Ship detection on the SAR images for marine monitoring has a wide usage. SAR technology helps us to have a better monitoring over intended sections, without considering atmospheric conditions, or image shooting time. In recent years, with advancements in convolutional neural network (CNN), which is one of the well-known ways of deep learning, using image deep features has increased. Recently, usage of CNN for SAR image segmentation has been increased. Existence of clutter edge, multiple interfering targets, speckle and sea-level clutters makes false alarms and false detections on detector algorithms. In this letter, constant false alarm rate is used for object recognition. This algorithm, processes the image pixel by pixel, and based on statistical information of its neighbor pixels, detects the targeted pixels. Afterward, a neural network with hybrid algorithm of CNN and multilayer perceptron (CNN–MLP) is suggested for image classification. In this proposal, the algorithm is trained with real SAR images from Sentinel-1 and RADARSAT-2 satellites, and has a better performance on object classification than state of the art.

Ship Classification in SAR Images Using a New Hybrid CNN–MLP Classifier

Recharge assessment is of critical importance for groundwater resources evaluation in arid/semiarid areas, as these have typically limited surface water resources. There are several models for water balance evaluation. One of them is WetSpass, which has the ability to simulate spatially distributed recharge, surface runoff, and evapotranspiration for seasonally averaged conditions. This paper presents a modified methodology and model, WetSpass-M, in which the seasonal resolution is downscaled to a monthly scale. A generalized runoff coefficient was introduced, enabling runoff estimation for different land-use classes. WetSpass-M has been calibrated and validated with observed streamflow records from Black Volta. Base-flow from simulated recharge was compared with base-flow derived via a digital filter applied to the observed streamflow and has shown to be in agreement. Previous studies have concluded that for this basin, small changes in rainfall could cause a large change in surface runoff, and here a similar behavior is observed for recharge rates. An advantage of the new model is that it is applicable to medium- and large-sized catchments. It is useful as an assessment tool for evaluating the response of hydrological processes to the changes in associated hydrological variables. Since monthly data for streamflow and climatic variables are widely available, this new model has the potential to be used in regions where data availability at high temporal resolution is an issue. The spatial–temporal characteristics of the model allow distributed quantification of water balance components by taking advantage of remote sensing data.

A distributed monthly water balance model: formulation and application on Black Volta Basin

Mapping per-pixel predicted accuracy of classified remote sensing images.

We investigated with remote sensing (APEX images) the coexistence of phytoplankton and macrophytes in three interconnected shallow and hypereutrophic fluvial lakes (Mantua Lakes, Northern Italy). High concentrations of chlorophyll-a, up to 60 mg m−3, were determined in the open water between well-developed stands of floating-leaved, submerged, and emergent macrophytes. Our data suggest a general inhibition of phytoplankton by macrophytes, evidenced by decreasing chlorophyll-a concentrations in proximity of macrophyte stands. Chlorophyll-a concentrations halved in the proximity of emergent stands (~6 mg m−3 within 21 m from the stand border) when compared to the outer zones (~13 mg m−3). Contrasting trends were observed for submerged stands, where concentrations decreased inwards from ~8 to ~3 mg m−3. Floating leaved stands had a neutral effect, chlorophyll-a being nearly constant in both inner and outer zones. Overall, remotely-sensed data allow evaluation of quantitative and spatially defined interactions of macrophytes and phytoplankton at the whole ecosystem scale.

Remote sensing of phytoplankton-macrophyte coexistence in shallow hypereutrophic fluvial lakes

Convolutional Neural Network (CNN) has attracted much attention for feature learning and image classification, mostly related to close range photography. As a benchmark work, we trained a relatively large CNN to classify SAR image patches into five different categories, where the image patches tiled and annotated from a typical TerraSAR-X spotlight scene of Wuhan, China. The neural network designed in this paper consists of seven layers, including one input layer, two convolutional layers where each followed by a max-pooling layer, as well as two fully-connected layers with a final five-class softmax. Using the toolkit caffe, we achieved the training and testing accuracy of 85.7% and 85.6% respectively, which is considerably better than the traditional feature extraction and classification based SVM method and shows great potential of CNN used for SAR image interpretation. In order to accelerate the training process, a very efficient GPU implementation was employed.

Convolutional Neural Network for SAR image classification at patch level

Objective and detailed mapping of urban land-cover types over large areas is important for hydrological modelling, as most man-made land-cover consist of sealed surfaces which strongly reduce groundwater recharge. Moreover, impervious surfaces are the predominant type in urbanized areas and can lead to increased surface runoff. Classification of man-made objects in urbanized areas is not straightforward due to similarity in spectral properties. This study examines the use of hyperspectral CHRIS-Proba images for complex urban land-cover classification of the Woluwe River catchment, Brussels, Belgium. Two methods are compared: 1) a multiscale region-based classification approach, which is based on a causal Markovian model being defined on a Multiscale Region Adjacency Tree and a set of nonparametric dissimilarity measures; and 2) a pixel based classification method with a Mahalanobis distance classifier. Multiscale region-based classification results in a Kappa value of 0.95 while pixel-based classification has a slightly lower Kappa value of 0.92. The impact of the classification method on the hydrology is estimated with the application of the WetSpass physically-based distributed water balance model. The model uncertainty is assessed with the use of a Monte Carlo simulation. Model results show that the region-based classification yields to a higher yearly recharge than the pixel-based classification. The overall uncertainty, quantified by the Monte Carlo method is lower for the region-based classification than for the pixel-based classification. The presented study indicates that the selection of the classification technique is of critical importance for the outcome of hydrological models.

I. Vanhamel

Papers

Impact of Urban Land-Cover Classification on Groundwater Recharge Uncertainty