Water Demand Forecasting in Umm Al-Quwain (UAE) Using the IWR-MAIN Specify Forecasting Model

IWR-MAIN software is used in this paper to forecast water demand in the Emirate of Umm Al-Quwain (UAQ), located in the northern part of the United Arab Emirates (UAE), for the next twenty 5 years. Two different databases are used. The first one provides average yearly water consumptions since 1980, while the second provides more detailed monthly water consumptions from 2000. The correlation between three different independent variables and water consumption is studied. These variables are population of UAQ, average temperature, and average rainfall. Results show that population is the most significant variable that affects water consumption in Umm Al-Quwain. Several calibration simulations are performed and each simulation is divided into two periods. In the first period the software “Statistical Package for the Social Sciences” (SPSS) is used to determine the correlation coefficients between the independent variables and actual water consumptions. These coefficients are used in IWR-MAIN over the second period to calculate values of water demand which are compared against actual water consumptions. Model calibration indicates that starting the calibration in 1999 in database one and 2006 in database 2 minimizes differences between actual and simulated water demands. Therefore, these simulations were used as the bases for several forecasting scenarios of water demand in Umm Al-Quwain. Results of one of these scenarios show that 50% increase in water demand is expected by the year 2015 and double of the current demand will be needed before 2025. In another forecasting scenario, it was found that by considering the expected increase in the income level, the water demand will increase by 40% in one decade. A new technique of using IWR-MAIN to separate estimates of metered demand, unmetered demand, and unaccounted water (losses) is also presented in this paper. Finally, results of a fourth scenario indicate that water demand in Umm Al-Quwain will be highly affected by the expected high migration rate due to the anticipated new developments in the emirate.     

A supervised hierarchical segmentation of remote-sensing images using a committee of multi-scale convolutional neural networks

This paper presents a supervised, hierarchical remote-sensing image segmentation technique using a committee of multi-scale convolutional neural networks. With existing techniques, segmentation is achieved through fine-tuning a set of predefined feature detectors. However, such a solution is not robust since the introduction of new sensors or applications would require novel features and techniques to be developed. Conversely, the proposed method achieves segmentation through a set of learnt feature detectors. In order to learn feature detectors, the proposed method exploits a committee of convolutional neural networks that perform multi-scale analysis on each band in order to derive individual confidence maps on region boundaries. Confidence maps are then inter-fused in order to produce a fused confidence map. Furthermore, the fused map is intra-fused using a morphological scheme into a hierarchical segmentation map. The proposed method is quantitatively compared to baseline techniques on a publicly available data set. The results presented in this paper highlight the improved accuracy of the proposed method.     

Temporal error concealment using motion field interpolation

An error concealment algorithm based on motion field interpolation is presented. For each pixel in a damaged block, the algorithm recovers a motion vector using bilinear interpolation of neighbouring motion vectors. This vector is then used to conceal the damaged pixel. Overlapped motion compensation is used to combine this algorithm with a boundary matching error concealment algorithm. Simulation results demonstrate the superior performance of the proposed algorithms     

Motion field interpolation for temporal error concealment

When transmitted over practical communication channels, compressed video can suffer severe degradation. One approach to combat the effect of channel errors is error concealment. It is an attractive choice because it does not increase the bit rate, it does not require any modifications to the encoder, it does not introduce any delays and it can be applied in almost any application. Conventional temporal concealment techniques estimate one concealment displacement for the whole damaged block and then use translational displacement compensation to conceal the block from a reference frame. The main problem with such techniques is that incorrect estimation of the concealment displacement can lead to poor concealment of the whole or most of the block. Two novel temporal concealment techniques are presented. In the first technique, motion field interpolation is used to estimate one concealment displacement per pel of the damaged block and then each pel is concealed individually. In this case, incorrect estimation of a concealment displacement will only affect the corresponding pel. On a block level this may affect few pels rather than the entire block. In the second technique, multi-hypothesis motion compensation is used to combine the first technique with a boundary matching temporal concealment technique to obtain a more robust performance. Simulation results, within both an isolated error propagation environment and an H.263 codec, show the superior subjective and objective performance of the proposed techniques when compared with conventional temporal concealment techniques     

Simplex minimisation for fast long-term memory motion estimation

Three algorithms for fast long-term memory block matching motion estimation are proposed. The algorithms are extensions of the fast single-reference simplex minimisation search (SMS) algorithm. The proposed algorithms represent different degrees of compromise between prediction quality and computational complexity. With a long-term memory of 50 frames the proposed algorithms have a computational complexity comparable to that of single-reference full-search while still maintaining the prediction gain of long-term memory motion estimation