Satellite-based onboard data processing is crucial for time-sensitive applications requiring timely and efficient rapid response. Advances in edge artificial intelligence are shifting computational power from ground-based centres to on-orbit platforms, transforming the “sensing-communication-decision-feedback” cycle and reducing latency from acquisition to delivery. The current research presents a framework addressing the strict bandwidth, energy, and latency constraints of small satellites, focusing on maritime monitoring. The study contributes three main innovations. Firstly, it investigates the application of deep learning techniques for direct ship detection and classification from raw satellite imagery. By simplifying the onboard processing chain, our approach facilitates direct analyses without requiring computationally intensive steps such as calibration and ortho-rectification. Secondly, to address the scarcity of raw satellite data, we introduce two novel datasets, VDS2Raw and VDV2Raw, which are derived from raw data from Sentinel-2 and Vegetation and Environment Monitoring New Micro Satellite (VEN $\mu$ S) missions, respectively, and enriched with Automatic Identification System (AIS) records. Thirdly, we characterize the tasks' optimal single and multiple spectral band combinations through statistical and feature-based analyses validated on both datasets. In sum, we demonstrate the feasibility of the proposed method through a proof-of-concept on CubeSat-like hardware, confirming the models' potential for operational satellite-based maritime monitoring.
Assessing the development of wildfire scars during a period of consecutive active fires and smoke overcast is a challenge. The study was conducted during nine months when Israel experienced massive pyro-terrorism attacks of more than 1100 fires from the Gaza Strip. The current project strives at developing and using an advanced Earth observation approach for accurate post-fire spatial and temporal assessment shortly after the event ends while eliminating the influence of biomass burning smoke on the ground signal. For fulfilling this goal, the Aerosol-Free Vegetation Index (AFRI), which has a meaningful advantage in penetrating an opaque atmosphere influenced by biomass burning smoke, was used. On top of it, under clear sky conditions, the AFRI closely resembles the widely used Normalized Difference Vegetation Index (NDVI), and it retains the same level of index values under smoke. The relative differenced AFRI (RdAFRI) set of algorithms was implemented at the same procedure commonly used with the Relative differenced Normalized Burn Ratio (RdBRN). The algorithm was applied to 24 Sentinel-2 Level-2A images throughout the study period. While validating with ground observations, the RdAFRI-based algorithms produced an overall accuracy of 90%. Furthermore, the RdAFRI maps were smoother than the equivalent RdNBR, with noise levels two orders of magnitude lower than the latter. Consequently, applying the RdAFRI, it is possible to distinguish among four severity categories. However, due to different cloud cover on the two consecutive dates, an automatic determination of a threshold level was not possible. Therefore, two threshold levels were considered through visual inspection and manually assigned to each imaging date. The novel procedure enables calculating the spatio-temporal dynamics of the fire scars along with the statistics of the burned vegetation species within the study area.
The bidirectional reflectance distribution function (BRDF) is crucial in determining the quantity of reflected light on the earth’s surface as a function of solar and view angles (i.e., azimuth and zenith angles). The Vegetation and ENvironment monitoring Micro-Satellite (VENµS) provides a unique opportunity to acquire data from the same site, with the same sensor, with almost constant solar and view zenith angles from two (or more) view azimuth angles. The present study was aimed at exploring the view angles’ effect on the stability of the values of albedo and of two vegetation indices (VIs): the normalized difference vegetation index (NDVI) and the red-edge inflection point (REIP). These products were calculated over three polygons representing urban and cultivated areas in April, June, and September 2018, under a minimal time difference of less than two minutes. Arithmetic differences of VIs and a change vector analysis (CVA) were performed. The results show that in urban areas, there was no difference between the VIs, whereas in the well-developed field crop canopy, the REIP was less affected by the view azimuth angle than the NDVI. Results suggest that REIP is a more appropriate index than NDVI for field crop studies and monitoring. This conclusion can be applied in a constellation of satellites that monitor ground features simultaneously but from different view azimuth angles.
18 May 2018 in IEEE Transactions on Automatic Control
This Note presents new results on the design of the process noise covariance matrix in Kalman filtering. The proposed design belongs to the realm of worst-case methods: the optimal covariance is sought such as to maximize an upper bound on the estimation error matrix under a total budget constraint. Optimality conditions are derived along with a numerical algorithm for iterative approximation of the optimal solution. The optimal solution features a few time points and the associated covariances. Broadly speaking the process noise is only "active" at the time points that enhance its controllability. The importance of the method is that it bounds from above the performances of any Kalman filter with a different design. Monte-Carlo simulations on a double harmonic system numerical example illustrate the good convergence, performances, and low sensitivity to initial guesses of the proposed method. Moreover, in terms of estimation accuracy, the Kalman filter with the optimal design outperforms a robust H_2 filter designed to cope with an equivalent uncertainty.
1 December 2019 in Journal of Guidance, Control, and Dynamics