VIEWS: 3 PAGES: 17 POSTED ON: 9/27/2010 Public Domain
Statistical Soil Moisture Analysis Kelly Wilhelm and David Smith Statistical Methods (105) Project Report May 5, 2008 Introduction: Soil, the weathered remains of rocks and organic matter provides both faunal and floral organisms a medium for ecological growth. A specific characteristic of soil is the pore space between nodules of clay, silts and clastic rock grains that allow for percolation of water from precipitation events. This critical attribute allows for the aqueous migration of nutrients to the roots of plants which are processed as inputs into the biological cycle. Thus, soil moisture is defined as the amount of water one can evaporate, by weight percent, of an original soil sample. Interestingly, soil moisture is highly variable and is influenced by many surface characteristics included topography changes, slope declination, vegetation cover, and potentially the amount of solar radiation received creating evaporative surface conditions. The focus of this project is to find and statistically represent the variability of soil moisture content within the grade of surface topography. Our goal is to analyze the changes in soil moisture content on three sections of a hill: summit, slope and foot (see figure 1). Then repeat the sampling system and compare the three sections between an east facing and west facing slopes for statistical differences (see figure 2). [ Figure 1] East Facing Hill Sample Field Summit Slope Transect Base Line [ Figure 2 ] West Facing Slope East Facing Slope Methods: Initially, two hills were selected for having similar characteristics of slope, vegetation cover, area and height, but varying in the overall facing direction of East and West (Figure 2). In order to complete a survey, transect methods of stratified random sampling were implemented. For each hill, a transect line from summit to base was selected for have traits of smooth surface transitions and lack of visible disturbance in surface flow down a slope (White line in figure 1). Samples collected were measured in a fifty foot by fifty foot square (the three locations in figure 1) area of sampling. For the three sampling locations found on a hill, fifteen samples were chosen to be collected. For each sample the program Microsoft Excel was used to randomly select the length down the transect line, whether to sample on the left or right side of the transect, and the overall width away from the transect. An example or our stratified random sample output can be seen in table 1. At each selected location a soil auger was used to extract the sample. The soil samples were subsequently labeled and stored in individual sandwich bags for further processing and to prevent evaporation of moisture. Since a direct measure of moisture in the soil is not possible, using a precise scale we measured the original sample weight, then evaporate all water, and re-measure the dried sample. Evaporation was accomplished by cooking the samples at 450 degrees for 8 hours. The potential weight difference will indicate how much moisture was evaporated from each sample as found in table 2 (appendix). [ Table 1 ] West Facing [Slope] East Facing [Slope] Sample # Length Right/Left Width Sample # Length Right/Left Width 1 10 R 11 1 1 R 0 2 5 L 1 2 22 R 5 3 27 R 24 3 10 R 4 4 40 R 6 4 6 L 7 5 50 R 8 5 20 R 9 6 35 L 17 6 2 R 3 7 11 R 22 7 19 L 1 8 4 R 4 8 20 L 0 9 30 L 2 9 21 L 3 10 21 L 2 10 14 R 3 11 2 L 24 11 25 R 19 12 33 L 11 12 8 L 11 13 2 R 2 13 17 R 5 14 49 R 7 14 4 L 15 15 3 R 13 15 23 L 20 Statistical Analysis: The soil sampling analysis was conducted under the hypothesis of finding differences in variability of soil moisture content within the grade of surface topography. Our initial hypothesis was to find variation within the three sections of a hill. This occurs as precipitation follows topography from high to low regions. High gradation of slopes significantly decreases the moisture holding capacity of the soil therefore yielding potential differences. Therefore, the mean soil moisture of samples collected from the summit, slope and base of a hill will potentially have statistically significant differences. This is represented by our Hypothesis: Ho: µtop = µslope = µbase Ha: µtop ≠ µslope ≠ µbase Another interesting statistical analysis would be the variability of soil moisture content within topography regions between two similar hills. Thus, by measuring the aforementioned regions in two hills, a statistical comparison of variation could be made to find a potential statistical difference between facing hills. This is stated by our second hypothesis: Ho: µtop1 = µtop2: µslope1 = µslope2: µbase1 = µbase2 Ha: µtop1 ≠ µtop2: µslope1 ≠ µslope2: µbase1 ≠ µbase2 The analyses will be entered through SAS statistical software. Initially, means procedure and box plot will be utilized for an overall statistical analysis. To find statistical differences between east and west facing locations and compiled means individual t tests will be conducted. For the comparison of multiple individual location means, ANOVA statistical analysis will be used to find the variance for more than two means. Tests of regression will be utilized. as evidence for potential sampling biasness. [ Table 3 ] The MEANS Procedure Analysis Variable : pwet Lower 95% Upper 95% hill Obs N Mean Median Std Dev CL for Mean CL for Mean Variance _______________________________________________________________________________________ EastBase 15 15 0.0026350 0.0025980 0.000721298 0.0022355 0.0030344 5.2027049E-7 EastSlop 15 15 0.0026373 0.0025305 0.000677979 0.0022619 0.0030128 4.5965512E-7 EastTop 15 15 0.0028267 0.0029243 0.000727137 0.0024240 0.0032294 5.2872833E-7 WestBase 15 15 0.0021408 0.0022641 0.000434937 0.0018999 0.0023816 1.8917028E-7 WestSlop 15 15 0.0023937 0.0022049 0.000683253 0.0020154 0.0027721 4.6683402E-7 WestTop 15 15 0.0025823 0.0025526 0.000539567 0.0022835 0.0028811 2.9113221E-7 Statistical Results: Means Procedure To begin, the analysis from the proc means procedure yielded data that supports the use of two independent t tests and ANOVA. The data collected is represented by a simple random sample, all sample locations have the same amount of sample individuals as well. The standard deviations do not display a high level of variance which ranges from .00043 to .00072. Thus, the largest standard deviation is smaller than twice the smallest sample standard deviation yielding a ration of 1.67 which is less than 2. This is a great indication for the data being in compliance with the rules that satisfy ANOVA. The overall dataset does not appear to have significant outliers as the mean and median of all samples are approximately equal with little variance. This data indicates the satisfaction of rules regarding two-sample t procedures and ANOVA statistical analysis at an alpha level of .05. Box Plots The Slope Location Box plot (appendix Figure 3) displays a comparison of the weight percent of water for each section of the hills. The analysis required the hill classification: Top, Slope, and Base. Since there were two hills made for a total of six box plots – three for each of the two hills. The mean for the weight percent of water was fairly homogenous throughout each sample; the means ranged from only 0.23%-0.29%. The sample homogeneity is unexpected since the water should percolate down to the bottom over time. The hill Top is displayed as contained the largest water weight percent for each individual hill (east or west alone), followed by Bottom, and slope always contained the least weight percent water. The data The East v West Box plot (appendix Figure 4) displays a comparison of the percent of water was within each sample for all samples between each hill. The mean percent water weight for the East facing hill was a bit more than 0.25%, while the mean percent water weight for the West facing hill was slightly less than 0.25%. The near identical mean for the two hills is not surprising since they were hills of similar slope, thus the same forces should have acted on both. The East facing hill had a wider range of water weight percentages compared to the West facing hill, but was generally not different from the West facing hill. The spread of the box plots satisfies the requirement for two-sample t producers to have equal populations and distributions in shape. T-test – Statistical Analysis Tops The first t test table (appendix table 4) displays the confidence limits of the mean and standard deviations between the East and West facing Top locations, and the difference between the two. The confidence interval for the difference in means is (-0.00023, 0.0007), and the confidence interval for the difference in standard deviation between the two slopes is (0.0006, 0.0009). With 44 degrees of freedom the p value was 0.2763, much less than the F value of 1.82. This indicates that the null cannot be rejected; although it does not prove the null either. t-Test – Slopes: The second t test table (appendix table 5) displays the confidence limits of the mean and standard deviations between the East and West facing Slope locations, and the difference between the two. The confidence interval for the difference in means is (-0.00027, 0.0008), and the confidence interval for the difference in standard deviation between the two slopes is (0.0005, 0.0009). With 44 degrees of freedom the p value was 0.9773, slightly less than the F value of 1.02. This indicates that the null cannot be rejected; although it does not prove the null either. t-Test – Bases: The third t test table (appendix table 6) displays the confidence limits of the mean and standard deviations between the East and West facing Base locations, and the difference between the two. The confidence interval for the difference in means is (0.0000487, 0.0009), and the confidence interval for the difference in standard deviation between the two slopes is (0.0005, 0.0008). With 44 degrees of freedom the p value was 0.0684, much less than the F value of 2.75. This indicates that the null cannot be rejected; although it does not prove the null either. t-Test - East versus West Total Slope: The fourth t test table (appendix table 7) displays the confidence limits of the mean and standard deviations between the all the samples of each the East and West facing hills, and the difference between the two. The confidence interval for the difference in means is (0.0000586, 0.0006), and the confidence interval for the difference in standard deviation between the two slopes is (0.0006, 0.0008). With 44 degrees of freedom the p value was 0.2148, much less than the F value of 1.46. This indicates that the null cannot be rejected; although it does not prove the null either. ANOVA statistical test: In order to avoid multiple P-values of lone tests in comparing the six sample areas, ANOVA was selected to find the differences in parameters of soil moisture means per location. The box plots help to indication a normal distribution of independent simple random samples, and identical sample populations per location collected help to reduce and increase the robustness of the ANOVA analysis. Our original null hypothesis stated no difference between the means of the top, slope and base. The resulting F-value of 2.08 (appendix table 8) indicates a value relatively close to 0 thus indicating support for the null hypothesis of the population means being similar. The total sample collection contained 90 observations from six sample locations producing 84 degrees of freedom. The completed result of the F-test produced a value of 2.09 with a p value of .0765. Thus, by finding the F distribution critical value with a numerator of 5 and a denominator of 84 the resulting p value is approximately 2.34. Thus, pr >F gives evidence that the null hypothesis cannot be rejected at the alpha level of 0.05. Being that the F-test is not significant at 0.05, the individual t-tests of the Bonferroni Correction also displayed each hill sample location as supporting the null hypothesis without findings of significant differences, as indicated by the same character value for all locations. Project Conclusions: While the original alternative hypothesis indicated a significant change in the soil moisture levels based on topography, our final statistical conclusions indicate that at a significance of 0.05 the multiple t-tests and ANOVA conclude in finding no significant differences. Thus, the experiment confirms the null hypothesis that the means within slopes and between two independent hills have approximately the same means at an alpha of 0.05. Though, the null hypothesis could not be rejected throughout the sampling process, there where many indications of potential experimental errors, and alterations that would potentially help similar studies. The distribution of the box plots for each individual hill hints at the potential of soil composition differences. The variation of clay and silts throughout the soil profile and sampling location can have a dramatic effect on moisture. Clay, having hydrophobic and soil moisture retention properties, could dramatically decrease the amount of water entrance into the soil. This could lead to precipitation creating overland flow without percolation into the ground. This would also allow water to flow on the slopes, and bases and potentially causing similarities in soil moisture means. In future analyses, the variable of overall soil composition should be collected and used for statistical comparisons. Ideally, more consistent sampling protocol should have been utilized, though time constraints prevented this in our research. Throughout, the sample procedure taking similar lengths of soil core was inhibited as varying moisture and soil consolidation made collecting similar masses unfeasible for our timeframe. One of our overall conclusions is that ground locations were the soil moisture seemed low yielded samples of lower mass. Also, that sample locations of extreme surface wetness, samples were surprisingly easy to collect, producing larger sample cores from the saturated soil. In order to examine potential sampling errors in collecting a regression analysis was used to examine statistical correlation of the original sample weight and the percentage of moisture evaporated from the samples and can be found in (figure 5). This regression compares the starting weight of a sample to the same sample’s evaporated water weight. The regression line was negative and had a moderately strong correlation (R Squared = .7789). It is surprising that that regression line was negative, because that means that as a sample became larger, a smaller total percentage of water was evaporated. We expected that even as the weight increased, the overall percentage of water would stay consistent. A possible reason for the smaller samples having larger percents of water weight is that the smaller samples could have been from the slopes, which were more soil rich and could not hold as much water as the more clay rich base and top; so this regression could be showing differences in sample composition. The second surprising fact about this scatter plot is that it had such a moderately strong correlation, with a R square of 0.78, it is difficult to deny that there is a correlation between starting weight and weight percent water. The reason for this moderately strong correlation is likely caused by the same compositional differences that created the negative slope. pwet - ei = 0. 0049 - 238E 7 w ght 0. 0045 Another graphical [ Figure 5 ] N 90 Rsq 0. 0040 representation of our soil 0. 7814 dj sq A R sampling can be found in the 0. 7789 ME R S 0. 0003 0. 0035 regression analysis of percent of soil moisture and the overall 0. 0030 change in soil mass from collection to evaporation. This 0. 0025 regression was preformed to examine for potential 0. 0020 correlation between soil sample mass loss and the percentage of 0. 0015 water loss from the samples. The results (Figure 6) yielded 0. 0010 absolutely no correlation 60 70 80 90 100 between the two factors, being 110 120 130 140 150 160 ei as the R squared statistic is w ght <0.0002. This can be contributed to the nature of weight lost was almost always falling between 20 and 28 grams. With this plot a great visual representation of the variability of sample weight loss from evaporation. pwet ei = 0. 0026 - 0. 0004 w ght 0. 0045 [ Figure 6 ] N 90 Rsq 0. 0002 0. 0040 Finally, the time dj sq A R - . 0111 constraints also made our ME R S 0. 0007 sampling schedule fall 1.5 0. 0035 days after heavy rainstorms in the area. Thus, the time 0. 0030 needed from the last rain fall event to find statistical differences in soil moisture 0. 0025 may need to be increase to allow for surface processes 0. 0020 to influence soil moisture. Though with the idealized 0. 0015 set of sampling procedures, 0. 100 0. 125 0. 150 0. 175 0. 200 we believe significant 0. 225 0. 250 0. 275 0. 300 0. 325 0. 350 ei difference can be found to w ght reject the null hypothesis, and perhaps a better understanding of hill surface features could be explained. Appendix Top Left – Kelly displaying our transect stratified random sample procedure. Top Right - An example of our soil weighing procedure. Bottom – All 90 samples placed in an oven at 450°F for 8 hours. [ Table 1 ] Facing % Soil Original Differ West Top 0.0017 136.62 0.225 Direction Location moisture Weight ence East Top 0.0033 67.64 0.236 West Top 0.0024 101.99 0.244 East Top 0.0035 69.11 0.246 West Top 0.0031 79.78 0.221 East Top 0.0024 90.74 0.246 West Top 0.0026 91.84 0.214 East Top 0.0017 128.5 0.234 West Top 0.0025 100.39 0.238 East Top 0.0037 65.13 0.254 West Top 0.0026 93.1 0.225 East Top 0.0017 128.88 0.242 West Top 0.0026 90.15 0.23 East Top 0.0023 100.54 0.234 West Top 0.003 85.21 0.234 East Top 0.0031 76.13 0.252 West Top 0.0024 96.84 0.258 East Top 0.0042 62.06 0.233 West Top 0.0018 119.3 0.229 East Top 0.0029 78.39 0.219 West Top 0.0018 119.38 0.222 East Top 0.0019 117.58 0.218 West Top 0.0036 68.32 0.239 East Top 0.0027 88.67 0.247 West Top 0.0031 78.75 0.214 East Top 0.003 72.3 0.247 West Top 0.0023 108.54 0.234 East Top 0.0033 70.34 0.254 West Top 0.0031 76.33 0.254 East Top 0.0027 93.51 0.239 West Slope 0.0031 76.35 0.242 East Slope 0.0022 112.66 0.233 West Slope 0.0015 156.94 0.23 East Slope 0.0034 68.54 0.234 West Slope 0.0033 70.75 0.258 East Slope 0.002 130.69 0.234 West Slope 0.0028 82.6 0.225 East Slope 0.0031 73 0.233 West Slope 0.0017 143.32 0.113 East Slope 0.0016 68.59 0.24 West Slope 0.0022 125.65 0.274 East Slope 0.0024 116.06 0.277 West Slope 0.0037 62.26 0.247 East Slope 0.0025 98.35 0.233 West Slope 0.0016 152.65 0.242 East Slope 0.0029 82.44 0.247 West Slope 0.0026 90.24 0.241 East Slope 0.0017 141.43 0.238 West Slope 0.0025 93.61 0.253 East Slope 0.0029 86.55 0.229 West Slope 0.003 81.39 0.231 East Slope 0.0025 91.39 0.244 West Slope 0.0016 143.6 0.243 East Slope 0.0031 78.86 0.231 West Slope 0.0021 112.91 0.229 East Slope 0.0025 90.64 0.235 West Slope 0.0021 116.49 0.243 East Slope 0.0025 97.61 0.249 West Slope 0.0021 121.26 0.344 East Slope 0.0043 80.55 0.252 West Base 0.0023 107.9 0.255 East Base 0.002 128.95 0.246 West Base 0.0017 133.1 0.249 East Base 0.0037 67.63 0.225 West Base 0.0017 122.33 0.259 East Base 0.0033 77.3 0.208 West Base 0.0025 86.25 0.291 East Base 0.0028 105.21 0.212 West Base 0.0016 131.33 0.239 East Base 0.0019 128.53 0.207 West Base 0.0028 81.69 0.238 East Base 0.0018 135.68 0.228 West Base 0.0019 109.63 0.253 East Base 0.0036 70.02 0.212 West Base 0.0023 91.4 0.258 East Base 0.0028 92.4 0.214 West Base 0.0017 124.3 0.266 East Base 0.0036 73.54 0.21 West Base 0.0023 96.17 0.234 East Base 0.0022 104.55 0.218 West Base 0.0025 95.95 0.252 East Base 0.0026 97.05 0.237 West Base 0.0026 85.33 0.238 East Base 0.0016 151.8 0.219 West Base 0.002 117.43 0.241 East Base 0.0021 113.47 0.237 West Base 0.0028 79.53 0.276 East Base 0.0032 86.53 0.222 West Base 0.0015 146.12 0.262 East Base 0.0024 108.26 0.226 0. 0045 [ Figure 3 ] 0. 0040 0. 0035 p w 0. 0030 e t 0. 0025 0. 0020 0. 0015 East Top E l ast S op East Base West Top W l est S op West Base hi l l 0. 0045 [ Figure 4 ] 0. 0040 0. 0035 p w 0. 0030 e t 0. 0025 0. 0020 0. 0015 East West f ace [ Table 4 ] hill=Top t Test The TTEST Procedure Statistics Lower CL Upper CL Lower CL Upper CL Variable face N Mean Mean Mean Std Dev Std Dev Std Dev Std Err pwet East 15 0.0024 0.0028 0.0032 0.0005 0.0007 0.0011 0.0002 pwet West 15 0.0023 0.0026 0.0029 0.0004 0.0005 0.0009 0.0001 pwet Diff (1-2) -23E-5 0.0002 0.0007 0.0005 0.0006 0.0009 0.0002 T-Tests Variable Method Variances DF t Value Pr > |t| pwet Pooled Equal 28 1.05 0.3049 pwet Satterthwaite Unequal 25.8 1.05 0.3056 Equality of Variances Variable Method Num DF Den DF F Value Pr > F pwet Folded F 14 14 1.82 0.2763 [ Table 5 ] hill=Slope The TTEST Procedure Statistics Lower CL Upper CL Lower CL Upper CL Variable face N Mean Mean Mean Std Dev Std Dev Std Dev Std Err pwet East 15 0.0023 0.0026 0.003 0.0005 0.0007 0.0011 0.0002 pwet West 15 0.002 0.0024 0.0028 0.0005 0.0007 0.0011 0.0002 pwet Diff (1-2) -27E-5 0.0002 0.0008 0.0005 0.0007 0.0009 0.0002 T-Tests Variable Method Variances DF t Value Pr > |t| pwet Pooled Equal 28 0.98 0.3355 pwet Satterthwaite Unequal 28 0.98 0.3355 Equality of Variances Variable Method Num DF Den DF F Value Pr > F pwet Folded F 14 14 1.02 0.9773 [ Table 6 ] hill=Base The TTEST Procedure Statistics Lower CL Upper CL Lower CL Upper CL Variable face N Mean Mean Mean Std Dev Std Dev Std Dev Std Err pwet East 15 0.0022 0.0026 0.003 0.0005 0.0007 0.0011 0.0002 pwet West 15 0.0019 0.0021 0.0024 0.0003 0.0004 0.0007 0.0001 pwet Diff (1-2) 487E-7 0.0005 0.0009 0.0005 0.0006 0.0008 0.0002 T-Tests Variable Method Variances DF t Value Pr > |t| pwet Pooled Equal 28 2.27 0.0309 pwet Satterthwaite Unequal 23 2.27 0.0327 Equality of Variances Variable Method Num DF Den DF F Value Pr > F pwet Folded F 14 14 2.75 0.0684 [ Table 7 ] East and West facing mean comparison The TTEST Procedure Statistics Lower CL Upper CL Lower CL Upper CL Variable/face N Mean Mean Mean Std Dev Std Dev Std Dev Std Err pwet East 45 0.0025 0.0027 0.0029 0.0006 0.0007 0.0009 0.0001 pwet West 45 0.0022 0.0024 0.0025 0.0005 0.0006 0.0007 0.0001 pwet Diff (1-2) 586E-7 0.0003 0.0006 0.0006 0.0006 0.0008 0.0001 T-Tests Variable Method Variances DF t Value Pr > |t| pwet Pooled Equal 88 2.42 0.0176 pwet Satterthwaite Unequal 85 2.42 0.0176 Equality of Variances Variable Method Num DF Den DF F Value Pr > F pwet Folded F 44 44 1.46 0.2148 [ Table 8 ] The ANOVA Procedure Class Level Information Class Levels Values hill 6 EastBase EastSlop EastTop WestBase WestSlop WestTop Number of Observations Read 90 Number of Observations Used 90 Dependent Variable: pwet Sum of Source DF Squares Mean Square F Value Pr > F Model 5 0.00000425 0.00000085 2.08 0.0765 Error 84 0.00003438 0.00000041 Corrected Total 89 0.00003863 R-Square Coeff Var Root MSE pwet Mean 0.109947 25.22755 0.000640 0.002536 Source DF Anova SS Mean Square F Value Pr > F hill 5 4.2470571E-6 8.4941141E-7 2.08 0.0765 Bonferroni (Dunn) t Tests for pwet NOTE: This test controls the Type I experimentwise error rate, but it generally has a higher Type II error rate than REGWQ. Alpha 0.05 Error Degrees of Freedom 84 Error Mean Square 4.093E-7 Critical Value of t 3.02146 Minimum Significant Difference 0.0007 Means with the same letter are not significantly different. Bon Grouping Mean N hill A 0.0028267 15 EastTop A A 0.0026373 15 EastSlope A A 0.0026350 15 EastBase A A 0.0025823 15 WestTop A A 0.0023937 15 WestSlope A A 0.0021408 15 WestBase data soilwet; input face$ hill$ pwet mass change; datalines; East Top 0.003326653 67.64 0.23561704 East Top 0.003530009 69.11 0.245514266 East Top 0.002439958 90.74 0.245675608 East Top 0.001667853 128.5 0.234429443 East Top 0.003656366 65.13 0.253909752 East Top 0.001748941 128.88 0.241568206 East Top 0.002292178 100.54 0.233721575 East Top 0.003076378 76.13 0.251731017 East Top 0.004159479 62.06 0.233064849 East Top 0.002924332 78.39 0.21860855 East Top 0.001887153 117.58 0.217540627 East Top 0.002691301 88.67 0.24721897 East Top 0.002953729 72.3 0.247492063 East Top 0.003328809 70.34 0.254376267 East Top 0.002717255 93.51 0.23922442 East Slope 0.002151702 112.66 0.23313687 East Slope 0.003354811 68.54 0.234484516 East Slope 0.001972497 130.69 0.234204947 East Slope 0.003085007 73 0.233414044 East Slope 0.001647328 68.59 0.240440971 East Slope 0.002360815 116.06 0.277039395 East Slope 0.002508084 98.35 0.232733697 East Slope 0.002941283 82.44 0.246708156 East Slope 0.001704792 141.43 0.237588652 East Slope 0.002928886 86.55 0.229462664 East Slope 0.00253109 91.39 0.244010321 East Slope 0.003082539 78.86 0.231337047 East Slope 0.002530549 90.64 0.234611638 East Slope 0.002486431 97.61 0.248776719 East Slope 0.004273845 80.55 0.252102919 East Base 0.001975571 128.95 0.245783133 East Base 0.00367964 67.63 0.224643125 East Base 0.003345444 77.3 0.207716832 East Base 0.002767152 105.21 0.211942029 East Base 0.001857147 128.53 0.207188 East Base 0.001752398 135.68 0.228424532 East Base 0.00361834 70.02 0.212076986 East Base 0.002788788 92.4 0.213785558 East Base 0.003611229 73.54 0.210297667 East Base 0.002240478 104.55 0.21773942 East Base 0.002598022 97.05 0.236685774 East Base 0.001570092 151.8 0.21938357 East Base 0.002119539 113.47 0.236821936 East Base 0.003185334 86.53 0.221803093 East Base 0.002415482 108.26 0.225636463 West Top 0.001724616 136.62 0.225014784 West Top 0.002407239 101.99 0.243958906 West Top 0.003079413 79.78 0.221401807 West Top 0.002552585 91.84 0.214319066 West Top 0.002529234 100.39 0.238139106 West Top 0.002594718 93.1 0.225403476 West Top 0.002592585 90.15 0.23045554 West Top 0.002954243 85.21 0.23420465 West Top 0.0024067 96.84 0.258137286 West Top 0.001832427 119.3 0.229238423 West Top 0.001822254 119.38 0.221891478 West Top 0.003618545 68.32 0.238637645 West Top 0.003142756 78.75 0.213554633 West Top 0.002343618 108.54 0.234148422 West Top 0.003134081 76.33 0.254090472 West Slope 0.003053528 76.35 0.242410794 West Slope 0.001494103 156.94 0.229938722 West Slope 0.003310317 70.75 0.2577856 West Slope 0.002825836 82.6 0.225205479 West Slope 0.001677651 143.32 0.112990232 West Slope 0.00220485 125.65 0.273996209 West Slope 0.003738093 62.26 0.246670056 West Slope 0.001616169 152.65 0.242479379 West Slope 0.002632853 90.24 0.241108676 West Slope 0.002451262 93.61 0.25349509 West Slope 0.002998038 81.39 0.231316337 West Slope 0.001610982 143.6 0.243089019 West Slope 0.002077864 112.91 0.229368932 West Slope 0.002135606 116.49 0.242700543 West Slope 0.002079028 121.26 0.344258225 West Base 0.002277879 107.9 0.254749903 West Base 0.001687777 133.1 0.248854059 West Base 0.001698004 122.33 0.258602846 West Base 0.002457299 86.25 0.291132022 West Base 0.001577614 131.33 0.238699136 West Base 0.002796236 81.69 0.23776533 West Base 0.001934479 109.63 0.253356184 West Base 0.00233901 91.4 0.257683983 West Base 0.001691856 124.3 0.265569758 West Base 0.00226411 96.17 0.234241989 West Base 0.002466762 95.95 0.252138073 West Base 0.002571002 85.33 0.238339921 West Base 0.002016707 117.43 0.240504098 West Base 0.002788924 79.53 0.27562695 West Base 0.001544186 146.12 0.261500092 ; run; proc sort data = soilwet ; by hill; run ; proc ttest data = soilwet; class face; var pwet; by hill; run; proc ttest data=soilwet; class face; var pwet; run; proc reg data = soilwet; model pwet = mass; plot pwet * mass / symbol = '.'; run; proc reg data = soilwet; model pwet = change; plot pwet * change / symbol = '.'; run; proc means data=soilwet n mean median std clm alpha = .05 var; class hill; var pwet; run; proc sort data=soilwet; by hill*face; run; proc boxplot data=soilwet; plot pwet*hill / boxstyle=schematic boxwidth=10; run; proc boxplot data = soilwet; plot pwet*face / boxstyle=schematic boxwidth=10; run; proc sort data = soilwet ; by hill ; run ; proc means data = soilwet ; by hill ; var pwet ; run ; proc anova data = soilwet ; class hill ; model pwet = hill ; run ; proc anova data = soilwet ; class hill ; model pwet = hill ; means hill / bon alpha = .05 ; run ;