Description and classification of generic landforms: from words to concepts for digital use and decision support

在过去的十年里，地球测量学工程的重点已经从关注性能和精度的计算问题转移到决策问题，目标是能够根据可用的信息做出决策。软件应用程序应该在语义级检索和处理数据，以做出自己的决策或帮助用户做出明智的决策。为了推理，空间实体被认为是具有自己的空间和语义属性和关系的对象。*对地形描述几乎没有兴趣，地形描述是最常见的空间数据之一，但没有被建模为对象。大多数过程能够根据一些数学描述符(如坡度)对地形数据进行分类，但不能对地形进行分类。事实上，像山脉或山谷这样的地貌有主观和模糊的定义，这些定义因个人而异。提供可在计算系统中处理的正式定义是一项困难的任务。地形术语仍然太隐含，无法处理和处理地形数据分类。这项工作需要定义一个完整的框架，使地貌描述正式化：1)在概念级别，其中给出领域专家批准的抽象定义；2)在逻辑级别，将定义转换为面向对象的概念以用于实施；以及3)在通信级别，用于数据可视化和交换。*因此，本研究方案提出了一个概念框架，其中地貌可以从正式的概念模型定义为允许在数字地形模型(DTM)上进行地形识别的模型。该框架建立在这样一个原则上，即地貌不是由几何属性来表征，而是由构成地形的显著特征来表征。为这一五年计划确定了三个目标：*1.构思一个通用的概念模型，从中可以得出地貌概念；*2.将概念模型转化为一个逻辑模型，其中可以从数字地面模型中提取显著特征；*3.通过两个应用程序演示该方法的适用性：1)自动化海图制作的应用程序和2)用于一般用途的在线地貌提取的Web服务。*预期在概念和应用研究层面都有几个交付成果：*-提供基于显著度的地貌描述的概念模型的原型*-用于从DTM提取地貌的一套算法和结构*-用于海图生产的自动海底特征分类的原型系统*-用于地貌分析的Web服务的系统架构*这两个系统都将应用于加拿大沿海地区的地貌研究。预计该方案将有助于更好地了解加拿大海域的环境现象。

In this last decade, the focus in geomatics engineering has shifted from computational issues looking at performance and accuracy to decisional issues where the objective is to be able to make decisions from the information available. Software applications are supposed to retrieve and process data at a semantic level to make their own decisions or to assist a user in making an informed decision. In order to reason, spatial entities are perceived as objects with their own spatial and semantic properties and relationships.***Little interest has been devoted to terrain description, one of the most often found spatial data but which is not modelled as an object. Most processes are able to classify terrain data according to some mathematical descriptors such as the slope but they cannot classify landforms. Indeed, landforms such as a mountain or a valley have subjective and vague definitions which vary with individuals. Providing a formal definition that can be processed in a computing system is a difficult task. Landform terminologies remain too implicit to be handled and processed for terrain data classification. Such a work requires the definition of a whole framework which formalises landform description 1) at a conceptual level where an abstract definition approved by domain experts is given, 2) at a logical level where the definition is translated into object-oriented concepts for implementation, and 3), at a communication level for data visualisation and exchange.***This research programme therefore proposes a conceptual framework where landforms can be defined from a formal conceptual model to a model that allows landform recognition on a digital terrain model (DTM). The framework is built around the principle that landforms are not characterised by geometrical attributes but by salient features structuring the terrain. Three objectives are identified for this five-year programme:***1. To conceive a generic conceptual model from which landform concepts can be derived;***2. To translate the conceptual model into a logical model where salient features can be extracted from a DTM;***3. To demonstrate the applicability of the approach through two applications: 1) an application automating nautical chart production and 2) a Web service for online extraction of landforms for general purpose.***Several deliverables are anticipated at both conceptual and applied research levels:***- a prototype providing a conceptual model for landform description based on saliencies***- a set of algorithms and structures for landform extraction from a DTM***- a prototype system for automatic undersea feature classification in chart production***- a system architecture for a Web service for landform analysis***Both systems will be applied to geomorphological studies of coastal areas of Canada. It is expected that the programme will contribute to a better understanding of environmental phenomena in maritime areas of Canada.********