The Yellow River Delta region is one of the estuarine deltas with the fastest land building speed, and it is an important region for the study of landscape pattern change due to its diverse variety of landscape types. By analyzing the dynamic degree, landscape type transfer matrix, and landscape indices of landscape types in the Yellow River Delta region in 2005, 2012, and 2018, this paper found that the area of construction land, salt fields, and breeding ponds in the Yellow River Delta region has increased to a large extent, with an increase in the aggregation degree and the utilization rate of this landscape type, and the landscape has developed toward the direction of aggregation and unification. The increase in construction land area mainly comes from the transfer of cropland area, part of which is occupied in order to adapt to urban expansion, and the salt fields and breeding ponds mainly come from the transfer of waters and mudflats, which can be seen as the main utilization direction of the water landscape. Moreover, unused land has increased with the degree of dispersion and fragmentation of development and utilization, so the exploitation and utilization of unused land still needs to be optimized. Through the analysis of the dynamic change in landscape pattern, we can explore the direction and extent of the evolution of landscape types, which has certain guiding significance for the sustainable use of land resources and the sustainable development of economy in the Yellow River Delta region.

An efficient, convenient, and accurate method for monitoring the distribution characteristics of soil salinity is required to effectively control the damage of saline soil to the land environment and maintain a virtuous cycle of the ecological environment. There are still problems with single-monitoring data that cannot meet the requirements of different regional scales and accuracy, including inconsistent band reflectance between multi-source sensor data. This article proposes a monitoring method based on the multi-source data fusion of unmanned aerial vehicle (UAV) multispectral remote sensing, Sentinel-2A satellite remote sensing, and ground-measured salinity data. The research area and two experimental fields were located in the Yellow River Delta (YRD). The results show that the back-propagation neural network model (BPNN) in the comprehensive estimation model is the best prediction model for soil salinity (modeling accuracy R² reaches 0.769, verification accuracy R² reaches 0.774). There is a strong correlation between the satellite and UAV imagery, while the Sentinel-2A imagery after reflectivity correction has a superior estimation effect. In addition, the results of dynamic analysis show that the area of non-saline soil and mild-saline soil decreased, while the area of moderately and heavily saline soils and solonchak increased. Additionally, the average area share of different classes of saline soils distributed over the land use types varied in order, from unused land > grassland > forest land > arable land, where the area share of severe-saline soil distributed on unused land changed the most (89.142%). In this study, the results of estimation are close to the true values, which supports the feasibility of the multi-source data fusion method of UAV remote sensing satellite ground measurements. It not only achieves the estimation of soil salinity and monitoring of change patterns at different scales, but also achieve high accuracy of soil salinity prediction in ascending scale regions. It provides a theoretical scientific basis for the remediation of soil salinization, land use, and environmental protection policies in coastal areas.