OBIA4RTM Demonstration Data

This dataset provides sample data demonstrating the capacities of the OBIA4RTM tool. OBIA4RTM combines radiative transfer modelling (RTM) of vegetation with object-based image analysis (OBIA). Its main purpose is to provide vegetation parameters such as Leaf Area Index (LAI) or leaf Chlorophyll a+b content (CAB) on a per-object rather than per pixel base. In this dataset, the OBIA4RTM tool was applied to two Sentinel-2 scenes covering an agricultural area in Southern Germany. Field parcels were used as image objects that were delineated from high-resolution ortho-photography and classified into vegetated and non-vegetated parcels using a Support Vector Machine trained on manually selected samples. For each of the two scenes - dating back on the 6th and 18th of July 2017 - the canopy RTM ProSAIL was run in forward mode and the synthetic spectra stored in a Lookup-Table (LUT). For parameter retrieval, the 5 closest matches between spectra in the LUT and a given observed satellite spectrum averaged per parcel were used. Matches were found in terms of the lowest Root Mean Squared Error (RMSE). The utilized vegetation parameterisation is provided additionally. The results include the Leaf Area Index (LAI), the Chlorophyll a+b content (CAB) of leaves and the fraction of brown leaves (Cbrown). In addition, the retrieval error in terms of RMSE is provided together with the average of the 5 best matching synthetic spectra in the LUT to a given object-based spectrum. This allows for evaluating the quality of the inversion results and enables user to further improve the results by applying a more appropiate vegetation parameterisation. The structure of the dataset (see below) is straightforward: - The "Field Parcels" folder contains an ESRI shapefile with the field parcels as well as the classification results for the two image acquisition dates - The "ProSAIL Parametersisation" directory provides the vegetation parameters used to run the ProSAIL model. - The actual results are stored as ESRI-shapefiles in "Retrieved Vegetation Parameters" folder containing the LAI, CAB, Fraction of brown leaves and the RMSE as well as inverted Sentinel-2 spectra - "Sentinel-2 data" contains the utilized Sentinel-2 data as GeoTiff clipped to the study area in Level-2A This information should allow for reproducing the results using the freely available base version of OBIA4RTM (for research and education) or within other software packages. All geodata is projected in UTM-Zone 32N, WGS-84.本数据集提供示例数据，用于展示OBIA4RTM工具的功能。OBIA4RTM将植被辐射传输模型（Radiative Transfer Modelling, RTM）与面向对象影像分析（Object-Based Image Analysis, OBIA）相结合，其核心目标是面向对象而非单个像素，获取叶面积指数（Leaf Area Index, LAI）、叶片叶绿素a+b含量（Chlorophyll a+b content, CAB）等植被参数。 本数据集将OBIA4RTM工具应用于覆盖德国南部农业区域的两景Sentinel-2影像。研究中以从高分辨率正射影像中勾勒出的农田地块作为影像对象，并通过基于手动选取样本训练的支持向量机（Support Vector Machine, SVM）将地块划分为植被覆盖与非植被覆盖两类。 两景影像分别采集于2017年7月6日与7月18日，研究中以冠层RTM模型ProSAIL的正向运行模式生成查找表（Lookup-Table, LUT）存储的合成光谱。在参数反演阶段，针对每个地块的平均观测卫星光谱，选取查找表中光谱相似度最高的5条合成光谱，以最低均方根误差（Root Mean Squared Error, RMSE）作为匹配准则。 本数据集同时提供所用的植被参数化方案。反演结果包含叶面积指数（LAI）、叶片叶绿素a+b含量（CAB）、枯叶占比（Cbrown），同时还提供了反演的均方根误差（RMSE），以及每个对象光谱对应的查找表中5条最佳匹配合成光谱的平均值，以此可评估反演结果的质量，并支持用户通过采用更适配的植被参数化方案进一步优化反演结果。 本数据集结构清晰，具体如下： - "Field Parcels（农田地块）"文件夹：包含存储农田地块信息的ESRI形状文件，以及两景影像采集日期对应的分类结果。 - "ProSAIL Parametersisation（ProSAIL参数化方案）"目录：提供运行ProSAIL模型所用的植被参数。 - "Retrieved Vegetation Parameters（反演植被参数）"文件夹：以ESRI形状文件存储实际反演结果，包含LAI、CAB、枯叶占比、RMSE以及反演得到的Sentinel-2光谱。 - "Sentinel-2 data（Sentinel-2数据）"：包含研究所用的Sentinel-2 Level-2A级数据，已裁剪至研究区域，格式为GeoTIFF。 所有地理数据均采用WGS-84坐标系下的UTM Zone 32N投影。本数据集可通过OBIA4RTM的免费开源基础版本（适用于科研与教育用途）或其他软件包复现研究结果。