Land cover information is necessary for a large range of environmental applications related to climate impacts and adaption, emergency response, wildlife habitat, etc. In Canada, a 2008 user survey indicated that the most practical land cover data is provided in a nationwide 30 m spatial resolution format, with an update frequency of five years. In response to this need, the Canada Centre for Remote Sensing (CCRS) has generated a 30 m land cover map of Canada for the base year 2010, as the first of a planned series of maps to be updated every five years, or more frequently. This land cover dataset is also the Canadian contribution to the 30 m spatial resolution 2010 Land Cover Map of North America, which is produced by Mexican, American and Canadian government institutions under a collaboration called the North American Land Change Monitoring System (NALCMS). This paper describes the mapping approach used for generating this land cover dataset for Canada from Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) Landsat sensor observations. The innovative part of the mapping approach is the local optimization of the land cover classifier, which has resulted in increased spatial consistency and accuracy. Training and classifying with locally confined reference samples over a large number of partially overlapping areas (i.e., moving windows) ensures the optimization of the classifier to a local land cover distribution, and decreases the negative effect of signature extension. A weighted combination of labels, which is determined by the classifier in overlapping windows, defines the final label for each pixel. Since the approach requires extensive computation, it has been developed and deployed using the Government of Canada’s High-Performance Computing Center (HPC). An accuracy assessment based on 2811 randomly distributed samples shows that land cover data produced with this new approach has achieved 76.60% accuracy with no marked spatial disparities.

Landsat is a fundamental data source for understanding historical change and its effect on environmental processes. In this research we test shallow and deep convolution neural networks (CNNs) for Landsat image super-resolution enhancement, trained using Sentinel-2, in three study sites representing boreal forest, tundra, and cropland/woodland environments. The analysis sought to assess baseline performance and determine the capacity for spatial and temporal extension of the trained CNNs. This is not a data fusion approach and a high-resolution image is only needed to train the CNN. Results show improvement with the deeper network generally achieving better results. For spatial and temporal extension, the deep CNN performed the same or better than the shallow CNN, but at greater computational cost. Results for temporal extension were influenced by change potentiality reducing the performance difference between the shallow and deep CNN. Visual examination revealed sharper images regarding land cover boundaries, linear features, and within-cover textures. The results suggest that spatial enhancement of the Landsat archive is feasible, with optimal performance where CNNs can be trained and applied within the same spatial domain. Future research will assess the enhancement on time series and associated land cover applications.

Mapping of surficial geology is an important requirement for broadening the geoscience database of northern Canada. Surficial geology maps are an integral data source for mineral and energy exploration. Moreover, they provide information such as the location of gravels and sands, which are important for infrastructure development. Currently, surficial geology maps are produced through expert interpretation of aerial photography and field data. However, interpretation is known to be subjective, labour-intensive and difficult to repeat. The expert knowledge required for interpretation can be challenging to maintain and transfer. In this research, we seek to assess the potential of deep neural networks to aid surficial geology mapping by providing an objective surficial materials initial layer that experts can modify to speed map development and improve consistency between mapped areas. Such an approach may also harness expert knowledge in a way that is transferable to unmapped areas. For this purpose, we assess the ability of convolution neural networks (CNN) to predict surficial geology classes under two sampling scenarios. In the first scenario, a CNN uses samples collected over the area to be mapped. In the second, a CNN trained over one area is then applied to locations where the available samples were not used in training the network. The latter case is important, as a collection of in situ training data can be costly. The evaluation of the CNN was carried out using aerial photos, Landsat reflectance, and high-resolution digital elevation data over five areas within the South Rae geological region of Northwest Territories, Canada. The results are encouraging, with the CNN generating average accuracy of 76% when locally trained. For independent test areas (i.e., trained over one area and applied over other), accuracy dropped to 59–70% depending on the classes selected for mapping. In the South Rae region, significant confusion was found between till veneer and till blanket as well as glaciofluvial subclasses (esker, terraced, and hummocky ice-contact). Merging these classes respectively increased accuracy for independent test area to 68% on average. Relative to the more widely used Random Forest machine learning algorithm, this represents an improvement in accuracy of 4%. Furthermore, the CNN produced better results for less frequent classes with distinct spatial structure.

Methods for effective wetland monitoring are needed to understand how ecosystem services may be altered from past and present anthropogenic activities and recent climate change. The large extent of wetlands in many regions suggests remote sensing as an effective means for monitoring. Remote sensing approaches have shown good performance in local extent studies, but larger regional efforts have generally produced low accuracies for detailed classes. In this research we evaluate the potential of deep-learning Convolution Neural Networks (CNNs) for wetland classification using Landsat data to bog, fen, marsh, swamp, and water classes defined by the Canada Wetland Classification System (CWCS). The study area is the northern part of the forested region of Alberta where we had access to two reference data sources. We evaluated ResNet CNNs and developed a Multi-Size/Scale ResNet Ensemble (MSRE) approach that exhibited the best performance. For assessment, a spatial extension strategy was employed that separated regions for training and testing. Results were consistent between the two reference sources. The best overall accuracy for the CWCS classes was 62–68%. Compared to a pixel-based random forest implementation this was 5–7% higher depending on the accuracy measure considered. For a parameter-optimized spatial-based implementation this was 2–4% higher. For a reduced set of classes to water, wetland, and upland, overall accuracy was in the range of 86–87%. Assessment for sampling over the entire region instead of spatial extension improved the mean class accuracies (F1-score) by 9% for the CWCS classes and for the reduced three-class level by 6%. The overall accuracies were 69% and 90% for the CWCS and reduced classes respectively with region sampling. Results in this study show that detailed classification of wetland types with Landsat remains challenging, particularly for small wetlands. In addition, further investigation of deep-learning methods are needed to identify CNN configurations and sampling methods better suited to moderate spatial resolution imagery across a range of environments.

Optical satellite imagery is often contaminated by the persistent presence of clouds and atmospheric haze. Without an effective method for removing this contamination, most optical remote sensing applications are less reliable. In this research, a methodology has been developed to fully automate and improve the Haze Optimized Transformation (HOT)-based haze removal. The method is referred to as AutoHOT and characterized with three notable features: a fully automated HOT process, a novel HOT image post-processing tool and a class-based HOT radiometric adjustment method. The performances of AutoHOT in haze detection and compensation were evaluated through three experiments with one Landsat-5 TM, one Landsat-7 ETM+ and eight Landsat-8 OLI scenes that encompass diverse landscapes and atmospheric haze conditions. The first experiment confirms that AutoHOT is robust and effective for haze detection. The average overall, user’s and producer’s accuracies of AutoHOT in haze detection can reach 96.4%, 97.6% and 97.5%, respectively. The second and third experiments demonstrate that AutoHOT can not only accurately characterize the haze intensities but also improve dehazed results, especially for brighter targets, compared to traditional HOT radiometric adjustment.

Global land cover information is required to initialize land surface and Earth system models. In recent years, new land cover (LC) datasets at finer spatial resolutions have become available while those currently implemented in most models are outdated. This study assesses the applicability of the Climate Change Initiative (CCI) LC product for use in the Canadian Land Surface Scheme (CLASS) through comparison with finer resolution datasets over Canada, assisted with reference sample data and a vegetation continuous field tree cover fraction dataset. The results show that in comparison with the finer resolution maps over Canada, the 300 m CCI product provides much improved LC distribution over that from the 1 km GLC2000 dataset currently used to provide initial surface conditions in CLASS. However, the CCI dataset appears to overestimate needleleaf forest cover especially in the taiga-tundra transition zone of northwestern Canada. This may have partly resulted from limited availability of clear sky MEdium Resolution Imaging Spectrometer (MERIS) images used to generate the CCI classification maps due to the long snow cover season in Canada. In addition, changes based on the CCI time series are not always consistent with those from the MODIS or a Landsat-based forest cover change dataset, especially prior to 2003 when only coarse spatial resolution satellite data were available for change detection in the CCI product. It will be helpful for application in global simulations to determine whether these results also apply to other regions with similar landscapes, such as Eurasia. Nevertheless, the detailed LC classes and finer spatial resolution in the CCI dataset provide an improved reference map for use in land surface models in Canada. The results also suggest that uncertainties in the current cross-walking tables are a major source of the often large differences in the plant functional types (PFT) maps, and should be an area of focus in future work.