Decision Support Systems in Forestry and Tree-Planting Practices: History
View Latest Version
Please note this is an old version of this entry, which may differ significantly from the current revision.
Contributor:
Neelesh Yadav
,
Shrey Rakholia
,
Reuven Yosef

Using deep neural networks (DNNs), a decision support system (DSS) can be trained to learn from a large dataset of tree data, including information about tree species, climate, soil conditions, and other factors that influence tree growth and survival. This is because the use of neural networks was proposed three decades ago to solve forest management problems by integrating forest knowledge with artificial intelligence (AI). AI greatly benefits sustainability and the preservation of ecosystem values, as increasing disruptions in a changing world can only be managed beyond human intelligence. Furthermore, despite the various DSSs and AI systems used, the appointment of appropriate project managers is crucial to the execution and subsequent success of a project.

  • decision support system
  • climate resilience
  • ecosystem services
  • deep neural networks
  • sustainability

1. Introduction

One of the earliest DSSs for tree plantations in forestry was developed at the University of Canterbury: a framework-based system coded in Prolog. The focus was on knowledge-based decision support by linking to the Forest Management Information System (FMIS) or Geographic Information System (GIS) databases, enabling location-based access to information about the field microenvironment such as soils, climate, elevation, and earlier land/crop use and current conditions, along with multiple management options for optimization [1].
Further efforts to develop a DSS for tree plantations began in the 2000s using a GIS with a focus on street and neighborhood tree plantations, while attempting to address management aspects such as DSS-based strategies to reduce energy, fuel, and pesticides/fertilizers for plantation management [2]. In addition, the focus was also expanded to include aspects such as soil-property-based tree planting, feasibility of the planting area, tree age, species diversity, shade, and canopy cover [3]. It is also important to conduct an existing urban tree cover (UTC) analysis prior to tree-planting decisions, using object-oriented satellite image analysis to identify existing vegetation cover and land-use types [4].
Mitigating a region’s hydrological problems also requires selecting appropriate species, prioritizing sites for re-vegetation, and simulating different hydro-climatological conditions annually. These aspects were incorporated into China’s bilingual GUI decision support tool for re-vegetation programs, ReVegIH, which could also reduce sediment load release through afforestation modeling [5].
A multilingual programming (C++ and Fortran)-based DSS known as the Motti Simulator, developed by the Natural Resources Institute Finland (Luke), has also been used for tree selection based on detailed forest stand dynamics and incorporating tree growth and yield models [6]. Additionally, simplified open-source and open-code DSSs such as PT2 (Prairie and Tree Planting Tool) have allowed users to explore and delineate areas of interest for tree/prairie planting or management using scaled dimensional drawing tools, and then select seeds/woody plants for the various soils with a drop-down menu. This also enabled the selection and calculation of financial costs and long-term management options [7].
Nevertheless, advances in machine learning in recent years have enabled the selection of tree species taking into account climate variability using MaxEnt to determine the suitability and resilience of trees in different climate scenarios. A recent example is the online platform “Which Plant Where” in Australia, which was developed using Python, Django, and PostgreSQL [8]. In addition, others use tree-selection tools developed by the United States Department of Agriculture (USDA) such as the Tree Advisor and the Woody Plant Selection Tool for multi-functional purposes, using MySQL and the Drupal framework [9]. In addition, a spatio-temporal urban tree DSS was developed using ensemble CAD and GIS tools. This integrates detailed 3D trees into urban design, allowing the testing of tree placement, species selection, solar exposure, etc. Valuable elements of computational botany and lighting engineering technology make this possible [10].
Although tree-planting decision support systems have addressed tree-selection ecosystem services such as UHI mitigation, only simple filtering techniques with limited variables that filter the attributes from tree databases have been used [11]. In addition, ensemble models that use higher-resolution datasets to infer the potential suitability and realized distribution of tree species through batch generalization are also proposed. This is a boosting method that uses random forest (RF), gradient-booster trees (GBTs), and generalized linear models (GLMs), which are further processed by the meta-learner [12].

2. The Need for an Ecosystem-Services-Focused DSS

It is crucial to understand the ecosystem services received from trees during selection and planting, as trees provide various regulatory (carbon sequestration, air pollution reduction) and provisioning services (timber, tree crops). Non-market values sometimes exceed commercial values and threats, such as forest fires and pests, and this must be taken into account for resilience [13]. Additionally, models such as the Natural Capital Protocol can be applied to improve agroforestry decision making and evaluation at the farm level. They describe the connection between a natural capital, its condition, the resulting ecosystem services, and the benefits that people derive from these services. Better benefit representation can also promote the public benefits of agroforestry at the farm level [14].
One of the most important ecosystem services is flood protection, which can be improved by riparian forests as part of agroforestry (e.g., riparian buffers), providing the same benefits at almost 30% of the cost compared to an engineered protection structure, as shown in a study in Germany [15]. Satellite datasets and IDF-based (Intensity, Duration and Frequency) flood models can provide valuable information about the flooding and water logging situation in regions experiencing monsoons and persistent floods. The areas affected by flooding and erosion can be identified using flood depth and flow velocity forecasts for 25-, 50- and 100-year return periods [16]. Therefore, the selection of tree species adapted to this water logging must be assessed based on the literature that evaluates parameters such as stomatal conductance and net photosynthesis, since some tree species show a reduction in these two processes after flooding [17]. In addition, trees such as poplars in riparian zones are very tolerant of flooding because nitrogen metabolism is not affected by flooding compared to species such as oak and beech, which are sensitive to successive flooding, and the depth and duration of flooding must also be taken into account in detail [18].
It is important to understand the dynamics of the UHI effect. There are regional and zonal differences, including in urban areas, because although trees are effective in reducing air temperature in areas with high building density, they are ineffective in built-up areas with low building density, and therefore high-density trees with taller trunks are recommended for built-up areas [19]. Changes in land use and land cover can influence local surface temperatures. For example, as previously irrigated croplands and forests transform into built-up urban areas over time, this can lead to increases in air and land surface temperatures (LSTs). Conversely, a transition from bare land cover to urban areas could reduce the average LST for semi-arid regions [20][21]. This highlights the significant influence of both vegetation and urban development on LSTs at the local scale. The vegetation has a cooling effect through transpiration, shading, and rainwater retention.
Similarly, urbanized zones contribute more to temperature reduction than regions with exposed ground or rocky terrain due to their surface properties and materials that promote convection more effectively [22]. There is a unique approach to UHI mitigation that involves creating a regional Heat Vulnerability Index (HVI) that incorporates socioeconomic (family income, age, building density) and environmental data (e.g., LST, vegetation) for decision making [23], which helps to increase urban tree canopy cover with the most suitable tree species. To mitigate UHI, urban areas need to be divided into high- and low-density areas because land use and tree availability are limited in cities.
Furthermore, nature-based solutions (NbSs) to air pollution can be implemented zone-wise by involving plantations. Air-pollution-tolerant species such as Shorea robusta, Ficus religiosa, and Mangifera indica have high tolerance to pollutants and high metal accumulation capacity in industrial areas. Dust removal and deposition are excellent in residential areas in Azadirachta indica, Dalbergia sissoo, and Ficus religiosa [24]. Tools for slope protection and landslide mitigation include Plant-Best, which was developed in the statistical programming language R [25].
Many factors influence tree plantation, including the value and placement of trees, particularly in urban areas. This includes public lands, parks, and roadsides, as well as private land, i.e., residential properties [26]. Kirkpatrick et al. [27] suggested that small fruit trees on private property were more aesthetically pleasing and practical. A study on agroforestry found that the management of forests involves significant uncertainty regarding future timber prices, tree growth, and the impact of climate-related changes on tree growth. Because most forest owners prefer to avoid risk and tree growth and timber prices are unpredictable, the study suggests the following implication: longer rotations should be compared to recommended guidelines. There may be a greater preference for mixed stands than deterministic calculations suggest; the concentration of timber revenues should be less focused on the final harvest, as currently recommended. The consistent retention of multiple timber assortments in the inventory is advantageous, which indicates that more uneven stand structures should be pursued [28].
Therefore, the suitability process must include mixed stands and not just monoculture recommendations. However, this may not be the case for all tree species as agarwood monoculture plantations could also be favorable in terms of growth, as they are endangered [29]. Nevertheless, plantation agriculture in tropical countries must be managed on the basis of polyculture systems and not monocultures since the ecosystem services provided by the former are much better, as they include the improvement of biodiversity, pollination, and biological pest control even in the context of small-scale silviculture [30]. Hirsch et al. [31] found species-specific tolerance to drought and traffic pollution in urban areas, suggesting the use of certain tree species along roads and in residential areas.
DSSs such as the FADSS (Florida Agroforestry DSS) dealt with economic and environmental services and used GIS databases that contained important datasets on tree attributes, infrastructure, climate, soil, and cropping, including critical levels such as key agroforestry management practices [32]. It is also important to include soil datasets on soil pH, sand content, etc., for tree species distribution models (SDMs) as soil variables are strong predictors of habitat suitability [33]. Soil datasets are often neglected in many SDMs, so these datasets should be some of the core variables in decision support systems. Finally, recent developments in tree selection DSSs include the Diversity for Restoration (D4R) tool, which allows users to make multiple selections from a menu for restoration objectives, ecosystem services, seeding zones, climate, and other environmental data on decision-making for individual and combined tree species selection [34].
Therefore, by incorporating rich ecosystem services, DSS tools are enriched with more data-driven and knowledge-driven capabilities, introducing complexities in these systems that can then be addressed and improved through the implementation of DNNs, as explained in the following section.

3. Proposed Use of DNN in DSS for Tree Selection/Plantation

In order to improve decision making in urban forestry for sustainable and livable cities, AI has been increasingly used as part of smart technology in recent years [35]. However, only half of the studies using AI manage to take into account aspects such as the limitations of methods, including robustness and lack of precision in some datasets, the combined use of discrete and continuous data variables, overfitting, collinearity, etc. [36]. The application of AI in forestry can be improved by incorporating the XAI (Explainable Artificial Intelligence), LTNL (Learning To Not Learn), and FUL (Feature Unlearning) methods which allow the qualitative and quantitative comparison of model accuracy and explanations through the use of predefined annotation matrices, i.e., expert knowledge that can improve these deep learning models. Therefore, the combination of XAI, FUL, and expert knowledge can improve the understanding of how the model works, instead of only obtaining simple model results [37].
In addition, the use of CNNs (convolutional neural networks) is increasing significantly with a large number of applications in agriculture/agroforestry DSSs generally based on frameworks such as Keras, Tensorflow, Tensorflow-Keras, PyTorch, Tensorflow-PyTorch, and Deeplearning 4j [38]. In addition, the applications of DNNs for intelligent geographic data analysis in DSSs in agriculture have shown promising results, especially when Back-Propagation Neural Network (BPNN)-based prediction models are used to predict agricultural indicator values [39][40]. In addition, DNN-based species distribution models show better results than traditional models, including DNNs built using bootstraps to improve the prediction performance of species distribution. These can be built in the Python environment using the Scikit-learn package with bootstrapping aggregation (bagging) performed in the R statistics package boot to train the DNN [41]. Regardless, CNN-based SDMs offer broader advantages, including better learning of non-linear environmental descriptors, compelling distribution predictions of environmental descriptors, and the use of high-dimensional data, enabling an improved collection of information about environmental landscapes structured on tensors, rather than local values of environmental factors [42]. Likewise, the ecosystem service component of a tree plantation DSS can be better understood and improved through these tensors [43], i.e., different functions of multiple vectors (as ecosystem services include multiple services and complex relationships, such as between the existing environment and land use) can be considered in one vector. Ecosystem services can be viewed as multi-linear functions of the vector [44].
TensorFlow uses the term “Tensor” to denote the primary data structure used in deep learning algorithms. This “Tensor” represents a multi-dimensional array of numerical values [45]. In addition, deep neural networks have been widely used in recent years. This rise in popularity of deep learning models is due to TensorFlow, an open-source deep learning framework, as this framework offers users the ability to rely on pre-defined, network-trained deep learning classification (and regression) models while enabling the customizable training of their personalized or custom datasets [46].
The TensorFlow Deep Neural Network (TF-DNN) is used in the Python environment as the primary model of this study because TF-DNN has been applied in GIS studies that have shown higher spatial prediction accuracy than other techniques such as random forest (RF), support vector machines (SVMs), and logistic regression (LR) [47]. The TF-DNN can be applied with semi-supervised learning with a multivariate multilayer perceptron with training datasets, where the soil, climate, and landscape environmental layers can be used to determine the land suitability of the plant species in the study, with the results providing continuously better decision-making potential when validated through K-fold cross-validation [48].
For the proper implementation of the TF-DNN, it is important to use multiple libraries, including TensorFlow, Keras, NumPy, and Matplotlib. Keras is used as a backend to build and implement the TF-DNN algorithm, while TensorFlow acts as a numerical computing library. The Numpy library is useful for many mathematical functions that operate on arrays, and Matplotlib is similarly used to visualize statistical outputs [49].
Therefore, the use of DNNs is crucial for improving the precision and effectiveness of DSSs and contributes to sustainable and informed tree selection and plantation strategies in both urban and regional environments.

This entry is adapted from the peer-reviewed paper 10.3390/land13020230

References

  1. Mason, E.G. Planning forest establishment operations with a computerised decision-support system: A case study analysis of decision-making over a full rotation. NZJ For. Sci 1996, 26, 222–234.
  2. Randall, T.A.; Churchill, C.J.; Baetz, B.W. A GIS-based decision support system for neighbourhood greening. Environ. Plan. B Plan. Des. 2003, 30, 541–563.
  3. Kirnbauer, M.C.; Kenney, W.A.; Churchill, C.J.; Baetz, B.W. A prototype decision support system for sustainable urban tree planting programs. Urban For. Urban Green. 2009, 8, 3–19.
  4. McGee, J.A., III; Day, S.D.; Wynne, R.H.; White, M.B. Using geospatial tools to assess the urban tree canopy: Decision support for local governments. J. For. 2012, 110, 275–286.
  5. McVicar, T.R.; Li, L.; Van Niel, T.G.; Zhang, L.; Li, R.; Yang, Q.; Zhang, X.P.; Mu, X.M.; Wen, Z.M.; Liu, W.Z.; et al. Developing a decision support tool for China’s re-vegetation program: Simulating regional impacts of afforestation on average annual streamflow in the Loess Plateau. For. Ecol. Manag. 2007, 251, 65–81.
  6. Saksa, T.; Uusitalo, J.; Lindeman, H.; Häyrynen, E.; Kulju, S.; Huuskonen, S. Decision support tool for tree species selection in forest regeneration based on harvester data. Forests 2021, 12, 1329.
  7. Tyndall, J. Prairie and Tree Planting Tool—PT 2 (1.0): A Conservation Decision Support Tool for Iowa, USA. Agrofor. Syst. 2021, 96, 49–64.
  8. Tabassum, S.; Beaumont, L.J.; Shabani, F.; Staas, L.; Griffiths, G.; Ossola, A.; Leishman, M.R. Which Plant Where: A Plant Selection Tool for Changing Urban Climates. Arboric. Urban For. 2023, 49, 190–210.
  9. Bentrup, G.; Dosskey, M.G. Tree Advisor: A Novel Woody Plant Selection Tool to Support Multifunctional Objectives. Land 2022, 11, 397.
  10. White, M.R.; Langenheim, N. A spatio-temporal decision support system for designing with street trees. In Geospatial Intelligence: Concepts, Methodologies, Tools, and Applications; IGI Global: Hershey, PA, USA, 2019; pp. 533–560.
  11. Werbin, Z.R.; Heidari, L.; Buckley, S.; Brochu, P.; Butler, L.J.; Connolly, C.; Bloemendaal, L.H.; McCabe, T.D.; Miller, T.K.; Hutyra, L.R. A tree-planting decision support tool for urban heat mitigation. PLoS ONE 2020, 15, e0224959.
  12. Bonannella, C.; Hengl, T.; Heisig, J.; Parente, L.; Wright, M.N.; Herold, M.; De Bruin, S. Forest tree species distribution for Europe 2000–2020: Mapping potential and realized distributions using spatiotemporal machine learning. PeerJ 2022, 10, e13728.
  13. Cavender-Bares, J.M.; Nelson, E.; Meireles, J.E.; Lasky, J.R.; Miteva, D.A.; Nowak, D.J.; Pearse, W.D.; Helmus, M.R.; Zanne, M.E.; Fagan, W.F.; et al. The hidden value of trees: Quantifying the ecosystem services of tree lineages and their major threats across the contiguous US. PLOS Sustain. Transform. 2022, 1, e0000010.
  14. Marais, Z.E.; Baker, T.P.; O’Grady, A.P.; England, J.R.; Tinch, D.; Hunt, M.A. A natural capital approach to agroforestry decision-making at the farm scale. Forests 2019, 10, 980.
  15. Barth, N.C.; Döll, P. Assessing the ecosystem service flood protection of a riparian forest by applying a cascade approach. Ecosyst. Serv. 2016, 21, 39–52.
  16. Quirogaa, V.M.; Kurea, S.; Udoa, K.; Manoa, A. Application of 2D numerical simulation for the analysis of the February 2014 Bolivian Amazonia flood: Application of the new HEC-RAS version 5. Ribagua 2016, 3, 25–33.
  17. Anderson, P.H.; Pezeshki, S.R. Effects of flood pre-conditioning on responses of three bottomland tree species to soil waterlogging. J. Plant Physiol. 2001, 158, 227–233.
  18. Kreuzwieser, J.; Fürniss, S.; Rennenberg, H. Impact of waterlogging on the N-metabolism of flood tolerant and non-tolerant tree species. Plant Cell Environ. 2002, 25, 1039–1049.
  19. Aboelata, A.; Sodoudi, S. Evaluating the effect of trees on UHI mitigation and reduction of energy usage in different built up areas in Cairo. Build. Environ. 2020, 168, 106490.
  20. Yosef, R.; Rakholia, S.; Mehta, A.; Bhatt, A.; Kumbhojkar, S. Land Surface temperature regulation ecosystem service: A case study of Jaipur, India, and the urban island of Jhalana Forest Reserve. Forests 2022, 13, 1821.
  21. Sahdev, S.; Yosef, R.; Rakholia, S.; Mehta, A.; Yadav, N. Bidecadal Analysis of Ecosystem Services and Urbanization Effects in the Western Himalayas: A Case Study from Haldwani, Uttarakhand. Remote Sens. Appl. Soc. Environ. 2023. accepted for publication.
  22. Rasul, A.; Balzter, H.; Smith, C. Applying a normalized ratio scale technique to assess influences of urban expansion on land surface temperature of the semi-arid city of Erbil. Int. J. Remote Sens. 2017, 38, 3960–3980.
  23. Johnson, D.P.; Stanforth, A.; Lulla, V.; Luber, G. Developing an applied extreme heat vulnerability index utilizing socioeconomic and environmental data. Appl. Geogr. 2012, 35, 23–31.
  24. Menon, J.S.; Sharma, R. Nature-based solutions for co-mitigation of air pollution and urban heat in Indian cities. Front. Sustain. Cities 2021, 3, 705185.
  25. Gonzalez-Ollauri, A.; Mickovski, S.B. Plant-Best: A novel plant selection tool for slope protection. Ecol. Eng. 2017, 106, 154–173.
  26. Kowarik, I. The Mediterranean tree Acer manspessulanum invades urban green spaces in Berlin. Dendrobiology 2023, 89, 20–26.
  27. Kirkpatrick, J.B.; Davison, A.; Daniels, G.D. Resident attitudes towards trees influence the planting and removal of different types of trees in eastern Australian cities. Landsc. Urban Plan. 2012, 107, 147–158.
  28. Pukkala, T.; Kellomäki, S. Anticipatory vs adaptive optimization of stand management when tree growth and timber prices are stochastic. For. Int. J. For. Res. 2012, 85, 463–472.
  29. Nath, P.C.; Nath, A.J.; Sileshi, G.W.; Das, A.K. Growth and Coppicing Ability of the Critically Endangered Agarwood (Aquilaria malaccensis Lam.) Tree in Monoculture and Polyculture in North East India. J. Sustain. For. 2022, 42, 947–959.
  30. Yahya, M.S.; Syafiq, M.; Ashton-Butt, A.; Ghazali, A.; Asmah, S.; Azhar, B. Switching from monoculture to polyculture farming benefits birds in oil palm production landscapes: Evidence from mist netting data. Ecol. Evol. 2017, 7, 6314–6325.
  31. Hirsch, M.; Boddeker, H.; Albrecht, A.; Saha, S. Drought tolerance differs between urban tree species but is not affected by the intensity of traffic pollution. Trees 2023, 37, 111–131.
  32. Ellis, E.A.; Nair, P.K.R.; Linehan, P.E.; Beck, H.W.; Blanche, C.A. A GIS-based database management application for agroforestry planning and tree selection. Comput. Electron. Agric. 2000, 27, 41–55.
  33. Henderson, A.F.; Santoro, J.A.; Kremer, P. Impacts of spatial scale and resolution on species distribution models of American chestnut (Castanea dentata) in Pennsylvania, USA. For. Ecol. Manag. 2023, 529, 120741.
  34. Fremout, T.; Thomas, E.; Taedoumg, H.; Briers, S.; Gutiérrez-Miranda, C.E.; Alcázar-Caicedo, C.; Lindau, A.; Kpoumie, H.M.; Vincenti, B.; Kettle, C.; et al. Diversity for Restoration (D4R): Guiding the selection of tree species and seed sources for climate-resilient restoration of tropical forest landscapes. J. Appl. Ecol. 2022, 59, 664–679.
  35. Nitoslawski, S.A.; Galle, N.J.; Van Den Bosch, C.K.; Steenberg, J.W. Smarter ecosystems for smarter cities? A review of trends, technologies, and turning points for smart urban forestry. Sustain. Cities Soc. 2019, 51, 101770.
  36. de Lima Araújo, H.C.; Martins, F.S.; Cortese, T.T.P.; Locosselli, G.M. Artificial intelligence in urban forestry—A systematic review. Urban For. Urban Green. 2021, 66, 127410.
  37. Cheng, X.; Doosthosseini, A.; Kunkel, J. Improve the deep learning models in forestry based on explanations and expertise. Front. Plant Sci. 2022, 13, 902105.
  38. Altalak, M.; Ammad Uddin, M.; Alajmi, A.; Rizg, A. Smart agriculture applications using deep learning technologies: A survey. Appl. Sci. 2022, 12, 5919.
  39. Zeng, C.; Zhang, F.; Luo, M. A deep neural network-based decision support system for intelligent geospatial data analysis in intelligent agriculture system. Soft Comput. 2022, 26, 10813–10826.
  40. Araujo, S.O.; Peres, R.S.; Filipe, L.; Manta-Costa, A.; Lidon, F.; Ramalho, J.C.; Barata, J. Intelligent data-driven decision support for agricultural systems-ID3SAS. IEEE Access 2023, 11, 115798–115815.
  41. Rew, J.; Cho, Y.; Hwang, E. A robust prediction model for species distribution using bagging ensembles with deep neural networks. Remote Sens. 2021, 13, 1495.
  42. Deneu, B.; Servajean, M.; Bonnet, P.; Botella, C.; Munoz, F.; Joly, A. Convolutional neural networks improve species distribution modelling by capturing the spatial structure of the environment. PLoS Comput. Biol. 2021, 17, e1008856.
  43. Unawong, W.; Yaemphum, S.; Nathalang, A.; Chen, Y.; Domec, J.-C.; Torngern, P. Variations in leaf water status and drought tolerance of dominant tree species growing in multi-aged tropical forests in Thailand. Sci. Rep. 2022, 12, 6882.
  44. Zhang, P.; Ren, H.; Dong, X.; Wang, X.; Liu, M.; Zhang, Y.; Zhang, Y.; Huang, J.; Dong, S.; Xiao, R. Understanding and Applications of Tensors in Ecosystem Services: A Case Study of the Manas River Basin. Land 2023, 12, 454.
  45. Singh, P.; Manure, A. Learn TensorFlow 2.0: Implement Machine Learning and Deep Learning Models with Python; Apress: New York, NY, USA, 2019.
  46. Pally, R.J.; Samadi, S. Application of image processing and convolutional neural networks for flood image classification and semantic segmentation. Environ. Model. Softw. 2022, 148, 105285.
  47. Truong, T.X.; Nhu, V.H.; Phuong, D.T.N.; Nghi, L.T.; Hung, N.N.; Hoa, P.V.; Bui, D.T. A New Approach Based on TensorFlow Deep Neural Networks with ADAM Optimizer and GIS for Spatial Prediction of Forest Fire Danger in Tropical Areas. Remote Sens. 2023, 15, 3458.
  48. Bhullar, A.; Nadeem, K.; Ali, R.A. Simultaneous multi-crop land suitability prediction from remote sensing data using semi-supervised learning. Sci. Rep. 2023, 13, 6823.
  49. Osah, S.; Acheampong, A.A.; Fosu, C.; Dadzie, I. Deep learning model for predicting daily IGS zenith tropospheric delays in West Africa using TensorFlow and Keras. Adv. Space Res. 2021, 68, 1243–1262.
More
© Text is available under the terms and conditions of the Creative Commons Attribution (CC BY) license; additional terms may apply. By using this site, you agree to the Terms and Conditions and Privacy Policy.
This entry is offline, you can click here to edit this entry!
ScholarVision Creations
Feedback