Recently, convolutional neural networks (CNNs) and self-attention mechanisms have been widely applied in plant disease identification tasks, yielding significant successes. Currently, the majority of research models for tomato leaf disease recognition rely solely on traditional convolutional models or Transformer architectures and fail to capture both local and global features simultaneously. This limitation may result in biases in the model’s focus, consequently impacting the accuracy of disease recognition. Consequently, models capable of extracting local features while attending to global information have emerged as a novel research direction. To address these challenges, we propose an Eff-Swin model that integrates the enhanced features of the EfficientNetV2 and Swin Transformer networks, aiming to harness the local feature extraction capability of CNNs and the global modeling ability of Transformers. Comparative experiments demonstrate that the enhanced model has achieved a further increase in training accuracy, reaching an accuracy rate of 99.70% on the tomato leaf disease dataset, which is 0.49~3.68% higher than that of individual network models and 0.8~1.15% higher than that of existing state-of-the-art combined approaches. The results show that integrating attention mechanisms into convolutional models can significantly enhance the accuracy of tomato leaf disease recognition while also offering the great potential of the Eff-Swin backbone with self-attention in plant disease identification.

High spatial and temporal resolution normalized difference vegetation index (NDVI) time series are indispensable for monitoring land surfaces dynamics in spatiotemporally heterogeneous areas. Spatiotemporal fusion (STF) is one of the most common methods used for producing such data. These methods require the use of one or two pairs of fine images (with fine spatial but rough temporal resolution, such as Landsat images) and coarse images [with fine temporal but rough spatial resolution, such as Moderate Resolution Imaging Spectroradiometer (MODIS)]. A coarse image at the prediction date is also required to predict the corresponding missing fine image in the time series. Recently, the proposed deep learning (DL)-based STF methods have achieved promising fusion performance but are challenged in areas with frequent cloud contamination and landcover change prediction, while they also suffer from unstable fusion performance. Moreover, current STF methods lack a quality assessment process for the fusion results. To address these limitations, in this article, we propose a multitask DL framework for STF of NDVI time series, which integrates two types of DL-based STF methods. Four experiments in two study sites were conducted to test the effectiveness of the proposed method, and the results indicate that it achieves accurate and stable fusion capable of predicting landcover changes even when image pairs obtained at long intervals are used. In addition, a fusion uncertainty estimation method is proposed, which has the potential to be used as a quality assessment metric.

Soil moisture (SM) is an essential Earth surface and climate system variable. Insights into its persistence and corresponding time scales can improve numerical modeling and climate system prediction. In this study, SM from 17 observation stations within the Babao River Basin, Northwest China, between 1 June and 31 August 2014 were used to investigate persistence and corresponding time scales via adaptive fractal analysis. By conducting an adaptive fractal analysis of net radiation and by estimating the complementary cumulative distribution function of precipitation intervals, the relation between meteorological factors and the persistence and time scales of SM are determined and discussed. Results show that persistence and corresponding time scales of SM could be described using a three‐phase concept diagram. (1) At short time scales (approximately 0–14 hr [4 cm], 0–15 hr [10 cm], or 0–19 hr [20 cm]), the persistence of SM at most observation stations shows a strong long‐range correlation or nonstationarity. This phenomenon is essentially due to evaporation influenced by net radiation processes and the effects of rain. (2) At moderate time scales (approximately 14–159 hr [4 cm], 15–161 hr [10 cm], or 19–143 hr [20 cm]), the persistence of SM mostly exhibits a weak long‐range correlation or antipersistence due to net radiation process uncertainty and the probability of precipitation. (3) At long time scales (approximately greater than 159 hr [4 cm], 161 hr [10 cm], and 143 hr [20 cm]), the persistence of SM dynamics exhibits antipersistence because a high probability of precipitation reverses changes in the SM persistence.

Soil erosion inflicts multiple and severe damage throughout the world. The importance of vegetation spatial patterns in conserving soil and water has been widely acknowledged. In this study, by using the leakiness index (LI), which indicates the soil and water conservation function of the landscape by integrating landscape patterns closely with hydrological processes, we analyzed the changes in this function of riparian vegetation under different patterns with the aim of identifying the optimal pattern for improving soil and water conservation in severely eroded riparian buffer zones. Prior to this, the relationship between the erosion modulus and LI was discussed to provide certain evidence for the potential application of LI to the study area given the limited empirical works. Results showed that LI illustrated a significantly linear correlation with the erosion modulus (R² = 0.636, p < 0.01), thereby suggesting a promising application of LI in the Beijiang riparian vegetation buffer zone. A comparison of the LI values regarding four different vegetation patterns indicated that under the premise of the same coverage (40%), the aggregation degree and patch orientation with low LI values exerted improved performance for soil and water conservation, so we selected the horizontal distribution and compact aggregation as the optimal pattern for vegetation regulation. The spatial variations of LI values in the study area showed that five regions were suffering from severe erosion, thus becoming the targeted area for regulation. The final regulation with the optimal vegetation pattern in severely eroded areas performed well given that the soil and water conservation was improved to a high level with a LI value less than or equal to 0.2. The results described in this study provide an alternative screening method to figure out the severe erosion areas needing improvement, a further understanding of the effect of vegetation pattern on soil and water conservation and a theoretical basis for the extended application of LI.