A wall–to–wall comparison of MapBiomas and JRC–Tropical Moist Forests across Amazonian biomes

Overview This repository contains the replication data, spatial layers, and analysis code for a spatial inter-comparison study evaluating thematic agreements and disagreements between the MapBiomas and JRC Tropical Moist Forest (JRC-TMF) products. The analysis focuses broadly on the Amazon Basin, with a high-resolution, localized case study centered on Parque do Xingu along the Arc of Deforestation. Repository Structure & File Descriptions 1. Processing & Analysis Code GoogleEarthEngineCode.txt (11.11 KB) Contains the JavaScript pipeline used in Google Earth Engine (GEE) for cloud-based data extraction, masking, intersection, and initial preprocessing of the raw MapBiomas and JRC-TMF assets. version08_06.R The primary R script used for post-processing, spatial filtering (e.g., noise reduction via terra::classify), area calculations, and generating final publication-quality multi-panel grid maps using ggplot2, tidyterra, and patchwork. 2. Spatial Data Layers (Raster/GeoTIFF) Amazon_Disagreement_Density_0p02deg.tif (6.06 MB) A spatially aggregated, basin-wide raster at a coarser resolution of 0.02° × 0.02°. Cells represent the density percentage of structural/thematic classification contradiction between the two products across the entire Amazon Basin. Xingu_Incoherence_Masks_30m.tif (1.24 MB) The 30-meter resolution spatial incoherence mask for the Parque do Xingu study area, explicitly isolating pixels where major map contradictions occur. Xingu_MapBiomas_Level1_30m.tif (688.42 KB) The baseline MapBiomas Level 1 classification layer at native 30-meter resolution, cropped to the bounding perimeter of the Parque do Xingu study region. Xingu_TMF_Defor_And_Other_30m.tif (951.89 KB) The JRC-TMF tracking layer showing 'Deforested Land' and 'Other Land Cover' categories at native 30-meter resolution, cropped to the Parque do Xingu study region. 3. Tabular Data TMF_MapBiomas_Raw_Area_m2.csv (3.93 KB) A raw data matrix containing pixel counts and area calculations (measured in square meters) quantifying cross-tabulated intersections between overlapping MapBiomas and JRC-TMF classification categories. Methodological Summary Data Extraction: Raw 30 m satellite-derived maps were filtered and paired chronologically using GoogleEarthEngineCode.txt. Aggregation & Masking: Regional structural contradictions were mapped at 30 m over Xingu, while a broader regional pattern map was created by aggregating data to a 0.02° grid cell resolution to display localized disagreement densities. Statistical Analysis & Mapping: The script version08_06.R filters out trace noise below a 0.5% threshold to clearly isolate true structural contradiction hotspots for final figures.数据集概述 本数据集仓库包含一项空间互比对研究的复现数据、空间图层与分析代码，该研究旨在评估MapBiomas与JRC热带湿润森林（JRC Tropical Moist Forest, JRC-TMF）产品间的专题一致性与不一致性。本研究整体以亚马孙流域为核心分析范围，同时围绕位于毁林弧带的欣古公园（Parque do Xingu）开展高分辨率本地化案例研究。 仓库结构与文件说明 1. 处理与分析代码 GoogleEarthEngineCode.txt（11.11 KB）：包含基于谷歌地球引擎（Google Earth Engine, GEE）的云端数据提取、掩膜、交集运算及原始MapBiomas与JRC-TMF数据源初始预处理所用的JavaScript流水线脚本。 version08_06.R：本研究的核心R脚本，用于完成后处理、空间滤波（例如通过terra::classify实现降噪）、面积计算，并借助ggplot2、tidyterra与patchwork包生成符合出版标准的多面板网格地图。 2. 空间数据图层（栅格/GeoTIFF格式） Amazon_Disagreement_Density_0p02deg.tif（6.06 MB）：空间聚合后的全流域栅格数据，分辨率为较粗的0.02°×0.02°，每个栅格单元代表整个亚马孙流域内两类产品间结构/专题分类不一致的密度百分比。 Xingu_Incoherence_Masks_30m.tif（1.24 MB）：欣古公园研究区的30米分辨率空间不一致掩膜图层，用于精准分离两类地图存在显著分类矛盾的像素单元。 Xingu_MapBiomas_Level1_30m.tif（688.42 KB）：原生分辨率为30米的基准MapBiomas一级分类图层，已裁剪至欣古公园研究区的边界范围。 Xingu_TMF_Defor_And_Other_30m.tif（951.89 KB）：JRC-TMF追踪图层，以30米原生分辨率展示“毁林地”与“其他土地覆盖”类别，已裁剪至欣古公园研究区范围。 3. 表格数据 TMF_MapBiomas_Raw_Area_m2.csv（3.93 KB）：原始数据矩阵，包含像素计数与面积计算结果（单位为平方米），用于量化两类产品重叠分类类别间的交叉制表交集情况。 方法学概述 数据提取：借助GoogleEarthEngineCode.txt脚本，对原始30米分辨率卫星衍生地图进行滤波处理，并按时间顺序完成配对。 聚合与掩膜：针对欣古研究区，以30米分辨率绘制区域结构分类矛盾图；同时通过将数据聚合至0.02°栅格单元分辨率，生成更宏观的区域格局图，以展示局部不一致的密度分布。 统计分析与制图：version08_06.R脚本会滤除0.5%阈值以下的微量噪声，以精准分离真实的结构分类矛盾热点区域，用于最终制图。