAR can provide fire scar boundary information which is otherwise labour intensive to collect in the field using a global positioning system (GPS). This article evaluates the potential of SAR intensity and InSAR coherence to detect a large peat moorland wildfire scar in the Peak District of northern England. A time-series of pre-fire and post-fire ERS-2 and advanced synthetic aperture radar (ASAR) Single Look Complex (SLC) data were pre-processed using SARScape 4.2 to produce georeferenced greyscale images. SAR intensity and InSAR coherence values were analysed against Coordinate Information on the Environment (CORINE) land‐cover classes and precipitation data. SAR intensity detected burnt peat well after a precipitation event and for previous fire events within the CORINE peat bog class. For the 18 April 2003 fire event, intensity increased to 0.84 dB post-fire inside the fire scar for the peat bog class. InSAR coherence peaked post-fire for moors and heathland and natural grassland classes inside the fire scar, but peat bog exposed from previous fires was less responsive. Overall, SAR was found to be effective for detecting the Bleaklow moorland wildfire scar and monitoring wildfire scar persistence in a degraded peat landscape up to 71 days later. Heavy precipitation amplified the SAR fire scar signal, with precipitation after wildfires being typical in UK moorlands. Further work is required to disentangle the effects of fire size, topography, and less generalized land‐cover classes on SAR intensity and InSAR coherence for detecting fire scars in degraded peat moorlands.     

James Sanborn’s Kryptos and the matrix encryption conjecture

James Sanborn’s sculpture, Kryptos, commissioned by the CIA, consists (in part) of four enciphered messages. These have attracted a tremendous amount of attention, and only the first three have been solved. In the present article, the authors provide a brief summary of each cipher and examine evidence that the fourth makes use of matrix encryption. They also provide results of brute force attacks for the 2 × 2 and 3 × 3 cases. Sanborn’s latest hint was of great value in testing these possibilities. Room for further testing is indicated for those wishing to continue the attack.     

星载高分辨率合成孔径雷达影像纠正技术

为检测TERRASAR、COSMO SkyMed、RADARSAT-2等星载高分辨率合成孔径雷达影像(SAR)在土地利用调查监测中的适用性,该文针对高分辨率SAR数据和产品特性,提出了控制点选取方法,分析了不同纠正模型的应用效果。试验表明高分辨率SAR几何纠正一般需要10～15个控制点,1m聚束模式纠正中误差约3m～5m,3m条带模式纠正中误差约5m～8m,分别满足1∶1万和1∶2.5万土地调查监测几何精度要求。研究结果为构建基于高分辨率SAR数据土地利用调查监测应用技术流程和促进高分辨率极化SAR数据业务化应用奠定了基础。     

Changes in the wetland vegetation growth patterns in large lakes on the Yangtze Plain

In this study, the vegetation emergence times (VET), an important phenological characteristics, were obtained for 25 large lakes on the Yangtze Plain between 2001 and 2014. This was carried out by extracting the normalized difference vegetation index (NDVI) time series from the moderate-resolution imaging spectroradiometer (MODIS) data using a decision tree method. This is the first comprehensive documentation of the changes in temporal and spatial distribution in wetland vegetation phenology for large lakes on the Yangtze Plain. The results showed that considerable changes in the VET occurred in 25 wetland ecosystems in the Yangtze Plain. Specifically, 76% of the lakes showed delayed trends in the VET, and 32% of them were statistically significant (p < 0.05). In contrast, 24% of the lakes displayed advanced trends in the VET and 17% of them were statistically significant (p < 0.05) over the past 14 years. An analysis of the driving factors of VET revealed that a VET change was more sensitive to temperature and sunshine duration than to precipitation for most of the lakes. The temperature in 1–2 months before VET had great effect on the vegetation growth, while such a pattern was not evident for sunshine duration for 5 months before VET. Furthermore, the amounts of chemical fertilizers used in nearby farmlands have also played an important role in the vegetation growth for some of the lakes. This record of change in vegetation phenology provides critical information for wetland ecosystem monitoring in the Yangtze Plain.     

An expert system for predicting Manning’s roughness coefficient in open channels by using gene expression programming

Manning’s roughness coefficient (n) has been widely used in the estimation of flood discharges or depths of flow in natural channels. Accurate estimation of Manning’s roughness coefficient is essential for the computation of flow rate, velocity. Conventional formulae that are greatly based on empirical methods lack in providing high accuracy for the prediction of Manning’s roughness coefficient. Consequently, new and accurate techniques are still highly demanded. In this study, gene expression programming (GEP) is used to estimate the Manning’s roughness coefficient. The estimated value of the roughness coefficient is used in Manning’s equation to compute the flow parameters in open-channel flows in order to carry out a comparison between the proposed GEP-based approach and the conventional ones. Results show that computed discharge using estimated value of roughness coefficient by GEP is in good agreement (±10%) with the experimental results compared to the conventional formulae (R ² = 0.97 and RMSE = 0.0034 for the training data and R ² = 0.94 and RMSE = 0.086 for the testing data).     

Sensitivity study of Radarsat-2 polarimetric SAR to crop height and fractional vegetation cover of corn and wheat

Increasing studies have been conducted to investigate the potential of polarimetric synthetic aperture radar (SAR) in crop growth monitoring due to the capability of penetrating the clouds, haze, light rain, and vegetation canopy. This study investigated the sensitivity of 16 parameters derived from C-band Radarsat-2 polarimetric SAR data to crop height and fractional vegetation cover (FVC) of corn and wheat. The in-situ measured crop height and FVC were collected from 29 April to 30 September 2013, at the study site in southwest Ontario, Canada. A total of 10 Radarsat-2 polarimetric SAR images were acquired throughout the same growing season. It was observed that at the early growing stage, the corn height was strongly correlated with the SAR parameters including HV (R² = 0.88), HH-VV (R² = 0.84), and HV/VV (R² = 0.80), and the corn FVC was significantly correlated with HV (R² = 0.79) and HV/VV (R² = 0.92), but the correlation became weaker at the later growing stage. The sensitivity of the SAR parameters to wheat variables was very low and only HV and Yamaguchi helix scattering showed relatively good but negative correlations with wheat height (R² = 0.57 and R² = 0.39) at the middle growing stage. These findings indicated that Radarsat-2 polarimetric SAR (C-band) has a great potential in crop height and FVC estimation for broad-leaf crops, as well as identifying the changes in crop canopy structures and phenology.     

Distorted spatial warping to compress large,high-resolution rasters for remote-sensing classification

Decisions made early in the data preparation phases of remote-sensing classification projects set fundamental limits on the value to society of the final products. The often-used approach of degrading/down-sampling high-resolution (e.g. 1 m pixel size) imagery to match lower-resolution data (e.g. Landsat 30 m) through averaging or majority-rule solves the problem of aligning pixels across bands of differing resolution, but does so by forgoing all ability to detect features smaller than 30 m in addition to potentially discarding up to 99% of the information content of the high-resolution data. The alternative of up-sampling coarser-resolution data into smaller-sized synthetic pixels creates its own set of problems, including potentially enormous file sizes, likely absence of meaningful variation over small spatial scales (which may generate matrix singularities fatal to the maximum likelihood classifier), and no assurance of meaningful improvement in classification accuracy despite guaranteed increases in computational time and resource requirements. We propose a new ‘warped space compression technique’ as a variation of vector quantization that analyses local variability in the finest-resolution data available to define acceptable pixel-based neighbourhood (N × N) sizes over which data can be averaged while minimizing overall information loss. Alternative neighbourhoods are aligned so that nine smaller ones nest within each progressively larger one as 3 × 3 squares, resulting in local data compression options of 3 × 3 (ninefold), 9 × 9 (81-fold), 27 × 27 (729-fold), and 81 × 81 (6561-fold). Our transformation process to ‘warped space’ created spatially distorted images with jagged east edges and little visually discernible relationship to the original data. We achieved compressions of 48- to 138-fold in disc storage and 292- to 785-fold in actual numbers of non-null pixels through our choice of cut-off values for accepting 3 × 3, 9 × 9, 27 × 27, or 81 × 81 neighbourhoods of tolerable variability, while otherwise retaining full (1 m) resolution data in regions three cells wide by three cells high. Medium-resolution data (e.g. Landsat 30 m) can be translated into the warped space defined by high-resolution data and composited with it for conducting remote-sensing classifications. When applied to a 71-band, 55-class remote-sensing classification of a 25,500 km² region centred on the Willamette Valley of Oregon, USA, classification accuracy increased from 64.4% in normal space to 71.3% in warped space. Unsupervised classification in warped space identified several additional categories that could be appended to the 55 existing ground-truth classes, leading to further increases in accuracy. Warped-space compression may be particularly beneficial for ecological studies where it could maintain high resolution in features of interest such as riparian buffers without creating exorbitantly large data files.