IntroductionDigging drainage ditches is common forestry practice across northern European boreal forests and in some parts of North America (Lõhmus et al. 2015). Ditching helps with lowering the groundwater level in the wet parts of the forest to improve soil aeration and support tree growth (Laurén et al. 2021; Sikström and Hökkä 2016). Some of the most drained countries include Finland, the Baltic States, and parts of Sweden, where the drained forest stands comprise 20%–25% of the total forest area (Nieminen et al. 2018). However, the extensive use of ditches over a long period of time has resulted in a major change in wetland and soil hydrology and impacted the overall ecosystem functions of these regions (Kuglerová et al. 2020).Recently, the intensive ditching practice has been identified as posing multiple environmental risks, particularly for degradation of wetland and soil, greenhouse gas emissions (Audet et al. 2017; Peacock et al. 2021), increased nutrient and sediment loadings to water bodies, and biodiversity loss (Holden et al. 2004; Lepistö et al. 2021; Lidman et al. 2017; Lõhmus et al. 2015; Nieminen et al. 2018). Research and actions were set up to minimize environmental loss and restore degraded land in the ditched forest landscape. However, such initiatives are significantly constrained by the lack of accurate and site-specific information of ditch networks. For example, a comparison between a national field inventory (Ståhl et al. 2011) and the best available high-resolution map of Sweden suggests that only 9% of the ditches that were inventoried by the national field inventory are mapped on current maps. In addition, 68% of the field-mapped channels were ditches, highlighting the scale of this lack of knowledge. A study in northern Sweden documented that accounting for all ditches nearly doubled the size of the stream network within a 68-km2 catchment (Hasselquist et al. 2018). These findings are a clear indication that there is a discrepancy between the potential significance of artificial water bodies, such as drainage ditches, and their low representation in scientific research and policy (Koschorreck et al. 2020). The increasing understanding of the environmental risks associated with forest ditches, together with the poor representation of ditch networks in existing maps of many forest landscapes, makes detailed mapping of these ditches a priority for sustainable land and hydrological management.The availability of high-quality remote sensing data has enabled successful mapping of different landscape features; accordingly, satellite-based hydrological analysis has become an effective tool for the assessment of water resources across the globe (McCabe et al. 2017). In open landscapes, such as agricultural lands or open peatlands, ditches can be detected from satellite images or aerial photos (Ayana et al. 2017). But mapping ditches in a forest landscape remains a significant challenge, given that such features are obscured by the tree canopy cover (Benstead and Leigh 2012; Hasselquist et al. 2018). A common solution for mapping small-scale watercourses in such landscapes (Benstead and Leigh 2012) is to model the accumulated flow from a digital terrain model and set a stream initiation threshold for extracting the stream networks. The stream initiation threshold refers to how large a drainage area is required to initiate a stream. However, the formation of stream heads varies depending on climate, topography, and soil conditions (Elmore et al. 2013; Jensen et al. 2017; Julian et al. 2012; Russell et al. 2015). Moreover, in landscapes where channels have been altered by humans, such as in the case of ditches, channel formation is no longer controlled by natural erosion. Hence, the flow accumulation–based hydrological modeling approach does not adequately capture a heavily modified ditch network in forested landscapes.More recently, deep learning approaches have emerged as a mainstay of data processing and analysis in the field of water resources (Sit et al. 2020). Airborne laser scanning (ALS) data of high point density are being integrated with powerful machine learning and statistical tools to map ditches. Most ditch detection studies used ALS-derived digital elevation models (DEMs) but focused only on smaller study areas of up to 150 ha (Cazorzi et al. 2013; Rapinel et al. 2015; Roelens et al. 2018a, b). In addition, those ditch detection studies were predominantly performed in open areas, such as French marshes (Rapinel et al. 2015) and vineyards (Bailly et al. 2008), Belgian grasslands and peri-urban areas (Roelens et al. 2018b), Italian and US agrarian landscapes (Cazorzi et al. 2013; Passalacqua et al. 2012), Chinese rice fields (Qian et al. 2018), or near roads in Finland (Kiss et al. 2015). In such open landscapes, ditch detection is comparatively less complex, as tree cover is rare, and ditches are usually well maintained in these populated, easily accessed, managed landscapes.These ditch detection methods using ALS data in open landscapes generally use data filtering to highlight ditches: for example, wavelet transformation (Bailly et al. 2008), relative elevation attribute (Cazorzi et al. 2013; Roelens et al. 2018b), geometric and Laplacian curvature (Passalacqua et al. 2012), linear filter (Rapinel et al. 2015), or topographic position index and standardized elevation index (Kiss et al. 2015). Data filtering follows a binary classification of raster layers into ditches and nonditches, while the classification can be performed on ALS point clouds, pixels, or linear objects depending on the chosen method. Different classification approaches have been used for binary classification, such as manual setting of thresholds using expert knowledge (Kiss et al. 2015; Passalacqua et al. 2012), logistic regression (Roelens et al. 2018b), and application of machine learning models (Bailly et al. 2011; Roelens et al. 2018a).Deep learning presents a new avenue for data-driven hydrological analysis and modeling, especially with high-resolution imagery. Semantic image segmentation using deep learning is on the rise for many applications, from autonomous driving to virtual or augmented reality systems (Garcia-Garcia et al. 2017). However, the use of deep learning for image analysis in hydrological research remains limited. Deep learning techniques use multilayer models that can effectively capture the underlying complexity in heterogeneous hydrological systems (Sit et al. 2020). Deep learning can be effective for extracting small-scale hydrological features and can reduce prediction uncertainty by being insusceptible to raw and noisy data (Shen et al. 2018). Here, we developed a novel methodology by combining two state-of-the-art technologies—ALS and deep learning—for detecting drainage ditches in forested landscapes. We then validated the methodology using data from multiple countries in Northern Europe. To the best of our knowledge, this approach to detecting ditches has not been reported in the literature previously.MethodWe trained a deep neural network on high-density ALS data and manually digitized ditches from 10 regions across Sweden. We set aside 20% of the data for testing the final model. In addition to this testing data, the model was applied to four additional test areas spread across Sweden, Finland, Latvia, and Poland [Fig. 1(a)].Training DatasetThe 10 digitized forest-dominated regions were selected to achieve a broad representation of different landscape properties concerning topography, soil conditions, runoff, land use, and tree species. A compact laser-based system (Leica ALS80-HP-8236) was used to collect the ALS data from an aircraft flying at 2,888–3,000 m. The ALS point clouds had a point density of one to two points per m2 and were divided into 55 tiles with a size of 2.5×2.5  km each. Combined, the tiles cover an area of 344  km2. DEMs with 0.5-m resolution were created from the ALS point clouds using a tin-gridding approach implemented in Whitebox tools 1.4.0 (Lindsay 2018). A high-pass median filter (HPMF) was applied to the DEMs to emphasize short-range variability in the topography. The HPMF algorithm, implemented in Whitebox tools (Lindsay 2018), operates by subtracting the value at the grid cell at the center of the window from the median value in the surrounding neighborhood with a kernel of 11 cells. Negative values indicate depressions, and positive values indicate ridges.LabelsDitches were manually digitized as vector lines by trained experts who were calibrated among themselves with regular meetings and discussions of “edge cases”. Multidirectional hill-shaded DEMs and a HPMF were used to visually separate local ridges from local depressions (e.g., ditch channels). Current and historical orthophotos and maps were used to corroborate edge cases to digitize the ditch network. Of the 55 tiles, 20% (n=11) were randomly selected for model testing and were not used to train the model [Fig. 1(b)]. The digitized vector lines have no width, so we used average ditch width from a field inventory during which 2,188 ditch channels were visited across Sweden (Ståhl et al. 2011). The average ditch width was 2 m, with a standard deviation of 1.3 m. Instead of flagging all pixels within ∼3  m of a vector line as ditch, we used the HPMF to create more natural ditch labels. Pixels within 3 m of a vector line and with a HPMF value less than −0.075 were flagged as ditch pixels. The threshold value of −0.075 was selected based on visual observations of ditch channels during digitalization of the ditches. Spurious pixels that were not connected to a ditch were removed using a majority filter with a three-cell kernel. In total 1,607 km of ditch channels were mapped in this manner (Ågren et al. 2022).Deep Learning ModelThe HPMF and labeled data from each of the 55 tiles were split into pairs of image chips with 512×512  pixels in each chip [Fig. 1(c)]. Image chips in which ditches made up less than 0.1% of all pixels in the chip were removed to combat the highly imbalanced class distribution: clearly, most pixels in the data are nonditch. This resulted in 2,367 pairs of image chips. We used TensorFlow 2.6 to build an encoder-decoder style deep neural network, shown in Fig. 2, to transform the filtered HPMF images into images highlighting the detected ditches. On the encoding path, the network learns a series of filters, organized in layers, that express larger and larger neighborhoods of pixels in fewer and fewer vectors of features. This downsampling forces the network to ignore noise and extract features relevant for ditch detection. In contrast to normal convolutional neural networks, which apply a filter to all feature vectors in a certain spatial neighborhood at once, we use Xception blocks (Chollet 2017). These blocks decouple the filtering of the spatial neighborhood within each feature dimension from the filtering across feature dimensions. This simplifies the learning problem for ditch detection, since there is no strong coupling between the two dimensions.After encoding the HPMF image into a spatially more compact representation, it was again decoded by a series of learned filters performing transposed convolutions into the final classification map. This map contains, for every pixel in the input image, the probability that the pixel belongs to a ditch. Although the neural network considers the neighborhood of the pixel to label that pixel, the procedure still leads to label discontinuities, either in the form of ditches disrupted by mislabeled nonditch pixels or in the form of areas of nonditch pixels in which single pixels are labeled as ditch pixels. To smooth out this type of noise, we use a conditional random field layer proposed by Zheng et al. (2015), which learns to penalize undue label discontinuities. This neural network model is trained using weighted cross-entropy loss to deal with the large class imbalance between ditch and nonditch pixels.EvaluationWe evaluated the trained model on the set-aside test tiles (n=11) as well as on four additional test areas in Sweden (a 68-km2 area scanned with 20 points per m2), Finland (a 70-km2 area scanned with five points per m2), Latvia (a 25-km2 area scanned with four points per m2), and Poland (a 44-km2 area scanned with four points per m2) [Fig. 1(a)]. Note that the test sites outside of Sweden were not digitized by the same team of experts that digitized the Swedish training and testing data.The datasets are highly imbalanced. Therefore, we used several metrics, such as Cohen’s κ and the Matthews correlation coefficient (MCC), to assess the accuracy of the model. Additional metrics were extracted to compare the performance of our model with 13 other ditch-mapping studies published between 2008 and 2020. Most of the previous studies reported inadequate accuracy measures for highly imbalanced binary prediction. Therefore, we included a large number of accuracy measures to enhance the confidence in our prediction.DiscussionMost previous ditch detection studies focused only on study areas smaller than 100  km2 within predominantly open areas. Here, we introduce a large dataset with 1,607 km of mapped ditch channels spread across 344  km2 in a landscape mainly dominated by forest. We demonstrate that our approach of combining deep learning with high-density ALS data to map drainage ditches has accuracy equal to or higher than that of previous studies. However, the studies by Graves et al. (2020), Kiss et al. (2015), Larson and Trivedi (2011), Rapinel et al. (2015), Stanislawski et al. (2018) reported only recall, which made them difficult to evaluate. Additional metrics on false positives would be helpful for a comparison with our results. It needs to be emphasized that we used only one topographical index to achieve this high accuracy, while other studies, such as Roelens et al. (2018a), relied on multiple indices for mapping drainage ditches in a smaller landscape (i.e., 68  km2). This substantially enhances the utility of our methodology, especially for implementing it at large spatial scales with limited computational resources. The model could map an area of 1  km2 in 8.6 s using a GeForce GTX 1080 Ti Graphics Card. Still, it will be worth exploring whether adding more indices besides HPMF can improve the prediction accuracy. For example, a recent study in Finland by Bhattacharjee et al. (2021) demonstrated that aerial photos may contain important information for mapping drainage ditches on peatlands. However, in areas hidden beneath canopy, it might be better to include topographical indices such as impoundment size index implemented in Whitebox tools 1.4.0 (Lindsay 2018), which can be included as multiband images. Thus, it seems worthwhile to explore these indices as additional covariates in the deep neural network model for future analysis.Considering the inherent challenges of the study region’s glaciated and forested landscape, the performance of our ditch mapping approach was highly successful. For instance, the landscape presented substantial challenges due to landscape features that appear similar to ditch channels in the ALS data. These include natural streams, historical channels from the latest deglaciation, ravines on sediment soils, and variability in larger-scale topography from steep mountains to flat mires or smaller agricultural areas interspersed in the forest landscape. Although our model produced a fair amount of false positives in the sedimentary deposits in the Swedish test catchment, the overall performance of the model was notably better than the previous ditch mapping studies that were conducted in relatively less challenging, more homogeneous landscapes. Some of the false positives in our prediction might even be actual ditches that are missing or misplaced in the ground-truth digitized data [blue lines in Figs. 3(d and f)]. This is likely because of the differences in digitization approaches in the international datasets. Hence, the accuracy measures calculated from the international validation datasets should be interpreted cautiously. The high MCC and recall values of ditch prediction in the international test sites indicate that the developed methodology performs well in all countries around the Baltic Sea, although the model was developed based on training data from Sweden. The trained model can be adjusted for local conditions with transfer learning using local data.Visual inspection of the test sites revealed that some of the false positives were natural stream channels, as demonstrated in Fig. 3(b). This suggests that it could be possible to train a deep neural network to detect natural stream channels in addition to drainage ditches. This needs to be explored in the future with additional ground-truth data on natural stream channels. A model that can map natural stream channels would complement traditional topographical modeling of accumulated flows from digital terrain models, especially if the model can map headwater stream channels, as these are often missing on current maps (Benstead and Leigh 2012), and deal with road embankments, which make small streams difficult to map using traditional topographical modeling of low accumulation (Lidberg et al. 2017).The management of water systems is key to sustainable development, but headwater streams (Bishop et al. 2008) and artificial drainage ditches are often not included in monitoring programs (Koschorreck et al. 2020). This has large implications for future research and practical land-use management, as the ditch networks have legacy effects on greenhouse gas balance, water resources, and biodiversity. Ditches possess particular characteristics that often make them emit large amounts of greenhouse gases (Peacock et al. 2021). Specifically, the accumulation of sediment and the development of anoxia result in favorable conditions for the production of the potent greenhouse gas methane (Roulet and Moore 1995). Emissions of nitrous oxide (N2O) and carbon dioxide (CO2) can be disproportionately large compared with the small areal extent of ditch surfaces (Audet et al. 2017; Peacock et al. 2019). In countries such as Sweden and Finland, which have significant areas of land drained for agriculture or forestry, these greenhouse gas emissions can make nonnegligible contributions to national greenhouse gas budgets (Koschorreck et al. 2020). Further, it is unclear whether peatland restoration or ditch cleaning to maintain forest growth is the best strategy to avoid negative environmental outcomes in the future. Mapping the drainage ditches in the forest landscape is necessary to address these important sustainability questions.References Audet, J., M. B. Wallin, K. Kyllmar, S. Andersson, and K. Bishop. 2017. “Nitrous oxide emissions from streams in a Swedish agricultural catchment.” Agric. Ecosyst. Environ. 236 (Jan): 295–303. Ayana, E. K., J. R. B. Fisher, P. Hamel, and T. M. Boucher. 2017. “Identification of ditches and furrows using remote sensing: Application to sediment modelling in the Tana watershed, Kenya.” Int. J. Remote Sens. 38 (16): 4611–4630. Bailly, J. S., P. Lagacherie, C. Millier, C. Puech, and P. Kosuth. 2008. “Agrarian landscapes linear features detection from LiDAR: Application to artificial drainage networks.” Int. J. Remote Sens. 29 (12): 3489–3508. Bailly, J. S., F. Levavasseur, and P. Lagacherie. 2011. “A spatial stochastic algorithm to reconstruct artificial drainage networks from incomplete network delineations.” Int. J. Appl. Earth Obs. Geoinf. 13 (6): 853–862. Balado, J., J. Martínez-Sánchez, P. Arias, and A. Novo. 2019. “Road environment semantic segmentation with deep learning from mls point cloud data.” Sensors 19 (16): 3466. Bhattacharjee, J., H. Marttila, A. T. Haghighi, M. Saarimaa, A. Tolvanen, A. Lepistö, M. N. Futter, and B. Kløve. 2021. “Development of aerial photos and LIDAR data approaches to map spatial and temporal evolution of ditch networks in peat-dominated catchments.” J. Irrig. Drain. Eng. 147 (4): 04021006. Bishop, K., I. Buffam, M. Erlandsson, J. Fölster, H. Laudon, J. Seibert, and J. Temnerud. 2008. “Aqua Incognita: The unknown headwaters.” Hydrol. Processes 22 (8): 1239. Broersen, T., R. Peters, and H. Ledoux. 2017. “Automatic identification of watercourses in flat and engineered landscapes by computing the skeleton of a LiDAR point cloud.” Comput. Geosci. 106 (Sep): 171–180. Cazorzi, F., G. D. Fontana, A. D. Luca, G. Sofia, and P. Tarolli. 2013. “Drainage network detection and assessment of network storage capacity in agrarian landscape.” Hydrol. Processes 27 (4): 541–553. Chollet, F. 2017. Xception: Deep learning with depthwise separable convolutions. Ithaca, NY: Cornell University Library. Flykt, J., F. Anderson, N. Lavesson, L. Nilsson, and M. Ågren. 2022. “Detecting ditches using supervised learning on high-resolution digital elevation models.” Expert Syst. Appl. 201 (Sep): 116961. Garcia-Garcia, A., S. Orts-Escolano, S. O. Oprea, V. Villena-Martinez, and J. Garcia-Rodriguez. 2017. “A review on deep learning techniques applied to semantic segmentation.” Preprint, submitted April 22, 2017. Graves, J., R. Mohapatra, and N. Flatgard. 2020. “Drainage ditch berm delineation using lidar data: A case study of Waseca County, Minnesota.” Sustainability 12 (22): 9600. Hasselquist, E. M., W. Lidberg, R. A. Sponseller, A. Ågren, and H. Laudon. 2018. “Identifying and assessing the potential hydrological function of past artificial forest drainage.” Ambio 47 (5): 546–556. Holden, J., P. J. Chapman, and J. C. Labadz. 2004. “Artificial drainage of peatlands: Hydrological and hydrochemical process and wetland restoration.” Prog. Phys. Geogr. 28 (1): 95–123. Jensen, C. K., K. J. McGuire, and P. S. Prince. 2017. “Headwater stream length dynamics across four physiographic provinces of the Appalachian highlands.” Hydrol. Processes 31 (19): 3350–3363. Julian, J. P., A. J. Elmore, and S. M. Guinn. 2012. “Channel head locations in forested watersheds across the mid-Atlantic United States: A physiographic analysis.” Geomorphology 177 (Dec): 194–203. Koschorreck, M., A. S. Downing, J. Hejzlar, R. Marcé, A. Laas, W. G. Arndt, P. S. Keller, A. J. P. Smolders, G. van Dijk, and S. Kosten. 2020. “Hidden treasures: Human-made aquatic ecosystems harbour unexplored opportunities.” Ambio 49 (2): 531–540. Kuglerová, L., J. Jyväsjärvi, C. Ruffing, T. Muotka, A. Jonsson, E. Andersson, and J. S. Richardson. 2020. “Cutting edge: A comparison of contemporary practices of riparian buffer retention around small streams in Canada, Finland, and Sweden.” Water Resour. Res. 56 (9): e2019WR026381. Larson, J., and M. Trivedi. 2011. “Lidar based off-road negative obstacle detection and analysis.” In Proc., 14th Int. IEEE Conf. on Intelligent Transportation Systems (ITSC), 192–197. New York: IEEE. Laurén, A., M. Palviainen, S. Launiainen, K. Leppä, L. Stenberg, I. Urzainki, M. Nieminen, R. Laiho, and H. Hökkä. 2021. “Drainage and stand growth response in peatland forests—description, testing, and application of mechanistic peatland simulator susi.” Forests 12 (3): 1–23. Lepistö, A., A. Räike, T. Sallantaus, and L. Finér. 2021. “Increases in organic carbon and nitrogen concentrations in boreal forested catchments—Changes driven by climate and deposition.” Sci. Total Environ. 780 (Aug): 146627. Lidberg, W., M. Nilsson, T. Lundmark, and A. M. Ågren. 2017. “Evaluating preprocessing methods of digital elevation models for hydrological modeling.” Hydrol. Processes 31 (26): 4660–4668. Lidberg, W., F. Westphal, and A. Ågren. 2021. Mapping-drainage-ditches-in-forested-landscapes-using-deep-learning-and-aerial-laser-scanning. San Francisco: GitHub, Inc. Lidman, F., Å. Boily, H. Laudon, and S. J. Köhler. 2017. “From soil water to surface water—how the riparian zone controls element transport from a boreal forest to a stream.” Biogeosciences 14 (12): 3001–3014. Lindsay, J. B. 2018. WhiteboxTools user manual. Guelph, Canada: Univ. of Guelph. Lõhmus, A., L. Remm, and R. Rannap. 2015. “Just a ditch in forest? Reconsidering draining in the context of sustainable forest management.” Bioscience 65 (11): 1066–1076. McCabe, M. F., B. Aragon, R. Houborg, and J. Mascaro. 2017. “CubeSats in hydrology: Ultrahigh-resolution insights into vegetation dynamics and terrestrial evaporation.” Water Resour. Res. 53 (12): 10017–10024. Nieminen, M., M. Palviainen, S. Sarkkola, A. Laurén, H. Marttila, and L. Finér. 2018. “A synthesis of the impacts of ditch network maintenance on the quantity and quality of runoff from drained boreal peatland forests.” Ambio 47 (5): 523–534. Passalacqua, P., P. Belmont, and E. Foufoula-Georgiou. 2012. “Automatic geomorphic feature extraction from lidar in flat and engineered landscapes.” Water Resour. Res. 48 (3): 1–18. Peacock, M., et al. 2021. “Small artificial waterbodies are widespread and persistent emitters of methane and carbon dioxide.” Global Change Biol. 27 (20): 5109–5123. Qian, T., D. Shen, C. Xi, J. Chen, and J. Wang. 2018. “Extracting farmland features from LiDAR-derived DEM for improving flood plain delineation.” Water 10 (3): 252. Rapinel, S., L. Hubert-Moy, B. Clément, J. Nabucet, and C. Cudennec. 2015. “Ditch network extraction and hydrogeomorphological characterization using LiDAR-derived DTM in wetlands.” Hydrol. Res. 46 (2): 276–290. Roelens, J., B. Höfle, S. Dondeyne, J. Van Orshoven, and J. Diels. 2018a. “Drainage ditch extraction from airborne LiDAR point clouds.” ISPRS J. Photogramm. Remote Sens. 146 (Dec): 409–420. Roelens, J., I. Rosier, S. Dondeyne, J. Van Orshoven, and J. Diels. 2018b. “Extracting drainage networks and their connectivity using LiDAR data.” Hydrol. Processes 32 (8): 1026–1037. Roulet, N. T., and T. R. Moore. 1995. “The effect of forestry drainage practices on the emission of methane from northern peatlands.” Can. J. For. Res. 25 (3): 491–499. Russell, P. P., S. M. Gale, B. Muñoz, J. R. Dorney, and M. J. Rubino. 2015. “A spatially explicit model for mapping headwater streams.” J. Am. Water Resour. Assoc. 51 (1): 226–239. Shen, C., et al. 2018. “HESS opinions: Incubating deep-learning-powered hydrologic science advances as a community.” Hydrol. Earth Syst. Sci. 22 (11): 5639–5656. Sikström, U., and H. Hökkä. 2016. “Interactions between soil water conditions and forest stands in boreal forests with implications for ditch network maintenance.” Silva Fenn. 50 (1): 1416. Sit, M., B. Z. Demiray, Z. Xiang, G. J. Ewing, Y. Sermet, and I. Demir. 2020. “A comprehensive review of deep learning applications in hydrology and water resources.” Water Sci. Technol. 82 (12): 2635–2670. Ståhl, G., et al. 2011. “National Inventory of Landscapes in Sweden (NILS)—Scope, design, and experiences from establishing a multiscale biodiversity monitoring system.” Environ. Monit. Assess. 173 (1): 579–595. Stanislawski, L., T. Brockmeyer, and E. Shavers. 2018. “Automated road breaching to enhance extraction of natural drainage networks from elevation models through deep learning.” Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 42 (4): 671–678. Zheng, S., S. Jayasumana, B. Romera-Paredes, V. Vineet, Z. Su, D. Du, C. Huang, and P. H. S. Torr. 2015. “Conditional random fields as recurrent neural networks.” In Proc., IEEE Int. Conf. on Computer Vision 2015, 1529–1537. New York: IEEE.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *