At the beginning of the project, it seemed reasonable to assume that rain gardens would have greater infiltration, greater short term storage of runoff, and greater rates of evapotranspiration than untreated vacant parcels. The advantages of rain gardens over mowed vegetation cover derived from the higher porosity of rain garden soils and higher rates of evapotranspiration by rain garden plant cover.
By pairing experimental and control vacant parcels in the three cities, the experimental design of the project provides data to test this assumption. Experimental parcels included primary and secondary rain gardens (site descriptions) and control parcels all had mowed lawns that were planted and maintained after home demolition. Although there were unexpected problems with monitoring all experimental and control parcels of the three neighborhoods, we were able to obtain data for three experimental and one control parcel in both Gary and Buffalo neighborhoods and continuous monitoring of one Cleveland experimental parcel.
As explained in a previous blog post (here), monitoring of soil moisture provides data to evaluate variability among controls and experimental parcels due to both infiltration and evapotranspiration. Soil moisture monitoring at three depths (3cm, 10 cm, and 20 cm) in the primary rain gardens of the experimental parcels and at a central location in the control parcels yields data for analysis of profile changes and soil moisture content in the upper 20 cm of soil. All rain gardens had a maximum depth of 25 cm. Native soil types for the parcels varied among the three neighborhoods, with Gary having sandy-loam soils, Cleveland clay dominated loams, and Buffalo with compacted loam soil. Figure 1 illustrates the differences in amplitude of variation of Buffalo and Gary parcel soils at 20 cm depth. The greater amplitude of variation in soil moisture in Gary parcels is likely the consequence of greater porosity of the underlying sandy soils. I used a depth-weighted average soil moisture to estimate the average soil moisture of the upper 20 cm of monitored profiles.
Parcel
|
r
|
p-value
|
B1
|
Mean Ratio
|
Buffalo E1
|
0.81
|
< 0.001
|
0.88
|
1.1
|
Buffalo E2
|
0.79
|
< 0.001
|
0.55
|
0.81
|
Buffalo E3
|
0.89
|
< 0.001
|
0.58
|
0.77
|
Buffalo C1
|
0.94
|
< 0.001
|
0.86
|
1.1
|
Gary E1
|
0.39
|
>0.05, NS
|
0.18
|
0.46
|
Gary E2
|
0.66
|
< 0.001
|
0.19
|
0.35
|
Gary E3
|
0.73
|
< 0.001
|
0.16
|
0.33
|
Gary C1
|
0.81
|
>0.05, NS
|
0.61
|
0.86
|
Cleveland Hulda
|
0.93
|
< 0.001
|
0.71
|
0.76
|
As discussed in previous post (here), the rate of decline of soil moisture following a rainfall event is a useful estimate of infiltration. Patterns of association between soil moisture increment and post-rainfall decline, however, are highly variable (see Figures 4 and 5, and Table 2). Only four of the nine parcels have statistically significant correlations, so inferences are limited. Of the four statistically significant correlations (Buf E2, Buf E3, Gary E1, and Gary E2), the Gary regression slopes are greater, reflecting the greater porosity of the native soils as in Table 1. Because mowed lawns ought to have lower infiltration rates than the more porous rain garden soils, the control parcels should have lower slopes as is the case in Figures 4 and 5. The pattern for Buffalo E1 (608 W. Utica) is anomalous. The mean decline rate following rainfall for this parcel is -1.43e-06 (1/s) which is higher than the mean of the other parcels in Buffalo. The co-location of a nursery on this parcel also complicates inference about its infiltration characteristics. The remaining parcels (Gary E3 and Cleveland Hulda) with non-significant correlations are generally consistent with expectations. Overall, there are indications that the rain gardens do have higher infiltration rates than control parcels, but the evidence is equivocal.
Parcel
|
Correlation Coeficient
|
p Value
|
B1
|
Mean decline rate (1/d)
|
Buffalo E1
|
0.34
|
0.13
|
6.83e-06
|
-1.43e-06
|
Buffalo E2
|
-0.35*
|
0.04
|
-1.59e-05
|
-7.86e-07
|
Buffalo E3
|
-0.51*
|
0.015
|
-4.36e-06
|
-3.971e-07
|
Buffalo C1
|
0.005
|
0.98
|
3.79e-08
|
-5.216e-07
|
Gary E1
|
-0.64*
|
4.7e-05
|
-8.60e-06
|
-5.06e-07
|
Gary E2
|
-0.38*
|
0.031
|
-5.67e-06
|
-4.05e-07
|
Gary E3
|
-0.17
|
0.39
|
-5.20e-06
|
-3.53e-07
|
Gary C1
|
-0.32
|
0.27
|
-1.62e-06
|
-2.48e-07
|
Cleveland Hulda
|
-0.44
|
0.12
|
-5.31e-06
|
-4.03e-07
|
Comparisons evapotranspiration losses of experimental and control parcels requires estimates of potential evapotranspiration. I prepared an R-script (RET_Calc.R) to estimate evapotranspiration potential based on the January 2005 ASCE-EWRI Task Committee Report (link) assuming a tall coverage crop. Figures 6 and 7 show patterns of variability of reference evapotranspiration rate based on local weather station data collected in the Aetna neighborhood of Gary and the PUSH neighborhood in Buffalo. The patterns vary with incident solar radiation, temperature and relative humidity. Mean values are 3.43 mm/d for Gary, 2.33 mm/d for Buffalo, and 4.18 mm/d for a weather station in Cleveland's Woodland Hills neighborhood. Sufficient data were available to estimate daily rates of soil moisture loss for seven multiday intervals in both Buffalo and Gary and four multiday intervals in Cleveland. Table 3 summarizes the variability in loss rates for all nine monitored parcels. As in Table 2, the Buffalo E1 parcel has an anomalously high ratio of mean loss rate to reference evapotranspiration. Otherwise, the control parcels seem to have equal if not higher rates of evapotranspiration than the rain garden parcels.
Figure 6. Estimated reference evapotranspiration based on the ASCE-EWRI recommended method. Weather station data were obtained from the Aetna weather station for June 27 to August 19, 2016. |
Parcel
|
Mean loss rate (1/d)
|
MLR (mm/d)
|
Ratio of MLR:RET
|
Buffalo E1
|
-5.25e-07
|
9.07
|
3.89
|
Buffalo E2
|
-8.85e-08
|
1.53
|
0.66
|
Buffalo E3
|
-5.32e-08
|
0.92
|
0.39
|
Buffalo C1
|
-8.25e-08
|
1.43
|
0.61
|
Gary E1
|
-1.40e-07
|
2.42
|
0.71
|
Gary E2
|
-1.05e-07
|
1.81
|
0.53
|
Gary E3
|
-1.83e-07
|
3.17
|
0.92
|
Gary C1
|
-2.29e-07
|
3.96
|
1.15
|
Cleveland Hulda
|
-1.71e-07
|
2.96
|
0.71
|