Latent heat flux estimation using trapezoidal relationship between MODIS land surface temperature and fraction of vegetation – application and validation in a humid tropical region

The present study was taken up with the objective of developing a methodology for estimation of actual evapotranspiration (AET) using only satellite data. Accordingly, an algorithm based on the popular Priestley–Taylor method was developed. While previous studies have assumed a triangular relationship between land surface temperature (LST) and fraction of vegetation (FV) to calculate the Priestley–Taylor parameter (φ), a trapezoidal relationship was adopted in the present study to enable applications in forested regions in the humid tropics. The developed algorithm was applied to the humid tropical Mae Klong region, Thailand, and latent heat flux (λ_ET) estimates were validated with measurements made at a flux tower located at the centre of the region. Moderate Resolution Imaging Spectroradiometer (MODIS) remote sensing satellite data products corresponding to the study area were used to derive various inputs required by the algorithm. Comparison of estimated and measured fluxes on five cloud-free days in 2003 yielded root mean square error (RMSE) of 64.73 W m^?2 which reduced to 18.65 W m^?2 when one day was treated as an outlier. The methodology developed in this study derived inputs only from satellite imagery and provided reasonably accurate estimates of latent heat flux at a humid tropical location.     

Multi-dimensional histogram-based information capacity analysis of urban heat island effect using Landsat 8 data

The vegetation and impervious surface area (ISA) are the most key indicators for the urban heat islands (UHI). This study used the normalized difference vegetation index (NDVI) as the indicator of vegetation abundance, the modified normalized difference impervious surface index (MNDISI) as the indicator of impervious surface fraction to estimate the land surface temperature (LST)–vegetation relationship and LST–ISA relationship. The land surface cover types were obtained by classification and regression tree (CART). Results demonstrate and verify that LST possessed strong negative correlations with NDVI and positive correlations with MNDISI at various spatial resolutions (30–960 m). Both correlation coefficients reached their strongest points at 30 m resolution, which is believed to be the operational scale of LST, NDVI and MNDISI. Further, information capacity (IC), as a spatial index, is used to characterize the spatial pattern of UHI. Results show that the IC of LST (LST_IC) possessed strong positive correlations with the IC of NDVI (NDVI_IC) and the IC of MNDISI (MNDISI_IC). It is suggested that the spatial pattern of UHI has a direct correspondence with the vegetation and ISA.     

The suitability of multi-temporal web-enabled Landsat data NDVI for phenological monitoring – a comparison with flux tower and MODIS NDVI

Three years of flux-tower-derived normalized difference vegetation index (NDVI) data were compared with contemporaneous 30 m web-enabled Landsat data (WELD) and with 500 m Moderate-Resolution Imaging Spectroradiometer (MODIS) nadir bidirectional reflectance distribution function-adjusted reflectance (NBAR) NDVI data to assess the relative suitability of these different resolutions of freely available satellite data for phenological monitoring. Comparisons were made at two flux tower sites in the United States with average to above average cloud cover. The WELD 30 m NDVI data were found to have higher correlation with the flux tower NDVI data than the MODIS 500 m NBAR NDVI data. The dates of vegetation green-up onset and maximum-greenness onset, derived using an established phenological metric extraction methodology, were generally closer between the flux tower and WELD NDVI data than between the flux tower and MODIS NBAR data. These results indicate that the WELD NDVI time series is suitable for 30 m scale phenological monitoring.     

Validation of satellite-derived tropical cyclone heat potential with in situ observations in the North Indian Ocean

Tropical cyclone heat potential (TCHP) is an important ocean parameter influencing cyclones and hurricanes. The best approach for computing TCHP is to use in situ measurements. However, since in situ data have both spatial and temporal limitations, there is a need for satellite-based estimations. One potential solution is to use sea surface height anomalies (SSHAs) from altimeter observations. However, any estimation derived from satellite measurements requires extensive regional validation. In this letter, we compare satellite-derived TCHP values with those estimated using in situ measurements of the North Indian Ocean collected during 1993–2009. All the available measurements collected from the conductivity temperature and depth (CTD) profiler, expendable CTD profiler (XCTD), bathythermograph (BT), expendable BT (XBT) and Argo floats were used to estimate in situ derived TCHP values. TCHP estimations from satellite observations and in situ measurements are well correlated, with coefficient of determination R ² of 0.65 (0.76) and a scatter index (SI) of 0.33 (0.25) on a daily (monthly) basis for the North Indian Ocean.     

Improved NDBI differencing algorithm for built-up regions change detection from remote-sensing data: an automated approach

Remote-sensing-based multi-temporal satellite images allow the mapping of changes in built-up areas over time to illustrate the urban development. An improved normalized difference built-up index (NDBI) has recently been promoted as a more effective algorithm to identify built-up regions, compared with the conventional NDBI approach. The conventional NDBI algorithm assumes that difference between the values of the binary NDBI and binary normalized difference vegetation index (NDVI) would indicate built-up areas. The modified NDBI approach improves this assumption by assigning higher positive difference values of continuous NDBI and NDVI to built-up region using an optimal threshold value. This article extends the concept of improved NDBI approach to automate the extraction of built-up changes using multi-temporal satellite images. An automated kernel-based thresholding algorithm is used to sort the difference values of multi-temporal built-up image, obtained from modified improved NDBI differencing algorithm, into built-up and no-built-up change regions for enhancing the efficiency of built-up change detection process. The improved NDBI differencing algorithm better detects built-up change regions than original NDBI differencing algorithm. As a case study, the proposed algorithm has been implemented on Landsat-5 Thematic Mapper (TM) images of a typical Indian city and surrounding areas for built-up change detection.     

Monitoring vegetation dynamics using the universal normalized vegetation index (UNVI): An optimized vegetation index-VIUPD

This paper propose a universal normalized vegetation index (UNVI), which is an improved vegetation index (VI) based on the universal pattern decomposition method (UPDM), termed VIUPD. We also derive new matrices to facilitate convenient calculation of the UNVI based on data from the MODIS and Landsat-TM, ETM, OLI satellite sensors. We compared the performance of the UNVI to the normalized difference vegetation index (NDVI), enhanced vegetation index (EVI) and modified soil adjusted vegetation index 2 (MSAVI2) to estimate the vegetation dynamics (chlorophyll content and leaf area index [LAI]). The results show that the UNVI was more sensitive to vegetation dynamics than the NDVI, EVI and MSAVI2. The UNVI has a higher LAI saturation point than the other three indices. The UNVI can be used to monitor global changes in above ground biomass (AGB) and gross primary production (GPP) with respect to a wide range of vegetation dynamics.     

In Vivo Comparison of Backscatter Techniques for Ultrasonic Bone Assessment at the Femoral Neck

Ultrasonic techniques are being developed to detect changes in cancellous bone caused by osteoporosis. The goal of this study was to test the relative in vivo performance of eight backscatter parameters developed over the last several years for ultrasonic bone assessment: apparent integrated backscatter (AIB), frequency slope of apparent backscatter (FSAB), frequency intercept of apparent backscatter (FIAB), normalized mean of the backscatter difference (nMBD), normalized slope of the backscatter difference (nSBD), normalized intercept of the backscatter difference (nIBD), normalized backscatter amplitude ratio (nBAR) and backscatter amplitude decay constant (BADC). Backscatter measurements were performed on the left and right femoral necks of 80 adult volunteers (age = 25 ± 11 y) using an imaging system equipped with a convex array transducer. For comparison, additional ultrasonic measurements were performed at the left and right heel using a commercially available heel-bone ultrasonometer that measured the stiffness index. Six of the eight backscatter parameters (all but nSBD and nIBD) exhibited similar and highly significant (p < 0.000001) left–right correlations (0.51 ≤ R ≤ 0.68), indicating sensitivity to naturally occurring variations in bone tissue. Left–right correlations for the stiffness index measured at the heel (R = 0.75) were not significantly better than those produced by AIB, FSAB and FIAB. The short-term precisions of AIB, nMBD, nBAR and BADC (7.8%–11.7%) were comparable to that of the stiffness index measured with the heel-bone ultrasonometer (7.5%).     

Revealing the potential of spectral and textural predictor variables in a neural network-based rainfall retrieval technique

Estimating rainfall areas and rates from geostationary satellite images has the opportunity of both, a high spatial and a high temporal resolution which cannot be achieved by other satellite-based systems until now. Most recent retrieval techniques are solely based on spectral channels of the satellites. These retrievals can be classified as ‘purely pixel-based’ because no information about the neighbourhood pixels is included. Assuming that precipitation is highly correlated with cloud processes and therefore with cloud texture, textural information derived from the neighbourhood of a pixel might give valuable information about the cloud type and hence about a respective probability of the rainfall rate. To study the potential of textural variables to improve optical rainfall retrieval techniques, rainfall areas and rainfall rates were estimated over Germany for the year 2010 using a neural network approach. In addition to the spectral predictor variables from Meteosat Second Generation (MSG), different Grey Level Co-occurance Matrix (GLCM) based textural variables were calculated from all MSG channels. Using recursive feature selection, models were trained and their performance was compared to spectral-only models. Contrary to the expectations, the performance of the models did not increase when textural information was included.     

An automated method for the detection of emperor penguin colonies from Landsat 8 imagery

Detailed information on the emperor penguin colonies is crucial for estimating total populations and analysing population migration. This study presents a new method for detecting colonies of emperor penguins by identifying areas covered with their faeces using Landsat 8 data. Top of Atmosphere (TOA) reflectance and Brightness Temperature (BT) of Landsat images are used as inputs. This method first uses normalized spectral indexes (normalized difference water index (NDWI), normalized difference faeces index (NDFI), normalized difference snow index (NDSI)), band ratios and an individual band reflectance to produce probability masks for areas covered by faeces differentiating from other land cover types. Then, after eliminating those pixels that have abnormal elevations and performing a median filter, the probability masks for those areas covered by faeces are used to derive the corresponding polygons. Subsequently, the geometric centre of the polygons of those areas covered by faeces is used as the location for a corresponding colony. For a widely distributed set of data around the Ross and Somov Seas, the overall classification accuracy is as high as 91% with a small standard deviation of 0.12.     

Tracking MODIS NDVI time series to estimate fuel accumulation

Patterns of post-fire recovery in southern California chaparral shrublands are important for understanding fuel available for future fires. Satellite remote sensing provides an opportunity to examine these patterns over large spatial extents and at high temporal resolution. The relatively limited temporal range of satellite remote sensing products has previously constrained studies of post-fire recovery to chronosequence approaches, in which space is substituted for time, but the lengthening satellite data record is easing this limitation. In this study, we tracked vegetation recovery using a pixel-explicit approach from 2000 to 2013 using normalized difference vegetation index (NDVI) data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor. We attempted to control for inter-annual precipitation variation and examined the influence of shrub cover on MODIS NDVI during the post-fire recovery process. We find strong variation in recovery trajectories associated with site differences, which would have been lost in a chronosequence approach. Shrub cover plays a larger role in explaining annual NDVI variation during the early stages of post-fire recovery, and is a less important factor in more mature stands.     

Using satellite data to estimate particulate air quality in a subtropical city: an evaluation of accuracy and sampling issues

Although satellite data are increasingly being used for particulate air quality studies, the applicability of satellite-derived aerosol optical depth (AOD or τ) products for use over tropical or subtropical cities with frequent cloud cover should be carefully examined. Using eight years of ground-based and satellite-based observations, we assess the accuracy and sampling issues of using satellite data to study particulate air quality over a typical subtropical city, Hong Kong, at monthly to yearly timescales. The validation of Moderate Resolution Imaging Spectroradiometer (MODIS) AOD products shows that 64.6% of the retrievals fall within an expected error envelope of ± (0.05 + 0.15τ) and that they have a low bias during the eight-year study period, thus suggesting that the accuracy of current satellite-derived AOD data still needs to be improved. In addition, the availability of satellite observations is typically less than 30% during the months in spring and summer and less than 35% over seasonal and yearly timescales due to the cloudy and rainy weather. Inadequate sampling issues result in large biases over monthly and seasonal timescales; however, satellite data do not have major sampling issues on the yearly timescale despite the positive bias due to the washout effects of rain.     

CNN-based estimation of pre- and post-earthquake height models from single optical images for identification of collapsed buildings

In this paper, a novel approach is proposed to identify collapsed buildings after an earthquake using pre-event satellite image as well as post-event satellite image and airborne LiDAR data. In this regard, a convolutional neural network-based method is proposed for estimating a DHM from a single satellite image. A structure reconstruction strategy is designed to improve estimated height values and objects geometry by using the local features of shallow layers and employing a progressive context fusion method. The post-event images and their corresponding LiDAR data are used to train the proposed network. Subsequently, the trained network is employed to estimate a digital height model (DHM) from the pre-event satellite image. Finally, by investigating the difference image of pre- and post-event DHMs, collapsed buildings are identified. It is observed that the quality, kappa coefficient and overall accuracy of the obtained results are 84.86%, 91.15% and 98.78%, respectively, demonstrating a promising performance of the proposed approach.     

Assessment of Water Status in Wheat (<Emphasis Type="Italic">Triticum aestivum</Emphasis> L.) Using Ground Based Hyperspectral Reflectance

Field experiments were conducted with four levels of irrigation and nitrogen on wheat for 2 years (2009–2010 and 2010–2011) to quantify and predict the crop water status using hyperspectral remote sensing. Hyperspectral reflectance in 350–2500 nm range was recorded at five growth stages. Based on highest correlation between relative leaf water content (RLWC) and reflectance in five water bands, the booting stage was identified as the most suitable stage for water stress evaluation. Ten hyperspectral water indices were calculated using the first year booting stage reflectance data and prediction models for RLWC and equivalent water thickness (EWT) based on these ten indices were developed. The prediction models for RLWC based on moisture stress index (MSI), normalized difference infrared index (NDII), normalized difference water index₁₆₄₀ (NDWI₁₆₄₀) and normalized multi-band drought index (NMDI) were identified as the most precise and accurate models as indicated by different validation statistics. The models developed for EWT based on water band index (WBI), MSI, NDWI₁₆₄₀ and NMDI were found to be most suitable and accurate. These indices were found to be insensitive to N stress treatments indicating their ability to detect water deficiency as the cause of plant stress. Thus, the study identified four hyperspectral water indices to assess the wheat crop water status at booting stage and developed their respective predictive models.     

Segmentation of age-related white matter changes in a clinical multi-center study

Age-related white matter changes (WMC) are thought to be a marker of vascular pathology, and have been associated with motor and cognitive deficits. In the present study, an optimized artificial neural network was used as an automatic segmentation method to produce probabilistic maps of WMC in a clinical multi-center study. The neural network uses information from T1- and T2-weighted and fluid attenuation inversion recovery (FLAIR) magnetic resonance (MR) scans, neighboring voxels and spatial location. Generalizability of the neural network was optimized by including the Optimal Brain Damage (OBD) pruning method in the training stage. Six optimized neural networks were produced to investigate the impact of different input information on WMC segmentation. The automatic segmentation method was applied to MR scans of 362 non-demented elderly subjects from 11 centers in the European multi-center study Leukoaraiosis And Disability (LADIS). Semi-manually delineated WMC were used for validating the segmentation produced by the neural networks. The neural network segmentation demonstrated high consistency between subjects and centers, making it a promising technique for large studies. For WMC volumes less than 10 ml, an increasing discrepancy between semi-manual and neural network segmentation was observed using the similarity index (SI) measure. The use of all three image modalities significantly improved cross-center generalizability compared to neural networks using the FLAIR image only. Expert knowledge not available to the neural networks was a minor source of discrepancy, while variation in MR scan quality constituted the largest source of error.     

Object-based classification with rotation forest ensemble learning algorithm using very-high-resolution WorldView-2 image

Machine learning algorithms reported to be robust and superior to the conventional parametric classifiers have been recently employed in object-based classification. Within these algorithms, ensemble learning methods that construct set of individual classifiers and combining their predictions to make final decision about unlabelled data have been successfully applied. In this study, performance and effectiveness of a novel ensemble learning algorithm, rotation forest (RotFor) aiming to build diverse and accurate classifiers, was investigated for the first time in object-based classification using a WorldView-2 (WV-2) satellite image. Also, the combination of satellite imagery and ancillary data (i.e. normalized difference vegetation index and principal components) were assessed. Random forest (RF), support vector machine (SVM) and nearest neighbour (NN) algorithms were also used as benchmark classifiers to evaluate the power of RotFor. The classification results confirmed that integration of ancillary data increased the classification accuracy in comparison to using solely spectral bands of WV-2. While RotFor and SVM generally produced similar results, they outperformed the RF and NN based on McNemar’s and Wilcoxon’s signed-rank test of statistical significance results.     

An artificial neural network ensemble to predict disposition and length of stay in children presenting with bronchiolitis.

BACKGROUND: Artificial neural networks apply complex non-linear functions to pattern recognition problems. An ensemble is a 'committee' of neural networks that usually outperforms single neural networks. Bronchiolitis is a common manifestation of viral lower respiratory tract infection in infants and toddlers. OBJECTIVE: To train artificial neural network ensembles to predict the disposition and length of stay in children presenting to the Emergency Department with bronchiolitis. METHODS: A specifically constructed database of 119 episodes of bronchiolitis was used to train, validate, and test a neural network ensemble. We used EasyNN 7.0 on a 200 Mhz pentium PC with a maths co-processor. The ensemble of neural networks constructed was subjected to fivefold validation. Comparison with actual and predicted dispositions was measured using the kappa statistic for disposition and the Kaplan-Meier estimations and log rank test for predictions of length of stay. RESULTS: The neural network ensembles correctly predicted disposition in 81% (range 75-90%) of test cases. When compared with actual disposition the neural network performed similarly to a logistic regression model and significantly better than various 'dumb machine' strategies with which we compared it. The prediction of length of stay was poorer, 65% (range 60-80%), but the difference between observed and predicted lengths of stay were not significantly different. CONCLUSION: Artificial neural network ensembles can predict disposition for infants and toddlers with bronchiolitis; however, the prediction of length of hospital stay is not as good.     

Integrating Mahalanobis typicalities with a neural network for rubber distribution mapping

Accurate rubber distribution mapping is critical to the study of its expansion and to provide a better understanding of the consequences of land-cover and land-use change on carbon and water cycles. Employing Mahalanobis typicalities as inputs to a hard classifier to enhance the capability of generalization has not previously been explored. This letter presents a novel approach by integrating Mahalanobis typicalities with the multi-layer perceptron (MLP) neural network for mapping of rubber. A case study from the Thai–Lao and Sino–Lao borders was conducted using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data. Different combinations of the nine ASTER bands including Visible and Near Infrared (VNIR) and Short-wave Infrared (SWIR), Normalized Difference Vegetation Index (NDVI) and Mahalanobis typicalities were used as input variables to the MLP. Results indicate that including Mahalanobis typicalities as input variables can improve the MLP's performance and increase the user's accuracy of rubber mapping.     

Detection of flow limitation in obstructive sleep apnea with an artificial neural network

During sleep, the development of a plateau on the inspiratory airflow/time contour provides a non-invasive indicator of airway collapsibility. Humans recognize this abnormal contour easily, and this study replicates this with an artificial neural network (ANN) using a normalized shape. Five 10 min segments were selected from each of 18 sleep records (respiratory airflow measured with a nasal cannula) with varying degrees of sleep disordered breathing. Each breath was visually scored for shape, and breaths split randomly into a training and test set. Equally spaced, peak amplitude normalized flow values (representing breath shape) formed the only input to a back propagation ANN. Following training, breath-by-breath agreement of the ANN with the manual classification was tabulated for the training and test sets separately. Agreement of the ANN was 89% in the training set and 70.6% in the test set. When the categories of 'probably normal' and 'normal', and 'probably flow limited' and 'flow limited' were combined, the agreement increased to 92.7% and 89.4% respectively, similar to the intra- and inter-rater agreements obtained by a visual classification of these breaths. On a naive dataset, the agreement of the ANN to visual classification was 57.7% overall and 82.4% when the categories were collapsed. A neural network based only on the shape of inspiratory airflow succeeded in classifying breaths as to the presence/absence of flow limitation. This approach could be used to provide a standardized, reproducible and automated means of detecting elevated upper airway resistance.     

Global and regional drought dynamics in the climate warming era

This article investigates whether the highest global temperature during 2001–2012 triggered some changes in drought area, frequency, intensity and duration. New satellite-based vegetation health (VH) technology and regional in situ data were used for this analysis. The VH indices were used to investigate trends in global and regional drought area for several drought intensities (starting from moderate-to-exceptional (ME)) during the warmest decade, after 2000. Two of the most recent strongest droughts, 2010 in Russia and 2011 in the USA, are also discussed. During 2001–2012, droughts of ME, severe-to-exceptional (SE) and extreme-to-exceptional (EE) severity covered 17–35%, 7–15% and 2–6% of the total area of the world, respectively. No trends in drought areas for these levels of severity were found. Regional analysis was performed on Ukraine (from both satellite and in situ data). Annual mean temperature of the entire country follows global warming tendency, although the intensity is twice stronger, 1.45°C over 50-year period. The droughts of SE and EE severity during the growing season normally affect 25–60% (up to 80% of the major crop area) and 5–10% (up to 20%) of the entire country, respectively, and the later leading up to 40% of losses in Ukrainian grain production.     

Characterization of variability at in situ locations for calibration/validation of satellite-derived land surface temperature data

Calibration and validation (cal/val) of data derived from satellite-based instruments is critical to providing accurate global measurements of environmental variables at useful spatial and temporal resolutions. In this letter, statistical models based on linear regressions employing various predictor variables were utilized to elucidate appropriate methods of characterizing variability near ground sites that might be used for calibration and validation. Regressions based on more complex statistics performed no better than those based on easily derived statistics, and the regression relations provided valuable information for assessing the potential quality of satellite-based measures of land surface temperature.