Search

There are a variety of land cover products generated from remote-sensing images. However, misclassification errors in individual products and inconsistency among them undermine their utilities for research and other applications. While it is worth developing advanced pattern classifiers and utilizing the images of finer spatial, temporal, and/or spectral resolution for increased classification accuracy, it is also sensible to increase map classification accuracy through effective map fusion by exploiting complementarity among multi-source products over a study area. This paper presents a novel fusion method that works by weighting multiple source products based on their map-reference cover type transition probabilities, which are predicted using random forest for individual map pixels. The proposed method was tested and compared with three alternatives: consensus-based weighting, random forest, and locally modified Dempster–Shafer evidential reasoning, in a case study, over Shaanxi province, China. For this case study, three types of land cover products (GlobeLand30, FROM-GLC, and GLC_FCS30) of two nominal years (2010 and 2020) were used as the base maps for fusion. Reference sample data for model training and testing were collected following a robust stratified random sampling design that allows for augmenting reference data flexibly. Accuracy assessments show that overall accuracies (OAs) of fused land cover maps have been improved (1~9% in OAs), with the proposed method outperforming other methods by 2~8% in OAs. The proposed method does not need to have the base products' classification systems harmonized beforehand, thus being robust and highly recommendable for fusing land cover products. [ABSTRACT FROM AUTHOR]

Accelerated land use change is a current challenge for environmental management worldwide. Given the urgent need to incorporate economic and ecological goals in landscape planning, cost-effective conservation strategies are required. In this study, we validated the benefit of fusing imagery from multiple sensors to assess the impact of landscape changes on ecosystem services (ES) and their economic values in the Long County, Shaanxi Province, China. We applied several landscape metrics to assess the local spatial configuration over 15 years (2004–2019) from fused imageries. Using Landsat-7 Enhanced Thematic Mapper Plus (ETM+), Landsat-8 Operational Land Imager (OLI) and Indian Remote Sensing Satellite System Linear Imaging Self Scanning Sensor 3 (IRS LISS 3) imageries fused for 2004, 2009, 2014 and 2019, we reclassified land use/land cover (LULC) changes, through the rotation forest (RF) machine-learning algorithm. We proposed an equivalent monetary metric for estimating the ES values, which also could be used in the whole China. Results showed that agriculture farmland and unused land decreased their spatial distribution over time, with an observed increase on woodland, grassland, water bodies and built-up area. Our findings suggested that the patterns of landscape uniformity and connectivity improved, while the distribution of landscape types stabilized, while the landscape diversity had a slight improvement. The overall ES values increased (4.34%) under a benefit transfer approach, mainly concerning woodland and grassland. A sensitivity analysis showed the selected economic value (EV) was relevant and suitable for the study area associated with our ES for LULC changes. We suggested that changes in landscape patterns affected the ESV trends, while the increases on some LULC classes slightly improved the landscape diversity. Using an interdisciplinary approach, we recommend that local authorities and environmental practitioners should balance the economic benefits and ecological gains in different landscapes to achieve a sustainable development from local to regional scales. [ABSTRACT FROM AUTHOR]