Computational and Data Challenges in Environmental Modelling Sensors and Instrumentation Dr Peter M Allan Director, Hartree Centre, STFC Overview

• Hartree Centre – what are we • Sensors and instrumentation • The data challenge

The Hartree Centre

• Part of the STFC National Laboratories – Close working with the Scientific Computing Department is essential to delivering our mission • Our mission – Provide market led research and innovation and development services for competitive advantage and wider economic impact by accelerating the application of High Performance Computing, Data Science, and Cognitive techniques The Hartree Centre Establishment

2012 Creation of Hartree Centre as industrial driver in HPC • 2011 Tildesley Report recommendation • Investment from BIS • Economic impact through Software & Modelling • Focus on industrial engagement

2015 Major Investment in Collaborative Research • Data Centric & Cognitive Computing • Embedded IBM Research Centre • Extended industrial & scientific reach

What is the Hartree Centre? Joint Centre for Collaborative R&D in the UK Research Innovation

Development Deployment IBM Research Division

STFC IBM UK Research

Industry

Public Sector

Science

Academia Government Story so far Mission Phase 1 – HPC Enablement ⁻ Industrial engagement leveraging STFC scientists ⁻ Leading Research infrastructure for software development and industrial support ⁻ Collaboration with IBM for global insights Mission Phase 2 ⁻ Research into Energy efficient computing ⁻ Expanding HPC into Analytics ⁻ Building global partnerships Results ⁻ Completed 100+ engagements ⁻ 33 Collaborative R&D projects completed ⁻ 27 published case studies ⁻ 6 joint IBM / Hartree research projects ⁻ Pioneering convergence of HPC & Big Data Analytics ⁻ Parallel Computing Centre ⁻ IBM / Benchmarking Centre ⁻ IBM / Development Centre ⁻ MOU and partnership with Lawrence Livermore ⁻ MSc with University of Liverpool Mission Phase 3 – R&D programme ⁻ Develop tools needed by industry in the future Hartree Centre Clients Sensors

• Location – Land, air, sea, beneath the surface, space, everywhere! • In situ or remote sensing • Space – Instruments on satellites – Mostly remote sensing • Everywhere – Internet of Things – Mostly in situ measurements

Sensors in Space

• Earth Observation – Mostly from satellites in low Earth orbit (several hundred km altitude) – Some from geostationary orbit (weather satellites) • Space Weather – Solar storms • Both remote sensing and in situ

Sierra Leone River Estuary © Copernicus Sentinel data (2015)/ESA Sensors in Space

• Sentinel satellites from ESA/EU – S-1 : Synthetic Aperture Radar (5GHz) – S-2 : Multispectral imager (optical, IR) – S-3 : Sea and Land Surface Temperature Radiometer Ocean and Land Colour Altimetry (dual band SAR) Microwave radiometer – S-4&5 : Composition of atmosphere – S-6 : Radar altimeter for ocean topography

Advantages of Sensors in Space

• Continuous stream of data covering the whole globe • Can be very accurate – SLSTR will measure temperatures accurate to 0.1K • Cost-effective if you need global coverage • Data often provided free at the point of use for non- commercial applications • Data is well curated

Disadvantages of Sensors in Space

• Repetition rate of data from a given place typically several days (but see later) • If something breaks, you generally can’t fix it • Long time from conception to delivery of first data

Faster delivery of data from space

• Geostationary Orbit – High repetition rate, but lower spatial resolution • Constellations of satellites – Surrey Satellites’ DMC – Planet Labs, ~100 cubesats • Relay satellites – European Data Relay System • Laser inter-satellite communications

Internet of Things

• Sensors are everywhere – Low cost – Rapid deployment – May be owned by individuals or organisations • Data are of variable or unknown accuracy • Sensors may be active at variable times • When are such uncontrolled sensors of practical use?

Data Intensive Solutions

• What type of data best address your problem? • Do you need multiple types of data to even get you started? • Do other types of data improve your solution or are they a distraction? • Who are the experts in data that you have not used before? • Who is going to use your solution and what will they do with it? Big Data Challenge

• Satellite data are large volume – Well structured, standard formats – How to access the large volume of data? • Provide computing next to the data (JASMIN/CEMS) • IoT sensors have high variety • The Challenge – How best to make use of the increasing number and variety of data sources? – There are many experts in the use of some data sources, but few in exploiting the full range Questions?