Checkerboard test

A checkerboard test is a way to examine the ability of your data to resolve locking on a fault. The method is to generate a known locking pattern, calculate expected values at observation points, add random noise to the observations using standard deviations taken from the true data, then invert the data to try to recover the known locking pattern. The generation of the known locking model data should include other parameters of the model near their best value (for example, poles should be approximately correct). You can also use this approach to estimate the resolution of any other parameter.

Steps

Set up your faults (FA:), blocks (BL:), poles (PO:), etc. as usual.
Add data using GP: or IS:
Poles should be non-zero (except reference) and close to your final estimates

Assign model name chk1

#checkerboard forward model 1
mo: chk1

#For example, the fault (for this example fault 9) has 14 nodes along strike and 8 downdip. 
# Set up the forward calculation by assigning locking indices in a checkerboard pattern. This
# example uses 4 nodes (2x2) for each patch of uniform locking

ft: 9 0 0 0 0 

nng: 9 14 8
1 1 2 2 1 1 2 2 1 1 2 2 1 1 
1 1 2 2 1 1 2 2 1 1 2 2 1 1
2 2 1 1 2 2 1 1 2 2 1 1 2 2
2 2 1 1 2 2 1 1 2 2 1 1 2 2
1 1 2 2 1 1 2 2 1 1 2 2 1 1 
1 1 2 2 1 1 2 2 1 1 2 2 1 1
2 2 1 1 2 2 1 1 2 2 1 1 2 2
2 2 1 1 2 2 1 1 2 2 1 1 2 2

#Assign full locking to nodes with index 1 and no locking to nodes with index 2.
nv: 9 1.0 0.0

#Set flags so fault 9 has locking but is not inverted.
ff:   9
fi:  -9

#Set iteration control for forward model only.
ic: 0

#Don't adjust poles
pi: 0

#The forward runs will produce estimated values of all your data. You add random 
#uncertainties to the estimates using the +rnd flag:

fl: +rnd
#This produces files in the chk1 directory ending with _rand.vec and _rand.insar

#You can reverse the locking pattern easily by:
nv: 9 0.0 1.0

CHK1: Locking pattern for forward model

#Solving for locking using randomized data

# chk2 inverse model for checkerboard chk1, InSAR plus GPS
mo: chk2
pf: "chk2/pio" 3
 
fi: 9
ff: 9
ft: 9 0 0 0 0 

# Assign nodes in checkerboard pattern, but each 2x2 set of nodes has
# different index

nng: 9 14 8
 1  1  2  2  3  3  4  4  5  5  6  6  7  7
 1  1  2  2  3  3  4  4  5  5  6  6  7  7
 8  8  9  9 10 10 11 11 12 12 13 13 14 14
 8  8  9  9 10 10 11 11 12 12 13 13 14 14
15 15 16 16 17 17 18 18 19 19 20 20 21 21 
15 15 16 16 17 17 18 18 19 19 20 20 21 21
22 22 23 23 24 24 25 25 26 26 27 27 28 28
22 22 23 23 24 24 25 25 26 26 27 27 28 28

#Starting locking model values

nvg: 9 14 8
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

# remove data already read in
cl: gps ins

# poles to estimate (optional)
pi: 2
 
# read randomized gps vectors 
gp1: CMNC "chk1/chk1_rand.vec"   1 1.0 0 0 0 1999.0 2016.0 0.3 2.5 0 1 1 0

# read randomized InSAR (INS1 is the name assigned to InSAR data in IS: for forward run)
is4: DESM "chk1/chk1_INS1_rand.insar"  1 1.0 2016 2017  0 0 3 0 0 0 2.0

#iteration control
ic: 1 2 1 2  

em:

Run model chk1 to generate randomized data. Then run chk2 to estimate the locking of model chk1.

CHK2: Recovered Locking pattern for model chk1