Willows are widely distributed in the northern hemisphere and have good adaptability to different living environment. The increasing of genome and transcriptome data provides a chance for comparative analysis to study the evolution patterns with the different origin and geographical distributions in the Salix phylogeny. Transcript sequences of 10 Salicaceae species were downloaded from public databases. All pairwise of orthologues were identified by comparative analysis in these species, from which we constructed a phylogenetic tree and estimated the rate of diverse. Divergence times were estimated in the 10 Salicaceae using comparative transcriptomic analysis. All of the fast-evolving positive selection sequences were identified, and some cold-, drought-, light-, universal-, and heat- resistance genes were discovered. The divergence time of subgenus Vetrix and Salix was about 17.6–16.0 Mya during the period of Middle Miocene Climate Transition (21–14 Mya). Subgenus Vetrix diverged to migratory and resident groups when the climate changed to the cool and dry trend by 14 Mya. Cold- and light- stress genes were involved in positive selection among the resident Vetrix, and which would help them to adapt the cooling stage. Universal- stress genes exhibited positive selection among the migratory group and subgenus Salix. These data are useful for comprehending the adaptive evolution and speciation in the Salix lineage.

Tree height is an important parameter for calculating forest carbon sink and assessing forest carbon cycle. In order to obtain forest tree height over a large area both efficiently and at a low cost, this study proposed an Interferometric Synthetic Aperture Radar (InSAR) combined with a machine learning method to estimate the tree canopy height. The forest height in the study area was obtained using Unmanned Aerial Vehicle (UAV) photogrammetry, which was considered to be the true canopy height. Two machine learning methods (Random Forest, Multi-layer perceptron) were used to establish the relationship between phase center height calculated by InSAR DEM differential interference method and coherent amplitude method with true canopy height. The topographic factor, backward scattering coefficient and coherence coefficient were introduced into the relationship model. It was found that the accuracy of tree height estimation using random forest and two InSAR methods can reach 0.95 and 0.94. The root-mean-square error was 1.76 m, 1.86 m, respectively. The accuracy of tree height estimation using multi-layer perceptron and two InSAR methods was 0.25 and 0.2. The root-mean-square error was 3.96 m and 4.13 m. The results indicated that the combination of InSAR and machine learning can estimate canopy height efficiently and at a low cost. Moreover, the integrated learning algorithm random forest demonstrated better stability and higher accuracy than the single learning algorithm multi-layer perceptron.

Implementing forest protection measures such as afforestation, can be an effective approach towards slowing down the increase of C02 concentration and attaining carbon neutrality. The estimation of forest parameters is of great significance in understanding regional and global climate cbange pattems, aod the forest leaf area lndex(LAI) is a crucial parameter. Current LAI products are mosdy generated by moderate-resolution remote sensing data which doesn't meet the precision requirements for mountain forest ecosystems. To overcome this issue, there is an urgent need for higber resolution LAI data. This paper proposes a data fusion method to map LAI in Wolong Nature Reserve that utilizes Sentinel-2 reftectance data, solar sensor geometry parameters, and vegetation indices(VI) extracted from the Google Earth Engine (GEE) platform, along with canopy beigbt data derived from canopy beigbt estimation models in previous studies, combined with GLASS LAI V6 to estimate LAI using the random forest algorithm. Tbe resulting LAI distribution map was plotted at a resolution of 20 m. The study demonstrated that incorporating canopy heights Into the estimation modelled to an R2 model accnracy of greater thao 0.83. Tbe 20-m resolution LAI map increased spatial details compared to the moderateresolution LAI map, making it more suitable for mountain forrst ecosystems that exhibit significant spatial heterogeneity.

Forests constitute a fundamental component of terrestrial carbon stocks and play a pivotal role in mitigating climate change through carbon sequestration. Accurate estimation of aboveground biomass (AGB) is essential for quantifying carbon budgets and informing ecosystem models. This study takes Wolong Nature Reserve in Sichuan Province, China, a mountainous area with high vegetation coverage and diverse forest types dominated by coniferous and mixed forests, as the study area, and constructs and evaluates AGB estimation models by integrating canopy height, leaf area index (LAI), vegetation indices (VIs), and topographic variables. Initially, univariate parametric models (linear, exponential, logarithmic, power, and polynomial) were established to relate canopy height to field-measured AGB. Subsequently, multivariate regression models incorporating VIs, LAI, and topographic metrics were developed. Finally, a decision tree-based machine learning framework was implemented to exploit the combined predictor set. Comparative analysis revealed that both canopy height-based and conventional multivariate regression models tended to overestimate AGB, limiting their applicability for large-scale assessments. In contrast, the optimized decision tree model, following parameter tuning and cross-validation, achieved superior predictive accuracy.

This study investigates the application of Spiking Neural Network (SNN) in seismic signal denoising by developing a Convolutional Neural Network (CNN) to SNN conversion framework. We focus on two challenges: optimal spike encoding strategy adaptation for seismic data; and denoising performance preservation during CNN-SNN conversion. Through systematic experiments on the public Marmousi 2 dataset and field data from Sichuan Basin, we demonstrate that SNN can feasibly serve as an alternative to traditional CNN for seismic denoising tasks. Our experiments show that both count-rate encoding (CR) and time-to-first-spike coding (TTFS) can effectively encode seismic data. However, when integrated with the proposed CNN-SNN conversion framework for seismic denoising, CR outperforms TTFS due to structural mismatches between TTFS and the framework in signal representation, temporal control, and weight mapping, which lead to information loss. Our neuron optimization strategy combines soft reset mechanisms with adaptive thresholding, demonstrating enhanced performance that narrows the gap between native CNN and converted SNN implementations. To enable rigorous evaluation, we introduce a baseline alignment scheme ensuring fair comparison between the native CNN and its CNN-SNN converted architecture. This work demonstrates the first successful application of SNN to seismic signal denoising, offering a bio-inspired alternative to conventional CNN while preserving comparable signal-to-noise ratio performance.