Dimensionality reduction (DR) is an important preprocessing step in hyperspectral image applications. In this paper, a superpixelwise kernel principal component analysis (SuperKPCA) method for DR that performs kernel principal component analysis (KPCA) on each homogeneous region is proposed to fully utilize the KPCA’s ability to acquire nonlinear features. Moreover, for the proposed method, the differences in the DR results obtained based on different fundamental images (the first principal components obtained by principal component analysis (PCA), KPCA, and minimum noise fraction (MNF)) are compared. Extensive experiments show that when 5, 10, 20, and 30 samples from each class are selected, for the Indian Pines, Pavia University, and Salinas datasets: (1) when the most suitable fundamental image is selected, the classification accuracy obtained by SuperKPCA can be increased by 0.06%–0.74%, 3.88%–4.37%, and 0.39%–4.85%, respectively, when compared with SuperPCA, which performs PCA on each homogeneous region; (2) the DR results obtained based on different first principal components are different and complementary. By fusing the multiscale classification results obtained based on different first principal components, the classification accuracy can be increased by 0.54%–2.68%, 0.12%–1.10%, and 0.01%–0.08%, respectively, when compared with the method based only on the most suitable fundamental image.

Aiming at the problem of poor image effect in traditional image automatic generation technology, this paper proposes an automatic generation method of multimedia animation image based on surface stitching. Firstly, the principle of surface stitching is analyzed. On this basis, the feature vectors in the acquired image basic information are matched. According to the matching process, the camera is positioned by using the initial position calibration method, and the pose offset parameters are obtained. The mutual information between the collected image information and the generated image is taken as the similarity measure, and the spatial transformation parameters of the surface image to be registered are obtained Finally, the automatic generation of multimedia animation image is realized. The experimental results show that the matching degree of feature points is more than 87%, the image accuracy is up to 90%, and the average time consumption is 167s, which has good performance.

Considering time-lag and accumulation effects of climate is crucial for accurately evaluating vegetation dynamics under global climate change. Most studies investigate these mechanisms by examining the explanatory power of the overall climatic conditions of the same month of the day at which the vegetation index is recorded, or the preceding month(s). This approach, referred to as monthly climate approach, risks underestimating the importance of within-month vegetation index variations. This study introduces an alternative approach, the EVI-date climate approach, which considers climate data from a specified time period up to the EVI recording day. The explanatory power is investigated for temperature, precipitation, and solar radiation at a global scale. EVI-date climate generally shows stronger vegetation-climate relationships than monthly climate. The relative improvement in adjusted R² ranges from 2.86 % to 39.3 %, and it is especially significant at northern high latitude when the EVI typically varies greatly (May, June, September). Using EVI-date climate, the highest explanatory power for vegetation greenness is generally found with the overall conditions over the 30 or 60 days preceding EVI recording date for temperature and solar radiation, and the preceding 60-120 days for precipitation. Overall, using the optimal time spans of climatic conditions preceding EVI recording date improves the explanatory power by 16.3 %-22.3 % compared to the use of the preceding 30-days climate only. By better aligning climate variables with vegetation greenness, climate-vegetation interaction predictions can be improved, enhancing our insights into ecosystem changes driven by climate shifts and human impact.

Spring phenology of vegetation affects key ecosystem processes and properties, yet uncertainties remain regarding the fitting performance and phenological sensitivity of different modelling approaches globally. This study compares a process-agnostic multiple linear regression (MLR) model and a process-based modified growing degree day (mGDD) model in capturing global variability in start-of-season (SOS) dates and their climate drivers from 2003 to 2024. Globally, SOS advances by an average of 1.15 days per decade. Both models exhibit lower root mean squared error (RMSE) between modeled and satellite-observed SOS in temperature-limited regions of the Northern Hemisphere than in water- or radiation-limited regions. Both models show that warming advances SOS mainly north of 30°N, while increased precipitation advances SOS mainly south of 50°N, with stronger SOS sensitivity to temperature under higher preseason temperature and to precipitation under lower preseason precipitation. Compared with the linear MLR model, the nonlinear mGDD model yields lower RMSE and stronger SOS sensitivity to temperature in regions with lower SOS variability, whereas MLR performs better and shows stronger temperature sensitivity in regions with higher SOS variability, and identifies stronger precipitation sensitivity across most analyzed areas (67.0%). MLR indicates that warming predominantly delays SOS south of 30°N, whereas mGDD implies more widespread delays, although advances remain dominant overall. For precipitation, MLR indicates a predominantly delaying effect on SOS north of 50°N, while mGDD suggests a mainly advancing effect. These findings highlight that estimates of SOS sensitivity to temperature and precipitation are strongly shaped by model structure, underscoring the critical role of model choice in global phenological analyses.