A knowledge-based approach to the statistical mapping of climate
A knowledge-based approach to the statistical mapping of climate
复制标题
DOI：
10.3354/cr022099
复制
发表时间：
2002-09-06
期刊：
影响因子：
1.1
通讯作者：
Pasteris, P
中科院分区：
文献类型：
--
作者：
Daly, C;Gibson, WP;Pasteris, P
近年来，对数字形式的空间气候数据的需求急剧增加。为了满足这一需要，已经使用了各种统计技术来促进与地理信息系统兼容的气候图的制作。然而，观测数据往往过于稀疏，缺乏代表性，无法直接支持创建真正代表当前知识状态的高质量气候图和数据集。一种有效的方法是利用有关气候空间格局及其与地理特征关系的丰富专家知识，即“地理空间气候学”，以帮助加强、控制和参数化统计技术。这里描述的是一个动态的基于知识的框架，它允许有效地积累、应用和改进气候知识，正如一个被称为PRISM（独立斜坡参数-海拔回归模型）的统计回归模型所表达的那样。最终目标是开发一个专家系统，能够重现知识渊博的气候学家用来绘制高质量气候图的过程，并具有一致性和可重复性的额外好处。然而，知识的积累和评估必须首先通过一个持续的模型应用过程；开发知识原型、参数和参数设置；测试;评估;和修改。本文描述了基于知识的气候制图框架的现状，并介绍了PRISM的具体算法，以演示如何应用和改进该框架以适应困难的气候制图情况。加权气候-海拔回归函数承认海拔对气候的主要影响。气候台站的权重考虑了除海拔以外的其他重要气候因素。从山坡到山脉的迎风和背风面，坡向和地形暴露在不同尺度上影响气候，通过将地形划分为地形面来模拟。沿海接近测量用于解释海岸线附近的急剧气候梯度。两层模式结构将大气划分为低层边界层和上层自由大气层，可以模拟温度逆温和中坡降水最大值。本文还估计了不同地形配置对地形增雨的影响。给出了气候制图实例。
The demand for spatial climate data in digital form has risen dramatically in recent years. In response to this need, a variety of statistical techniques have been used to facilitate the production of GIS-compatible climate maps. However, observational data are often too sparse and unrepresentative to directly support the creation of high-quality climate maps and data sets that truly represent the current state of knowledge, An effective approach is to use the wealth of expert knowledge on the spatial patterns of climate and their relationships with geographic features, termed 'geospatial climatology', to help enhance, control, and parameterize a statistical technique. Described here is a dynamic knowledge-based framework that allows for the effective accumulation, application, and refinement of climatic knowledge, as expressed in a statistical regression model known as PRISM (parameter-elevation regressions on independent slopes model). The ultimate goal is to develop an expert system capable of reproducing the process a knowledgeable climatologist would use to create high-quality climate maps, with the added benefits of consistency and repeatability. However, knowledge must first be accumulated and evaluated through an ongoing process of model application; development of knowledge prototypes, parameters and parameter settings; testing; evaluation; and modification. This paper describes the current state of a knowledge-based framework for climate mapping and presents specific algorithms from PRISM to demonstrate how this framework is applied and refined to accommodate difficult climate mapping situations. A weighted climate-elevation regression function acknowledges the dominant influence of elevation on climate. Climate stations are assigned weights that account for other climatically important factors besides elevation. Aspect and topographic exposure, which affect climate at a variety of scales, from hill slope to windward and leeward sides of mountain ranges, are simulated by dividing the terrain into topographic facets. A coastal proximity measure is used to account for sharp climatic gradients near coastlines. A 2-layer model structure divides the atmosphere into a lower boundary layer and an upper free atmosphere layer, allowing the simulation of temperature inversions, as well as mid-slope precipitation maxima. The effectiveness of various terrain configurations at producing orographic precipitation enhancement is also estimated. Climate mapping examples are presented.