Emergency weather network

三个独立的设备系统被要求支持加拿大西部山区的天气预报研究。 (1)一个是计算机服务器（4个处理器，3 GHz，1 TB 10磁盘RAID阵列，16 GB内存），它将取代一台过时的、故障的计算机，该计算机拥有一个名为“紧急天气网”的天气数据库。 该服务器需要可靠地存储从大约3000个气象站连续接收的每小时现场数据的高达TB的在线数据库。 对于每一次观测，它还存储了来自16个不同数值天气预报（NWP）模型（求解大气流体动力学方程的计算机代码）的成对预报。 它还存储了站点、观测和预报的相同组合的校准系数和验证统计数据。 (2)第二个系统是一个实验性的计算机硬件，它包括2个现场可编程门阵列（FPGA）沿着4个传统的计算机CPU。 FPGA是一种特殊的计算机芯片，可以通过硬件配置以极快的速度执行重复计算。 这项研究的目的是实现更快的数值计算，从而以更高的分辨率进行更准确和及时的预报，更好地近似云、湍流和山脉效应等物理过程。 (3)第三个系统是一套10个廉价的商品气象站，将部署在大温哥华地区。 通过我们以前的NWP模型运行，我们发现了许多小尺度天气模式，这些模式会导致短距离内天气的极端变化（危险的雾，降雨，污染物输送，强风）。 我们希望通过这个200万人口的人口中心的地面气象站网络提供的实地数据来证实和更好地预测这些天气模式。

Three independent equipment systems are requested to support weather-forecast research for mountainous western Canada.      (1) One is a computer server (4 processors at 3 GHz, a 1 TB 10-disk RAID array, 16 GB of memory), which will replace an obsolete, failing computer that holds a weather database called Emergency Weather Net.  This server needs to reliably store up to a terabyte of online databases of hourly field data received continuously from about 3000 weather stations.  For each observation it also stores paired forecasts from 16 different numerical weather prediction (NWP) models (computer codes that solve the fluid-dynamics equations for the atmosphere).  It also stores calibration coefficients and verification statistics for the same combination of sites, observations, and forecasts.      (2) The second system is an experimental piece of computer hardware that includes 2 Field Programmable Gate Arrays (FPGAs) along with 4 conventional computer CPUs.  FPGAs are special computer chips that can be hardware configured to perform repeated calculations exceptionally fast.  The point of this research is to enable faster numerical computation, thereby allowing more accurate and timely forecasts at finer resolution with better approximations to physical processes such as clouds, turbulence, and mountain effects.      (3) The third system is a set of 10 inexpensive, commodity weather stations, to be deployed in the Greater Vancouver Area.  Via our previous NWP model runs we discovered many small-scale weather patterns that cause extreme variations in weather (hazardous fog, rainfall, pollutant transport, strong winds) across short distances.  We hope to confirm and better predict these weather patterns with field data from a more closely-spaced network of surface weather stations in this population centre of 2 million people.