公告：凡是在
2025 年 4 月 15 日前註冊使用 Earth Engine 的非商業專案，都必須
驗證非商業用途資格，才能繼續存取 Earth Engine。
Concepts
透過集合功能整理內容
你可以依據偏好儲存及分類內容。
這個 API 提供簡單一致的介面，可存取各種地理空間圖像資料。地理空間是指與地表位置相關聯的資料。圖片資料是指以像素值統一格線結構呈現的資料，與傳統圖片非常相似。
這類資料最常見的例子是衛星圖像，與使用者在 Google 地圖和地球上看到的圖像非常相似。不過，許多其他資料集具有相同結構，包括網格天氣和氣候資料集、地形和土地覆蓋資料集，以及人口密度資料集。
驗證
Earth Engine API 使用 OAuth 2.0 通訊協定進行驗證及授權。您可以使用 OAuth 2.0，以各種語言發出 API 呼叫，十分方便。如果您先前未使用 OAuth 2.0 存取 Google API，建議先詳閱「使用 OAuth 2.0 存取 Google API」一文。
這個 API 使用 Earth Engine OAuth 2.0 範圍 https://www.googleapis.com/auth/earthengine.readonly。如果您自行設定服務帳戶憑證，則必須明確要求該範圍。如果您使用 Compute Engine 虛擬機器執行個體中提供的預設應用程式憑證，則需要設定 VM 執行個體來要求該範圍。
素材資源
Earth Engine 資料模型的核心是類似檔案系統的資產結構。資產主要有三種，詳情請見下文：地理空間光柵資料本身會儲存在「圖片」中，圖片可以分組為大型「集合」，圖片和集合則會進一步整理成「資料夾」階層。
舉例來說，以下路徑說明特定 Landsat 影像：
LANDSAT/LC8_L1T/LC81180562013193LGN00
在這個範例中，LANDSAT 是包含所有 Landsat 資料的頂層資料夾，LC8_L1T 則是用於識別特定校正和地形校正 Landsat 8 影像的集合，而 LC81180562013193LGN00 則是用於識別該集合中的特定影像。素材資源路徑可包含英文字母、數字、底線和連字號，並以正斜線分隔。
每張圖片都有中繼資料，方便您找出感興趣的資料。這些圖片專屬中繼資料屬性通常包括：
- 圖片的足跡幾何，例如衛星圖像或地形資料集的空間範圍。
- 圖片的時間戳記，例如衛星圖片的擷取時間，或是氣候模型資料集中的日期或月份。
- 任意鍵/值屬性，例如 Landsat 衛星圖像的預估雲量百分比。
像素和地圖投影
在 Earth Engine 中，每張圖片都包含一或多個資料「波段」。這些波段可能對應到一般 RGB 衛星圖像的紅、綠和藍色波段，也可能對應到完全不同的事物，例如天氣資料集中的溫度和降水量。每個頻帶中的像素都有特定資料類型，例如 uint8 或 float32。
像素格線中的像素會透過稱為「地圖投影」的數學函式，對應到地球表面的點。這類關係通常包含兩個部分。首先，座標參考系統 (CRS) 會定義 2D 座標，代表地球曲面上的點。不同的座標參考系統具有不同的屬性，因此適合在不同應用程式中使用不同類型的資料。座標參考系統通常會使用標準 ID 代碼識別，常見的例子包括等距圓柱投影、Web Mercator 和 UTM 座標。
然後，像素座標通常會透過仿射轉換與空間座標參考系統相關聯，藉此控制像素格線的實體比例和原點。在 Earth Engine 中，我們會為每張圖片儲存金字塔資料：金字塔的底層包含原始資料，解析度與原始資料相同，金字塔的較高層則儲存解析度較低的概覽資料。金字塔的這些層級是由仿射轉換所描述，其比例相差連續的兩個因素。
集合和資料夾
圖片通常會以大型集合的形式提供，例如特定衛星取得的所有圖片集合，或是全球溫度估計值隨時間變化的集合。Earth Engine 會建立中繼資料索引，讓您根據中繼資料篩選或查詢集合中的圖片，包括依位置和時間篩選。通常，單一集合中的所有圖片都具有相同結構，也就是相同的波段和中繼資料屬性。
圖片和集合本身會依資料夾階層排列。資料夾與傳統檔案系統中的一般資料夾或目錄非常相似，是其他資產 (即圖片、集合和其他資料夾) 的簡單容器。並非所有圖片都包含在集合中，部分資料集 (例如特定全球地形模型) 可能只包含單一圖片。無論如何，每個資產都會由特定路徑唯一識別，該路徑會指出資產在 Earth Engine 資料目錄中的位置。
除非另有註明，否則本頁面中的內容是採用創用 CC 姓名標示 4.0 授權，程式碼範例則為阿帕契 2.0 授權。詳情請參閱《Google Developers 網站政策》。Java 是 Oracle 和/或其關聯企業的註冊商標。
上次更新時間：2025-07-26 (世界標準時間)。
[null,null,["上次更新時間：2025-07-26 (世界標準時間)。"],[[["\u003cp\u003eThis API provides access to a wide variety of geospatial image data, including satellite imagery and gridded datasets like weather, climate, terrain, and population density.\u003c/p\u003e\n"],["\u003cp\u003eEarth Engine uses OAuth 2.0 for authentication, requiring the \u003ccode\u003ehttps://www.googleapis.com/auth/earthengine.readonly\u003c/code\u003e scope.\u003c/p\u003e\n"],["\u003cp\u003eEarth Engine data is organized as assets, including images, collections, and folders, with images containing bands of pixels and metadata.\u003c/p\u003e\n"],["\u003cp\u003eImages in Earth Engine are georeferenced using map projections and coordinate reference systems for accurate location representation.\u003c/p\u003e\n"],["\u003cp\u003eCollections allow filtering and querying of images based on metadata, facilitating efficient data discovery and analysis.\u003c/p\u003e\n"]]],["This API offers access to geospatial image data, structured as pixel grids, through a filesystem-like asset system. Access requires OAuth 2.0 authentication with a specific scope. Data is organized into **images**, **collections**, and **folders**. Each image has metadata, including footprint geometry and timestamps. Images have bands of data with defined pixel data types, linked to Earth's surface via a map projection. Images can be filtered in collections using metadata. A pyramid of data is also available.\n"],null,["# Concepts\n\nThis API provides simple, uniform access to a wide variety geospatial image data. By *geospatial* we mean that the data is associated with locations on the Earth's surface. By *image data* we mean data that is structured as a uniform grid of pixel values, much like a traditional image.\n\nThe most familiar example of this type of data is satellite imagery, much like users are accustomed to seeing in Google Maps and Earth. However, many other datasets have the same structure, including for example gridded weather and climate datasets, terrain and land cover datasets, and population density datasets.\n\nAuthentication\n--------------\n\nThe Earth Engine API uses the **OAuth 2.0** protocol for authentication and authorization. There are convenient libraries for using OAuth 2.0 to issue API calls in a wide range of languages. If you have not used OAuth 2.0 to access Google APIs before, you may wish to familiarize yourself with the documentation on [Using OAuth 2.0 to Access Google APIs](https://developers.google.com/identity/protocols/OAuth2).\n\nThis API uses the Earth Engine OAuth 2.0 **scope** `https://www.googleapis.com/auth/earthengine.readonly`. If you are configuring service account credentials yourself then you will need to request that scope explicitly. If you are using the default application credentials available in Compute Engine virtual machine instances, you will need to configure your VM instance to request that scope.\n\nAssets\n------\n\nAt the heart of the Earth Engine data model is a filesystem-like structure of **assets** . There are three primary types of assets, described in more detail below: geospatial raster data itself is stored in **images** , images can be grouped into large **collections** , and images and collections are further organized in a hierarchy of **folders**.\n\nFor example, the following path describes a particular Landsat image: \n\n LANDSAT/LC8_L1T/LC81180562013193LGN00\n\nIn this example, `LANDSAT` is a top-level folder that contains all the Landsat data, `LC8_L1T` identifies a particular collection of calibrated and terrain-corrected Landsat 8 images, and `LC81180562013193LGN00` identifies a particular image within that collection. Asset paths may contain letters, numbers, underscores, and hyphens, separated by forward slashes.\n\nEach image has metadata that makes it easy to identify data of interest. These per-image metadata properties typically include:\n\n- The footprint geometry of the image, such as the spatial extent of a satellite image or terrain dataset.\n- The timestamp of the image, such as the satellite image acquisition time, or the day or month in a climate model dataset.\n- Arbitrary key/value properties, such as the estimated cloud cover percentage of a Landsat satellite image.\n\nPixels and Map Projections\n--------------------------\n\nIn Earth Engine each image consists of one or more **bands** of data. These might correspond to the red, green, and blue bands of an ordinary RGB satellite image, or they might correspond to something completely different, such as temperature and precipitation in a weather dataset. The **pixels** in each band have a particular data type, e.g. `uint8` or `float32`.\n\nPixels in a pixel grid correspond to points on the Earth's surface via a mathematical function known as a **map projection** . This relationship typically has two components. First a **coordinate reference system** (CRS) defines 2D coordinates that represent points on the curved surface of the Earth. Different coordinate reference systems have different properties that make them appropriate to use with different types of data in different applications. Coordinate reference systems are typically identified using [standard identifier codes](http://epsg.io/); common examples are [equirectangular](https://en.wikipedia.org/wiki/Equirectangular_projection), [web Mercator](https://en.wikipedia.org/wiki/Web_Mercator), and [UTM](https://en.wikipedia.org/wiki/Universal_Transverse_Mercator_coordinate_system) coordinates.\n\nPixel coordinates are then typically related to the spatial coordinate reference system via an **affine transformation** that controls the physical scale and origin of the pixel grid. In Earth Engine we store a **pyramid** of data for each image: the base level of the pyramid contains the original data in its native resolution, and higher levels of the pyramid store reduced-resolution overview data. These levels of the pyramid are described by affine transformations whose scales differ by successive factors of two.\n\nCollections and Folders\n-----------------------\n\nImages often come in large **collections** of data, such as a collection of all images acquired by a particular satellite, or a collection of global temperature estimates over time. Earth Engine builds metadata indexes that allow you to **filter** or **query** the images in collections based on their metadata, including filtering by location and time. Typically, all images within a single collection have the same structure, i.e. the same bands and the same metadata properties.\n\nImages and collections are themselves organized into a hierarchy of **folders**. A folder is much like an ordinary folder or directory in a traditional filesystem: it is a simple container for other assets, i.e. for images, collections, and other folders. Not all images are contained in collections: some datasets, such as a particular global terrain model, may consist of only a single image. Regardless, every asset is uniquely identified by a particular path that identifies its location in the Earth Engine data catalog."]]
