These are our introduction slides for the Ariel Re / Lloyd's / Argo US flood model comparison workshop.

1. KatRisk US Flood Model
London, November 2017
KatRisk LLC
752 Gilman St.
Berkeley, CA 94710
510-984-0056
www.katrisk.com
KatRisk Deutschland GmbH
Wilhelmstr. 6
79098 Freiburg, Germany
0761-5146-7600

2. 2
KatRisk USA Flood Model Highlights
 Correlated flood, TC wind (Atlantic and Pacific), and storm surge models
 10m inland flood and storm surge hazard resolution
 2d hydraulic model with no lower threshold for catchment size (all locs can flood)
 50k years of correlated simulations for all perils
 Global sea surface temperature correlation of cat models across continents
 Flexible model execution architecture to investigate model assumptions and
sensitivities (climate change, correlations to global teleconnections)
Harvey Event Footprint Stochastic Hurricane Tracks Storm Surge Score

3. 3
KatRisk USA Flood, Wind, Storm Surge AAL
 Ran test portfolio (from this workshop) and KatRisk Industry
Exposure for hurricane wind, storm surge and inland flood
 Test portfolio represents about 1% of value compared to the
KatRisk Industry Exposure, but different LOBs, building types,
construction, etc. makes comparison not 1:1
 Result: The KatRisk inland flood GU AAL is about as large as the
combination of Atlantic hurricane wind and storm surge losses
 About 18% of inland flood losses are caused by tropical cyclones
Exposure / Peril Flood GU Storm Surge GU Hurricane Wind GU
Model Comparison AAL $207 Million $49 Million $192 Million
Industry Loss AAL $15 - $20 Billion $4 – $6 Billion $12 - $16 Billion

4. 4
KatRisk SpatialKat Software Features
1.2 million locations Ariel Re test portfolio,
GU & GR, 10 samples, 50k years
Inland Flood Storm Surge TC wind
Runtime 25 cores (2x Xeon E5-2690 v4) 4 minutes 1 minute 4 minutes
Runtime 4 cores (Intel i7-7700k) = Desktop 20 minutes 5 minutes 19 minutes
 Enables cat modeling on laptop / desktop / server / cloud
– Open Source based efficient C++ and R based engine with memory efficient kernel (< 1GB)
– Location aware repeatable sparse antithetic latin-hypercube sampling enables coherent
measures of risk with fast loss convergence (2 to 1000 samples)
– View and change hazard, vulnerability, location level and regional flood defenses
– User specified spatial secondary uncertainty correlation with KatRisk defaults
– Easy file formats to implement OASIS type and other cat models
 True multi-peril: wind, storm surge, and inland flood. Flexible inuring order of contracts,
multiple policy files (by peril) with high performance in-memory execution
 Financial contracts up to Net Pre Cat Loss
Requires > 50MB/sec r/w speed per core
SATA III SSD sufficient for up to 10 cores
Low run-times allow performing multiple runs and sensitivity studies
even on local laptop / desktop. Hosted API for location loss analytics.