Dynamic Landscape Fragmentation and the Driving Forces on Haitan Island, China

Island ecosystems have distinct and unique vulnerabilities that place them at risk from threats to their ecology and socioeconomics. Spatially exhibiting the fragmentation process of island landscapes and identifying their driving factors are the fundamental prerequisites for the maintenance of island ecosystems and the rational utilization of islands. Haitan Island was chosen as a case study for understanding landscape fragmentation on urbanizing Islands. Based on remote sensing technology, three Landsat images from 2000 to 2020, landscape pattern index, transect gradient analysis, and moving window method were used in this study. The results showed that from 2000 to 2020, impervious land increased by 462.57%. In 2000, the predominant landscape was cropland (46.34%), which shifted to impervious land (35.20%) and forest (32.90%) in 2020. Combining the moving window method and Semivariogram, 1050 m was considered to be the best scale to reflect the landscape fragmentation of Haitan Island. Under this scale, it was found that the landscape fragmentation of Haitan Island generally increased with time and had obvious spatial heterogeneity. We set up sampling bands along the coastline and found that the degree of landscape fragmentation, advancing from the coast inland, was decreasing. Transects analysis showed the fragmentation intensity of the coastal zone: the north-western and southern wooded zones decreased, while the concentration of urban farmland in the north-central and southern areas increased. The implementation of a comprehensive experimental area plan on Haitan Island has disturbed the landscape considerably. In 2000, landscape fragmentation was mainly influenced by topography and agricultural production. The critical infrastructure construction, reclamation and development of landscape resources have greatly contributed to the urbanisation and tourism of Haitan Island, and landscape fragmentation in 2013 was at its highest. Due to China’s “Grain for Green Project” and the Comprehensive Territorial Spatial Planning policy (especially the protection of ecological control lines), the fragmentation of Haitan Island was slowing. This study investigated the optimal spatial scale for analyzing spatiotemporal changes in landscape fragmentation on Haitan Island from 2000 to 2020, and the essential influencing factors in urban islands from the perspective of natural environment and social development, which could provide a basis for land use management and ecological planning on the island.

Extreme events driving year-to-year differences in gross primary productivity across the US

Solar-induced chlorophyll fluorescence (SIF) has previously been shown to strongly correlate with gross primary productivity (GPP); however this relationship has not yet been quantified for the recently launched TROPOspheric Monitoring Instrument (TROPOMI). Here we use a Gaussian mixture model to develop a parsimonious relationship between SIF from TROPOMI and GPP from flux towers across the conterminous United States (CONUS). The mixture model indicates the SIF–GPP relationship can be characterized by a linear model with two terms. We then estimate GPP across CONUS at 500 m spatial resolution over a 16 d moving window. We observe four extreme precipitation events that induce regional GPP anomalies: drought in western Texas, flooding in the midwestern US, drought in South Dakota, and drought in California. Taken together, these events account for 28 % of the year-to-year GPP differences across CONUS. Despite these large regional anomalies, we find that CONUS GPP varies by less than 4 % between 2018 and 2019.

An improved method for recognizing pediatric epileptic seizures based on advanced learning and moving window technique

In this work, advanced learning and moving window-based methods have been used for epileptic seizure detection. Epilepsy is a disorder of the central nervous system and roughly affects 50 million people worldwide. The most common non-invasive tool for studying the brain activity of an epileptic patient is the electroencephalogram. Accurate detection of seizure onset is still an elusive work. Electroencephalogram signals belonging to pediatric patients from Children’s Hospital Boston, Massachusetts Institute of Technology have been used in this work to validate the proposed method. For determining between seizure and non-seizure signals, feature extraction techniques based on time-domain, frequency domain, time-frequency domain have been used. Four different methods (decision tree, random forest, artificial neural network, and ensemble learning) have been studied and their performances have been compared using different statistical measures. The test sample technique has been used for the validation of all seizure detection methods. The results show better performance by random forest among all the four classifiers with an accuracy, sensitivity, and specificity of 91.9%, 94.1%, and 89.7% respectively. The proposed method is suggested as an improved method because it is not channel specific, not patient specific and has a promising accuracy in detecting epileptic seizure.

SINS attitude algorithm based on moving-window overdetermined polynomial fitting of gyro outputs

The attitude algorithm is the most important part of the whole strapdown inertial navigation (SINS) processing. It calculates the attitude of certain parameterization by integrating the gyro outputs or measurements in a specifically tailored way according to the attitude kinematic differential equation. The measurements or some angular velocity models obtained by fitting these measurements are often assumed free of errors in order to assess the numerical errors only. However, the gyro outputs and hence the models from them are by no means free of measurement errors. It is more often than not that the measurement errors dominate the numerical ones in practice. In this study, with coping with the measurement errors as the focus, we aim to improve the angular velocity model which is used as input in an attitude integration algorithm. This is achieved by exploiting the potential of overdetermined least-squares polynomial fitting. In order to avoid reducing the update rate by incorporating more measurements, the moving window trick is employed to re-use measurements in the previous update interval. The conventional attitude algorithm with second-order approximation in solving the differential equation of the equivalent rotation vector is employed as an example; however, the proposed method can be readily applied to other parameterizations such as direction cosine matrix, quaternion or Rodrigues parameters, and other high order approximations in solving the differential equation widely studied recently.

Coupling Normalization with Moving Window in Backpropagation Neural Network (BNN) for Passive Microwave Soil Moisture Retrieval

A common practice to capture the non-stationary characteristics of the time series data in Artificial Neural Network (ANN) is by randomly dividing the whole set of available data into training, validation and testing, i.e. the data in validation and testing are represented in the training data. Consequently, the usability of the developed model on data not represented by the training data used during the network model development process is always doubtful. In this work, we present a back-propagation neural network (BNN) model trained using one-day history data to predict soil moisture data at 1 km resolution for two future dates. Specifically, high soil moisture values were observed in the training data while the testing data were characterized by drier conditions due to minimal or no rainfall. Our model uses separate mean and standard deviation statistics values from the training and testing data, respectively, to the z-normalized data. With data pre-processed using this method, the BNN model next uses a moving window of size 4 km × 4 km to capture the spatial variability of the soil moisture throughout the 40 km × 40 km study area. The coupling of the normalization and moving window method managed to achieve average soil moisture with Root Mean Square (RMSE) of 3.67% and correlation coefficient, R2 of 0.89. By only using the suggested normalization without the moving window method, the BNN model managed to achieve an average RMSE of barely 5.82% with R2 = 0.83. When comparing with the normal practice of using the same mean and standard deviation statistics of the training data in the testing data, the retrieval accuracy of the BNN model deteriorates to 8.86% with R2 = 0.32. The experiment results demonstrated that the proposed coupling method performed better in terms of both RMSE and R2 for soil moisture retrieval.