Optimised U-Net for Land Use–Land Cover Classification Using Aerial Photography
Convolutional Neural Networks (CNN) consist of various hyper-parameters which need to be specifed or can be altered when defning a deep learning architecture. There are numerous studies which have tested diferent types of networks (e.g. U-Net, DeepLabv3+) or created new architectures, benchmarked against well-known test datasets. However, there is a lack of real-world mapping applications demonstrating the efects of changing network hyper-parameters on model performance for land use and land cover (LULC) semantic segmentation. In this paper, we analysed the efects on training time and classifcation accuracy by altering parameters such as the number of initial convolutional flters, kernel size, network depth, kernel initialiser and activation functions, loss and loss optimiser functions, and learning rate. We achieved this using a well-known top performing architecture, the U-Net, in conjunction with LULC training data and two multispectral aerial images from North Queensland, Australia. A 2018 image was used to train and test CNN models with diferent parameters and a 2015 image was used for assessing the optimised parameters. We found more complex models with a larger number of flters and larger kernel size produce classifcations of higher accuracy but take longer to train. Using an accuracy-time ranking formula, we found using 56 initial flters with kernel size of 5×5 provide the best compromise between training time and accuracy. When fully training a model using these parameters and testing on the 2015 image, we achieved a kappa score of 0.84. This compares to the original U-Net parameters which achieved a kappa score of 0.73.
Detecting Banana Plantations in the Wet Tropics, Australia, Using Aerial Photography and U-Net
Bananas are the world's most popular fruit and an important staple food source. Recent outbreaks of Panama TR4 disease are threatening the global banana industry, which is worth an estimated $8 billion. Current methods to map land uses are time- and resource-intensive and result in delays in the timely release of data. We have used existing land use mapping to train a U-Net neural network to detect banana plantations in the Wet Tropics of Queensland, Australia, using high-resolution aerial photography. Accuracy assessments, based on a stratified random sample of points, revealed the classification achieves a user’s accuracy of 98% and a producer's accuracy of 96%. This is more accurate compared to existing (manual) methods, which achieved a user’s and producer's accuracy of 86% and 92% respectively. Using a neural network is substantially more efficient than manual methods and can inform a more rapid respond to existing and new biosecurity threats. The method is robust and repeatable and has potential for mapping other commodities and land uses which is the focus of future work.
A Multifaceted Approach to Developing an Australian National Map of Protected Cropping Structures
Abstract: As the global population rises, there is an ever-increasing demand for food, in terms of volume, quality and sustainable production. Protected Cropping Structures (PCS) provide controlled farming environments that support the optimum use of crop inputs for plant growth, faster production cycles, multiple growing seasons per annum and increased yield, while offering greater control of pests, disease and adverse weather. Globally, there has been a rapid increase in the adoption of PCS. However, there remains a concerning knowledge gap in the availability of accurate and up-to-date spatial information that defines the extent (location and area) of PCS. This data is fundamental for providing metrics that inform decision making around forward selling, labour, processing and infrastructure requirements, traceability, biosecurity and natural disaster preparedness and response. This project addresses this need, by developing a national map of PCS for Australia using remotely sensed imagery and deep learning analytics, ancillary data, field validation and industry engagement. The resulting map presents the location and extent of all commercial glasshouses, polyhouses, polytunnels, shadehouses and permanent nets with an area of >0.2 ha. The outcomes of the project revealed deep learning techniques can accurately map PCS with models achieving F-Scores > 0.9 and accelerate the mapping where suitable imagery is available. Location-based tools supported by web mapping applications were critical for the validation of PCS locations and for building industry awareness and engagement. The final national PCS map is publicly available through an online dashboard which summarises the area of PCS structures at a range of scales including state/territory, local government area and individual structure. The outcomes of this project have set a global standard on how this level of mapping can be achieved through a collaborative, multifaceted approach.
Identification of dust transport pathways from Lake Eyre, Australia using Hysplit
The HYbrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT_4) is used to create seasonal climatologies (1980–2000) of air parcel trajectories from the Southern Hemisphere's most active dust source Lake Eyre, Australia. Daily trajectories were computed forward for eight days from an origin centered over Lake Eyre at 500 m above the ground surface. Trajectory density maps were then created within a GIS for five levels; 0–500 m agl., 500–1000 m agl., 1000–1500 m agl., 1500–2000 m agl. and 2000–5000 m agl. These show that air parcel trajectories originating from Lake Eyre can affect regions many thousands of kilometers from the Australian continent in a relatively short period of time. Importantly, under favourable atmospheric conditions these air parcels have the potential to transport dust and other aerosols. During the austral winter, trajectories extended north to the southern Philippines highlighting the potential for dust from central Australia to affect most of Indonesia. This includes the tropical rainforests of Borneo and New Guinea, and the coral reefs of northern Australia and the Indonesian archipelago. We also show the potential for transport of dust from Lake Eyre to the Antarctic and much of the South Pacific and Southern Oceans. Accordingly, dust from Lake Eyre may affect biogeochemical cycles, sediment budgets, atmospheric processes and a wide range of ecosystems over a region much larger than previously thought. This highlights the need for further research to confirm the deposition of dust in the areas mapped by the present study.
A vertical profile of PM10 dust concentrations measured during a regional dust event identified by MODIS Terra, western Queensland, Australia
Accurate determination of the spatiotemporal properties of dust plumes and their dust concentrations is essential for calibration of satellite products and the initialization and validation of numerical models that simulate the physical properties and affects of dust events. In this paper, we present a 500 m vertical profile of PM10 dust concentrations measured during a regional dust event in western Queensland, Australia. PM10 dust concentrations within the haze were found to be >20 times background ambient values and decreased with height following an exponential function. We apply an over-land algorithm to MODIS Terra satellite images of the dust haze to enhance its visual appearance against the bright land surface and define its size. In conjunction with the measured attenuation of dust concentrations with height we calculate the PM10 dust load of the plume to be ~60% of that which would have been calculated assuming a constant dust concentration up to the dust ceiling height. Results extend previous findings from tower-based studies made close to the surface and confirm that atmospheric dust concentrations decrease rapidly with increasing height, thereby enabling more accurate calculation of atmospheric dust loads during synoptic-scale dust outbreaks.
A Paddock to reef monitoring and modelling framework for the Great Barrier Reef: Paddock and catchment component
Targets for improvements in water quality entering the Great Barrier Reef (GBR) have been set through the Reef Water Quality Protection Plan (Reef Plan). To measure and report on progress towards the targets set a program has been established that combines monitoring and modelling at paddock through to catchment and reef scales" the Paddock to Reef Integrated Monitoring, Modelling and Reporting Program (Paddock to Reef Program). This program aims to provide evidence of links between land management activities, water quality and reef health. Five lines of evidence are used: the effectiveness of management practices to improve water quality; the prevalence of management practice adoption and change in catchment indicators; long-term monitoring of catchment water quality; paddock & catchment modelling to provide a relative assessment of progress towards meeting targets; and finally marine monitoring of GBR water quality and reef ecosystem health. This paper outlines the first four lines of evidence.
Pre-Processing Training Data Improves Accuracy and Generalisability of Convolutional Neural Network Based Landscape Semantic Segmentation
Data pre-processing for developing a generalised land use and land cover (LULC) deep learning model using earth observation data is important for the classification of a different date and/or sensor. However, it is unclear how to approach deep learning segmentation problems in earth observation data. In this paper, we trialled different methods of data preparation for Convolutional Neural Network (CNN) training and semantic segmentation of LULC features within aerial photography over the Wet Tropics and Atherton Tablelands, Queensland, Australia. This was conducted by trialling and ranking various training patch selection sampling strategies, patch and batch sizes, data augmentations and scaling and inference strategies. Our results showed: a stratified random sampling approach for producing training patches counteracted class imbalances" a smaller number of larger patches (small batch size) improves model accuracy" data augmentations and scaling are imperative in creating a generalised model able to accurately classify LULC features in imagery from a different date and sensor" and producing the output classification by averaging multiple grids of patches and three rotated versions of each patch produced a more accurate and aesthetic result. Combining the findings from the trials, we fully trained five models on the 2018 training image and applied the model to the 2015 test image. The output LULC classifications achieved an average kappa of 0.84, user accuracy of 0.81, and producer accuracy of 0.87. Future research using CNNs and earth observation data should implement the findings of this project to increase LULC model accuracy and transferability.
Alternatives to Landsat-5 Thematic Mapper for operational monitoring of vegetation cover: considerations for natural resource management agencies
La possibilite´ que le capteur Thematic Mapper (TM) de Landsat 5 ne puisse fournir des images jusqu'au moment du lancement de la mission LDCM (« Landsat Data Continuity Mission ») pre´vu pour de´cembre 2012 suscite des inquie´tudes. La crainte vient du manque de combustible et des proble`mes sporadiques du satellite et du capteur. Le mauvais fonctionnement du correcteur de lignes de balayage (SLC) du capteur ETM+ (« Enhanced Thematic Mapper ») de Landsat 7 en 2003 fait que les images de ETM+ renferment des bandes de donne´es manquantes. Conse´quemment, on craint que les images de qualite´ supe´rieure de Landsat ne soient pas disponibles pour fins de suivi durant les anne´es 2011 et 2012. Les agences de suivi des ressources naturelles font ainsi face au de´fi conside´rable de de´terminer une alternative ade´quate aux images de TM de Landsat 5 en cas de besoin. Bien qu'il existe une litte´rature sur les alternatives possibles, il y a peu d'e´tudes qui donnent des directives aux agences de suivi sur les proble´matiques a` explorer lorsque l'on conside`re une source de donne´es alternatives. On a entrepris une e´tude qui a permis d'identifier les donne´es de ETM+ de Landsat 7, HRVIR de SPOT 4, HRG de SPOT 5 et de LISS-III de IRS-P6 comme e´tant les alternatives les mieux adapte´es pour le suivi du couvert de ve´ge´tation dans le Queensland et la Nouvelle-Galles du Sud (NSW) dans l'est de l'Australie. On a trouve´ qu'aucun des capteurs n'e´tait ide´al a` cause d'une combinaison de l'un ou de plusieurs des facteurs suivants dont la qualite´ radiome´trique re´duite, les volumes de donne´es accrus, les besoins de traitement supple´mentaires et les couˆ ts d'acquisition plus e´leve´s. On a trouve´ que le faible couˆ t et la facilite´ d'acce`s des donne´es de ETM+ de Landsat 7 rendaient cette option plus viable aux plans technique et e´conomique pour le suivi annuel de l'e´tendue et deschangements de la ve´ge´tation ligneuse dans l'est de l'Australie. Cependant, ETM+ de Landsat 7 ne peut eˆtre conside´re´ comme une option dans plusieurs re´gions du monde duˆ a` l'importance du couvert nuageux. Nous avons utilise´ les expe´riences acquises durant ce travail pour recommander une proce´dure que les agences de suivi des ressources naturelles peuvent mettre de l'avant pour explorer les proble´matiques relie´es au choix d'une alternative aux donne´es de TM de Landsat 5 pour le suivi du couvert.