Bug in HazardCurveLocal

[Kevin Milner]

Steps to reproduce: run Hazard Curve Application, select Mean UCERF2, set Location to "34.10688, -118.22060", hit compute. Interesting this problem isn't present with the default site.

Other info: The problem is with UCERF2 source 3337, rup 77, which is a Point2Vert_SS_FaultPoisSource source with 99 points. Can also be reproduced with the following code:

CB_2008_AttenRel imr = new CB_2008_AttenRel(null);
imr.setParamDefaults();
MeanUCERF2 ucerf = new MeanUCERF2();
ucerf.updateForecast();
ProbEqkSource src = ucerf.getSource(3337);
ProbEqkRupture theRup = src.getRupture(77);
imr.setEqkRupture(theRup);
imr.setIntensityMeasure(PGA_Param.NAME);
Site site = new Site(new Location(34.10688, -118.22060));
site.addParameterList(imr.getSiteParams());
imr.setAll(theRup, site, imr.getIntensityMeasure());

Application: HazardCurveLocal?
Version: 1.2.3
Bulid Type: Development
Java Version: 1.6.0_29 (Sun Microsystems Inc.)
Operating System: Linux (arch: amd64, version: 3.0.0-14-generic)

Exception:

java.lang.IllegalArgumentException: Internally computed Area is empty
	at com.google.common.base.Preconditions.checkArgument(Preconditions.java:88)
	at org.opensha.commons.geo.Region.createArea(Region.java:616)
	at org.opensha.commons.geo.Region.initBorderedRegion(Region.java:662)
	at org.opensha.commons.geo.Region.<init>(Region.java:185)
	at org.opensha.sha.faultSurface.utils.GriddedSurfaceUtils.getPropagationDistances(GriddedSurfaceUtils.java:116)
	at org.opensha.sha.faultSurface.AbstractEvenlyGriddedSurface.setPropagationDistances(AbstractEvenlyGriddedSurface.java:198)
	at org.opensha.sha.faultSurface.AbstractEvenlyGriddedSurface.getDistanceRup(AbstractEvenlyGriddedSurface.java:214)
	at org.opensha.sha.imr.param.PropagationEffectParams.DistanceRupParameter.calcValueFromSiteAndEqkRup(DistanceRupParameter.java:130)
	at org.opensha.sha.imr.param.PropagationEffectParams.PropagationEffectParameter.setValue(PropagationEffectParameter.java:103)
	at org.opensha.sha.imr.attenRelImpl.CB_2008_AttenRel.propEffectUpdate(CB_2008_AttenRel.java:230)
	at org.opensha.sha.imr.attenRelImpl.CB_2008_AttenRel.setPropagationEffectParams(CB_2008_AttenRel.java:222)
	at org.opensha.sha.imr.attenRelImpl.CB_2008_AttenRel.setEqkRupture(CB_2008_AttenRel.java:206)
	at org.opensha.sha.calc.HazardCurveCalculator.getHazardCurve(HazardCurveCalculator.java:363)
	at org.opensha.sha.gui.HazardCurveApplication.computeHazardCurve(HazardCurveApplication.java:1337)
	at org.opensha.sha.gui.HazardCurveApplication.run(HazardCurveApplication.java:928)
	at java.lang.Thread.run(Thread.java:662)

Metadata:

Cacluation Type = Probabilistic

IMR Param List:

---

IMR = Campbell & Bozorgnia (2008); Gaussian Truncation = None; Tectonic Region = Active Shallow Crust; Component = Average Horizontal (GMRotI50); Std Dev Type = Total

Site Param List:

---

Longitude = -118.2206; Latitude = 34.10688; Vs30 = 760.0; Depth 2.5 km/sec = 1.0

IMT Param List:

---

IMT = SA; SA Period = 1.0; SA Damping = 5.0

Forecast Param List:

---

Eqk Rup Forecast = WGCEP (2007) UCERF2 - Single Branch; Rupture Offset = 5.0; Floater Type = Along strike & centered down dip; Background Seismicity = Include; Treat Background Seismicity As = Two perpendicular faults; Apply CyberShake? DDW Corr = false; Probability Model = WGCEP Preferred Blend

TimeSpan? Param List:

---

Duration = 30.0; Start Year = 2012

Calculation Settings:

---

Maximum Distance = 200.0; Num Event Sets = 1; Use Mag-Distance Filter? = false; null; Set TRT From Source? = false; If source TRT not supported by IMR = Use TRT value already set in IMR; Pt Src Dist Corr = None

[Peter Powers]

Code formatting issue: brackets!

Ned, please confirm that the fixed form below was your original intent. The problem crops up for any locations that are within half the fault discretization distance of a background source. The bg source/surface was being incorrectly having its border/trace converted to a region. Fixed in [8443]

The following lines from GriddedSurfaceUtils?:

if(frankelTypeSurface)
	if(isDistJB_ReallyZero(surface,distJB))
		distJB=0;
else {
	Region reg = new Region(surface.getPerimeter(),BorderType.MERCATOR_LINEAR);
	if(reg.contains(loc)) 
		distJB=0;
}

look fine and dandy except that the 'else' pairs with the second if, not the first. Reformatted to:

if (frankelTypeSurface) {
	if (isDistJB_ReallyZero(surface, distJB)) distJB = 0;
} else {
	Region reg = new Region(surface.getPerimeter(),
		BorderType.MERCATOR_LINEAR);
	if (reg.contains(loc)) distJB = 0;
}