Update Landlab Landslide component to read depth to groundwater table instead of recharge

[ChristinaB]

• [x] Change code

• [x] Test code

• [ ] Pull request to Landlab

[ChristinaB]

Compare landslide probability to average depth to water table (during max saturated extent events/storm).

[RondaStrauch]

Current code: relative wetness Rw = min[R.a/T.sin(theta),1] where R comes from VIC runoff + baseflow

Rw also equals the ratio of groundwater depth to soil depth. (Rw = hw/hs) Thus, if we extract hw from DHSVM, then we use this to calculate Rw.

So we don't need transmissivity (or Ksat) or specific contributing area (a) as we have in the code already.

[RondaStrauch]

• [ ] We should understand how transmissivity is considered within DHSVM

[ChristinaB]

To do by April 5, 2018

[ChristinaB]

@RondaStrauch
I talked with Erkan and I took these notes for updating. I will change the code for now to hs/hw just to see how it looks, but we may miss convergence zones. Another idea is to calculate recharge based on DHSVM depth to water table using DHSVM equations.

[ChristinaB]

@RondaStrauch I talked to Erkan again about this and we agreed that there are three ways to run the model improvement.

0. 150 dtw and 150 m soil depth. This is what we have currently running.
1. low resolution recharge or dtw (150 m) and high res soil depth (soil evolution model).
2. high res dtw by estimating recharge from the 150 dtw and calculating Re using 30 m wetness index (no soil evoluation model needed).
3. high res dtw and soil evolution model.

We don't know which of these will really be needed until we start looking at results.

[ChristinaB]

we can replace q (eq2) with ar (eq 3) and solve for 30m recharge with a smooth high res topographic gradient. from http://www.hydrology.bee.cornell.edu/Papers/WalterHP02.pdf

[RondaStrauch]

@ChristinaB - Sorry took me so long to look and think about this. I would rather not go through the effort involved with the soil evolution model for initial application...lets just get this running and we can improve later. So what I'm understanding is based on Topmodel, we would estimate recharge using our 150m dtw via the equation Recharge = dtwtan(slope)/soildeptha. The soil depth in this case would be SSURGO resampled at 30 m, but really at MU resolution (i.e., repeated for every 30m in MU). The slope and 'a' would be at 30 m resolution. This would give us higher resolution topographic gradient influenced recharge. Think this your improvement #2. We could compare this to results from just using dtw at 150m and soil depth at 30m (i.e., dtw is repeated for every 30m in 150m grid) and the dtw/hs=Rw formula.

[ChristinaB]

You got it. Option #2 coming right up! Erkan will be happy with this as well. I agree with the approach of getting SOMETHING running before overcomplicating in directions that we are not sure are needed. If results are not acceptable, we have a few backup plans in our back pockets. I have big back pockets.

[ChristinaB]

@RondaStrauch Could you check the input parameters in this component update? Thanks. I'll retest after that. Or let me know here what the defaults should be. Thanks! C

[ChristinaB]

there are three ways to run the model improvement.

0. 150 dtw and 150 m soil depth. This is what we have currently running.

1. low resolution recharge or dtw (150 m) and high res soil depth (soil evolution model).

2. high res dtw by estimating recharge from the 150 dtw and calculating Re using 30 m wetness index (no soil evoluation model needed).

3. high res dtw and soil evolution model.

We don't know which of these will really be needed until we start looking at results.

With recharge as an input (see Line 676): if self.groundwater__recharge_distribution == 'uniform': self._rel_wetness = ((self._Re)/self._T)*(self._a/np.sin( np.arctan(self._theta)))

With depth to water table as an input: elif self.groundwater__depth_distribution == 'uniform': self._rel_wetness = ((self._hs_mode-self._De)/self._hs_mode)

[ChristinaB]