For Pure Scots Pine Forests: Ground-Based Forest Parameters and a Dataset Comprising Landsat 8, Sentinel-2, Sentinel-1, and UAV-Based Features

The dataset consists of a multi-component structure that combines locally measured forest variables with spectral and spatial attributes derived from different remote sensing sources. The dataset includes measurements obtained from a total of 184 sample plots belonging to pure Scots pine (Pinus sylvestris L.) stands located in the Western Black Sea Region of Türkiye.

The ground-based dataset includes the variables stand volume (V), breast height area (BA), number of trees (N), dominant tree height (Hdom), and mean diameter at breast height (dq), which were measured or calculated for each sample plot. These variables constitute the dependent variables of the dataset as fundamental forestry parameters representing the structural and production characteristics of pure Scots pine stands.

The remote sensing component includes band values derived from Landsat 8 OLI and Sentinel-2 satellite images and texture features based on GLCM (Gray-Level Co-occurrence Matrix) derived from these bands. Each feature is defined by an abbreviation system consisting of four components. These components represent, respectively, (i) the calculated texture metric, (ii) the relevant satellite sensor and band number, (iii) the texture window size used, and (iv) the texture direction.

In this coding structure, the first component (b1–b8) represents the GLCM-based texture measures calculated for each band or index. Accordingly, b1 denotes the mean, b2 the variance, b3 the homogeneity, b4 the contrast, b5 the dissimilarity, b6 the entropy, b7 the second moment, and b8 the correlation measures.

The second component indicates which satellite sensor and band the relevant tissue or spectral feature was derived from. In Landsat 8 OLI images, this refers to bands L2, L3, L4, L5, L6, and L7; in Sentinel-2 images, it refers to bands S2, S3, S4, S5, S6, S7, S8, S8A, S11, and S12.

The third component represents the window size used in texture calculations. In this context, seven different window sizes were used: 3×3, 5×5, 7×7, 9×9, 11×11, 13×13, and 15×15.

The fourth component represents the texture direction used in GLCM calculations. Texture measures were calculated for four different directions: 0°, 45°, 90°, and 135°.

Finally, for Sentinel-1 data, only the band values and backscatter (dB) values for VV and VH polarizations were calculated; for UAV data, only the reflectance values and the vegetation indices derived from these values are included.

本数据集为多组分结构，融合了实地实测森林变量与源自多种遥感源的光谱、空间属性。数据集涵盖土耳其黑海西部地区纯苏格兰松（Pinus sylvestris L.）林分的184个样地的实测数据。
地面实测数据集包含每个样地的林分蓄积量（V）、胸高断面积（BA）、林木株数（N）、优势木平均高（Hdom）以及平均胸径（dq），这些变量均为表征纯苏格兰松林分结构与生产特性的基础林业参数，作为本数据集的因变量。
遥感组分包含由Landsat 8 OLI与Sentinel-2卫星影像提取的波段值，以及基于这些波段生成的灰度共生矩阵（Gray-Level Co-occurrence Matrix，GLCM）纹理特征。所有特征均采用四组分缩写系统命名，各组分依次代表：(i) 计算得到的纹理指标，(ii) 对应的卫星传感器与波段编号，(iii) 所用纹理窗口尺寸，(iv) 纹理计算方向。
在该编码结构中，第一组分（b1–b8）代表针对各波段或指数计算得到的基于GLCM的纹理指标，其中b1为均值、b2为方差、b3为同质性、b4为对比度、b5为相异性、b6为熵、b7为二阶矩、b8为相关性。
第二组分指明提取相关纹理或光谱特征所用的卫星传感器与波段。其中Landsat 8 OLI影像对应波段L2、L3、L4、L5、L6、L7；Sentinel-2影像对应波段S2、S3、S4、S5、S6、S7、S8、S8A、S11、S12。
第三组分代表纹理计算所用的窗口尺寸，本次研究共采用7种不同窗口尺寸：3×3、5×5、7×7、9×9、11×11、13×13及15×15。
第四组分代表GLCM计算所用的纹理方向，本次针对0°、45°、90°、135°四个方向计算了纹理指标。
此外，针对Sentinel-1数据，仅提取了VV与VH极化方式下的波段值及后向散射（dB）值；针对无人机（UAV）数据，仅包含反射率值及由其衍生的植被指数。