USE OF OPEN DATA CUBE WITHIN AN OPERATIONAL

USE OF OPEN DATA CUBE WITHIN AN OPERATIONAL SERVICE TO GOVERNMENT AND INDUSTRY IN UGANDA Earth Observation Systems for Sustainable Development Goals Simon Reid - Chief Technical Officer, RHEA Group 11 April 2018 1

engineering the world with you 25 YEARS IN SPACE DATA SYSTEMS AND CYBER SECURITY. BESPOKE ENGINEERING SOLUTIONS THAT APPLY TO OUR EVERYDAY LIVES. RHEA Group © 2

THE DFMS CONSORTIUM Funded by Project leader Ugandan partners Specialist organisations Uganda Data Cube Overview 3

DATA FOR RELIABLE EARLY WARNINGS OF ADVERSE ENVIRONMENTAL CONDITIONS Gathering data Satellite Data Hydrological & Meteorological Soil Moisture & Maps Crop Type and Condition Land Cover … A comprehensive forecast model Derived indices and weather forecasts, combined with advanced models to predict future conditions over a range of timeframes Sharing information with decision makers Data are made easily accessible for decision makers to assess the potential impact. Farmers and others mitigate the impact of anticipated flood and drought conditions

A NATIONAL DATA SHARING PROJECT Ministry of Water & Environment National Meteo Authority Ministry of Agriculture National Agriculture Research Disaster Risk Information & Emergency Coordination District Users Commercial Users

Open Source system to: • Store large amounts of EO data; • Data in multi dimensional arrays (latitude, longitude, time …). • Python based API for high performance querying and data access; • Easy Exploratory Data Analysis (especially over a time series); • Allow scalable continent scale processing of the stored data; • Track the provenance of all the contained data to allow for quality control and updates TIME Uganda Data Cube Overview • RHEA is extending its use and contributing to the public repository 6

ANALYSIS READY DATA Loca lised (not globa l) pro duct s Sentinel-1, 2, 3 Landsat 8 MODIS SMOS … Land Cover Soil Moisture Surface Temperature Vegetation and Leaf Indices Water Height Water Extent …. NDVI Uganda Data Cube Overview

DFMS FORECAST PRODUCTS Weather forecast Wh Environmental forecast Precipitation, Temperature, en w Pressure etc Soil Moisture Evapotranspiration Runoff Base flow Drought Index Flood Index ill t he variables in ( + other rain multiple dimensions ) s star DAILY WEEKLY t, in SEASONAL Frequency of update 12 hourly 6 hourly 7 days Timeframe 0 -36 hours ~4 km 0 to months ~60 km Spatial resolution Uganda Data Cube Overview 0 -7 days 20 km my par ish 12 hourly 6 0 - 36 hours , 6 fhourly or m y 0 -7 days ~4 km or 20 km or Parish Sub District 7 days cr 0 o- 6 months p? ~60 km or District 8

SYSTEM ARCHITECTURE Interface Services Platform Data Website Weather Forecast Data Flow Hydrology Forecast Cloud Resource Manager Application Soil and Vegetation Error Manager Ground Data Service Engine System Configuration and Other Data API Analytics Security Engine History EO Infrastructure and Service Manager System Logs 9

Platform – process flows and resource management • Manages processing lifecycle • Dynamically creates / allocates cloud resources on demand initiates processing as required • Apache Nifi supports the dependency graph & user interface • Error handling 10

MULTIPLE DATA CUBES Use cases and design • Adjacent country instances collaborating • Satellite/Instrument providers providing ARD • Regional Cache of ARD • Remote / bad / intermittent network connections • On-demand or specific cubes per domain interest • Physically separate or shared on same cloud infrastructure 11

OPEN DATA CUBE Development & Interest areas • Ingesting different types of datasets into datacube in an easy way. • Direct Indexing and ingestion of dataset stored in Amazon S 3 buckets • Support for OGC Standards - performant WMS/WCS 1. 0/WCS 2. 0 webservers • Authorization and authentication service. • An external standard directory/catalog • Data Cube on demand • Federation 12

UGANDA DATA CUBE SUMMARY Complex scenario with many actors, processes, data types Open Data Cube helps us capture and manage that complexity Open Data Cube and ARD is a first step It’s the derived ( local ground truthed ) data that’s are important and real power arises from combining diverse datasets We use ODC for development ( cross-verification ) and operations ODC Integrated into a wider cloud-agnostic platform, managing processing chain, compute resources, security and billing etc. Enriching flow of information between stakeholders and building an accessible store of key environment information as a national asset DFMS Overview 13

s. reid@rheagroup. com www. dfms. co. uk s. reid@rheagroup. com