Convolutional neural network (CNN)-based deep learning (DL) is a powerful, recently developed image classification approach. With origins in the computer vision and image processing communities, the accuracy assessment methods developed for CNN-based DL use a wide range of metrics that may be unfamiliar to the remote sensing (RS) community. To explore the differences between traditional RS and DL RS methods, we surveyed a random selection of 100 papers from the RS DL literature. The results show that RS DL studies have largely abandoned traditional RS accuracy assessment terminology, though some of the accuracy measures typically used in DL papers, most notably precision and recall, have direct equivalents in traditional RS terminology. Some of the DL accuracy terms have multiple names, or are equivalent to another measure. In our sample, DL studies only rarely reported a complete confusion matrix, and when they did so, it was even more rare that the confusion matrix estimated population properties. On the other hand, some DL studies are increasingly paying attention to the role of class prevalence in designing accuracy assessment approaches. DL studies that evaluate the decision boundary threshold over a range of values tend to use the precision-recall (P-R) curve, the associated area under the curve (AUC) measures of average precision (AP) and mean average precision (mAP), rather than the traditional receiver operating characteristic (ROC) curve and its AUC. DL studies are also notable for testing the generalization of their models on entirely new datasets, including data from new areas, new acquisition times, or even new sensors.

Historic topographic maps, which are georeferenced and made publicly available by the United States Geological Survey (USGS) and the National Map’s Historical Topographic Map Collection (HTMC), are a valuable source of historic land cover and land use (LCLU) information that could be used to expand the historic record when combined with data from moderate spatial resolution Earth observation missions. This is especially true for landscape disturbances that have a long and complex historic record, such as surface coal mining in the Appalachian region of the eastern United States. In this study, we investigate this specific mapping problem using modified UNet semantic segmentation deep learning (DL), which is based on convolutional neural networks (CNNs), and a large example dataset of historic surface mine disturbance extents from the USGS Geology, Geophysics, and Geochemistry Science Center (GGGSC). The primary objectives of this study are to (1) evaluate model generalization to new geographic extents and topographic maps and (2) to assess the impact of training sample size, or the number of manually interpreted topographic maps, on model performance. Using data from the state of Kentucky, our findings suggest that DL semantic segmentation can detect surface mine disturbance features from topographic maps with a high level of accuracy (Dice coefficient = 0.902) and relatively balanced omission and commission error rates (Precision = 0.891, Recall = 0.917). When the model is applied to new topographic maps in Ohio and Virginia to assess generalization, model performance decreases; however, performance is still strong (Ohio Dice coefficient = 0.837 and Virginia Dice coefficient = 0.763). Further, when reducing the number of topographic maps used to derive training image chips from 84 to 15, model performance was only slightly reduced, suggesting that models that generalize well to new data and geographic extents may not require a large training set. We suggest the incorporation of DL semantic segmentation methods into applied workflows to decrease manual digitizing labor requirements and call for additional research associated with applying semantic segmentation methods to alternative cartographic representations to supplement research focused on multispectral image analysis and classification.

Small-scale forestry builds upon interactions among local stakeholders. Forest management entails multiple social situations such as consultations or cooperative engagements between owners and forest professionals. Successful social endeavours rest on positive social capital as operationalised via trust. Based on qualitative in-depth interviews with forest owners, managers and other forestry stakeholders, this study explores how trust influences the social relationships in a local context of Southern Swedish forestry. Most strikingly, the analysis reveals large differences in owners' trust towards two major actors: the Swedish Forest Agency (SFA) and the forest owner association (FOA) Södra. Permanence of personnel, a client-based approach, and personal features of SFA's local forest officer lead to strong local anchoring and high trust towards SFA. Södra proved to be a trustful partner in the aftermath of calamities; however its industrial priorities seem to erode owners' trust. The empirical findings of this study demonstrate the importance of recognising personal relationships and the catalysing role of bonding social capital in order to understand the local forest management situations. Our results are useful for forestry organisations and policy-makers willing to comprehend the local context and implement best practices in small-scale forestry.

Convolutional neural network (CNN)-based deep learning (DL) has a wide variety of applications in the geospatial and remote sensing (RS) sciences, and consequently has been a focus of many recent studies. However, a review of accuracy assessment methods used in recently published RS DL studies, focusing on scene classification, object detection, semantic segmentation, and instance segmentation, indicates that RS DL papers appear to follow an accuracy assessment approach that diverges from that of traditional RS studies. Papers reporting on RS DL studies have largely abandoned traditional RS accuracy assessment terminology; they rarely reported a complete confusion matrix; and sampling designs and analysis protocols generally did not provide a population-based confusion matrix, in which the table entries are estimates of the probabilities of occurrence of the mapped landscape. These issues indicate the need for the RS community to develop guidance on best practices for accuracy assessment for CNN-based DL thematic mapping and object detection. As a first step in that process, we explore key issues, including the observation that accuracy assessments should not be biased by the CNN-based training and inference processes that rely on image chips. Furthermore, accuracy assessments should be consistent with prior recommendations and standards in the field, should support the estimation of a population confusion matrix, and should allow for assessment of model generalization. This paper draws from our review of the RS DL literature and the rich record of traditional remote sensing accuracy assessment research while considering the unique nature of CNN-based deep learning to propose accuracy assessment best practices that use appropriate sampling methods, training and validation data partitioning, assessment metrics, and reporting standards.

Monitoring wildland fire burn severity is important for assessing ecological outcomes of fire and their spatial patterning as well as guiding efforts to mitigate or restore areas where ecological outcomes are negative. Burn severity mapping products are typically created using satellite reflectance data but must be calibrated to field data to derive meaning. The composite burn index (CBI) is the most widely used field-based method used to calibrate satellite-based burn severity data but important limitations of this approach have yet to be resolved. The objective of this study was focused on predicting CBI from point cloud and visible-spectrum camera (RGB) metrics derived from single-scan terrestrial laser scanning (TLS) datasets to determine the viability of TLS data as an alternative approach to estimating burn severity in the field. In our approach, we considered the predictive potential of post-scan-only metrics, differenced pre- and post-scan metrics, RGB metrics, and all three together to predict CBI and evaluated these with candidate algorithms (i.e., linear model, random forest (RF), and support vector machines (SVM) and two evaluation criteria (R-squared and root mean square error (RMSE)). In congruence with the strata-based observations used to calculate CBI, we evaluated the potential approaches at the strata level and at the plot level using 70 TLS and 10 RGB independent variables that we generated from the field data. Machine learning algorithms successfully predicted total plot CBI and strata-specific CBI; however, the accuracy of predictions varied among strata by algorithm. RGB variables improved predictions when used in conjunction with TLS variables, but alone proved a poor predictor of burn severity below the canopy. Although our study was to predict CBI, our results highlight that TLS-based methods for quantifying burn severity can be an improvement over CBI in many ways because TLS is repeatable, quantitative, faster, requires less field-expertise, and is more flexible to phenological variation and biomass change in the understory where prescribed fire effects are most pronounced. We also point out that TLS data can also be leveraged to inform other monitoring needs beyond those specific to wildland fire, representing additional efficiency in using this approach.