Skip to content
RichardWagener edited this page Jul 8, 2020 · 3 revisions

BNL Solar Nowcasting System

Overview

In order to predict Solar Power with a lead time of 30 min, a network of pairs of hemispheric sky-cameras are installed in the vicinity of Solar PV installations. Images are collected every 30s. Then corrected for orientation and geometric distortion. Triangulation of cloud edges as seen from the two cameras in each pair gives cloud heights limited to two distinct cloud layers. Tracking of cloud edges in successive images gives wind speed and direction in those two cloud layers and from that the ability to predict what areas will be shaded in the near term future.

Installation

Pre-requisites

The code is written in python and requires at least version 3.5. It requires the following packages which were installed with the anaconda3 package manager conda .

Code install

Plan to put these into a package install script

Separation of Privileges

::
groupadd nowcast useradd nowcast

Set up developer (with github id) with sudo privileges to run commands as user nowcast.

::
umask 002 sudo -u nowcast mkdir -p ~nowcast/code sudo chmod g+ws ~nowcast/code cd ~nowcast/code git clone https://github.com/BNL-NowCasting/SolarForecasting # or if git already setup, to get latest updates git pull

Configuration

::
cd ~nowcast/run cp ~nowcast/code/SolarForecasting/downloader/image_downloader.conf.dist image_downloader_{SITE-CODE}.conf

Edit conf file to update directory paths. Make sure destination directories (cachepath, latest, imagepath, and logpath, etc., see :doc:`DataTree`) in conf file are writable by user nowcast.

Processing Pipeline

Image Collection

image_downloader.py