Evaluation of river basins requires land-use and land-cover (LULC) change detection to determine hydrological and ecological conditions for sustainable use of their resources. This study assessed LULC changes over 28 years (1990–2018) in the Wami–Ruvu Basin, located in Tanzania, Africa. Six pairs of images acquired using Landsat 5 TM and 8 OLI sensors in 1990 and 2018, respectively, were mosaicked into a single composite image of the basin. A supervised classification using the Neural Network classifier and training data was used to create LULC maps for 1990 and 2018, and targeted the following eight classes of agriculture, forest, grassland, bushland, built-up, bare soil, water, and wetland. The results show that over the past three decades, water and wetland areas have decreased by 0.3%, forest areas by 15.4%, and grassland by 6.7%, while agricultural, bushland, bare soil, and the built-up areas have increased by 11.6%, 8.2%, 1.6%, and 0.8%, respectively. LULC transformations were assessed with water discharge, precipitation, and temperature, and the population from 1990 to 2018. The results revealed decreases in precipitation, water discharge by 4130 m³, temperature rise by 1 °C, and an increase in population from 5.4 to 10 million. For proper management of water-resources, we propose three strategies for water-use efficiency-techniques, a review legal frameworks, and time-based LULC monitoring. This study provides a reference for water resources sustainability for other countries with basins threatened by LULC changes.

Studying land use changes and ecological risk assessment in Yongjiang River Basin in Zhejiang Province, China, provides theoretical references for optimal configuration of land resources and maintaining stability of ecosystems. Given impacts of land use changes on landscape patterns in the Yongjiang River Basin, ecological risk assessment indexes were constructed and used to analyze temporal and spatial variation characteristics of ecological risk within different periods. Results show that (1) the construction land area was increased quickly, while the cultivated area decreased sharply. A prominent characteristic of land use changes was manifested by transforming cultivated area and forestland into construction land. The utilized degree of the land increased continuously. Spatially, the land utilized degree in northern regions was higher than that in southern regions and the degree in eastern regions was higher than that in western regions. (2) The ecological risk in the Yongjiang River Basin was intensified and the area of high ecological risk was expanded by 893.96 km². Regions with low and relatively low ecological risks concentrated in western and southern regions of the Basin, whereas regions with high ecological risks were mainly in northern and eastern regions. Landscapes in cities and towns at a high economic development level are highly sensitive to human activities. (3) Transformation of ecological risk is complicated. Land area with the ecological risk changing from a low level to a high level was 4.15 times that with the ecological risk changing from a high level to a low level. There were 15 transformation directions among different ecological risk regions.

The landscape grain effect reflects the spatial heterogeneity of a landscape and it is used as a research core of landscape ecology. The landscape grain effect can be used to not only explore spatiotemporal variation characteristics of a landscape pattern, but also to disclose variation laws of ecological structures and functions of landscapes. In this study, the sensitivity of landscape pattern indexes to grain sizes 50–1000 m was studied based on landscape data in Yancheng Coastal Wetland acquired in 1991, 2000, 2008, and 2017. Response of the grain effect to landscape changes was analyzed and an optimal grain size for analysis in the study area was determined. Results indicated that: (1) among 27 indexes (12 in a class level and 15 in a landscape level), eight indexes were highly sensitive to grains, ten indexes presented moderate sensitivity, eight indexes presented low sensitivity, and one was unresponsive. It was shown that the area-margin index and the shape index were more sensitive to the different grain sizes. The aggregation index had some differences in the grain size change, and the diversity index had a low response degree to the grain size. (2) Landscape indexes showed six different responses to different grains, including slow reduced response, fast reduced and then slow reduced response, monotonically increased response, fluctuating reduced response, up-down responses, and stable response, which indicated that the landscape index was closely related to the spatial grain. (3) From 1991 to 2017, variation curves of the landscape grain size of different landscape types could be divided into four types: fluctuation rising type, fluctuation type, monotonous decreasing type, and monotonous rising type. Different grain size curves had different interpretations of landscape changes, but in general, Yancheng Coastal Wetland’s landscape tended to be fragmented and complicated, internal connectivity was weakened, and dominant landscape area was reduced. Natural wetlands were more sensitive to grain size effects than artificial wetlands. (4) The landscape index at the 50 m grain size had a strong response to different grain size changes, and the loss of landscape information was the smallest. Therefore, it was determined that the optimal landscape grain size in the study area was 50 m.

Detecting the stability of nearshore subaqueous geomorphology is a crucial challenge for ensuring early warning and controlling the stability of riverbank slopes. Acquiring nearshore subaqueous geomorphology data using unmanned ship-mounted acoustic multibeam systems is difficult, costly, and time-consuming. Moreover, it is often influenced by weather conditions. The limited availability of nearshore subaqueous geomorphology samples suitable for model training, combined with the high similarity between targets of nearshore unstable geomorphology and the background, poses significant challenges for traditional detection methods. In response to issues such as high similarity in subaqueous geomorphology images, large-scale variations in target size, and a scarcity of samples, this study proposes a nearshore subaqueous geomorphology instability detection framework based on Few-shot learning. Firstly, a feature extraction network is designed, replacing the backbone network with a Swin Transformer network. This network employs a feature pyramid network to extract multi-scale geomorphology features containing global information from the query set, facilitating the fusion of features across deep and shallow layers. Secondly, a weight adjustment module is devised to transform the support set into weight coefficients with class attributes. This adjustment helps in adapting the distribution of geomorphology features for detecting new class objects. Experimental results demonstrate that the proposed detection framework achieves desirable performance in terms of average precision and average recall indicators.Keywords: Subaqueous Geomorphology; Stability Detection; Few-shot learning