PDF
Soil Volcanic Ash and Douglas-Fir (Pseudotsuga menziesii [Mirb.] Franco var. glauca) Productivity in North and Central Idaho Item Info
![Volcanic ash distribution, thickness, and its role in Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var. glauca) site index (SI) were determined for a forested region of north central Idaho. Ash thickness (AT) and Douglas-fir SI measurements were collected from local Natural Resource Conservation Service soil surveys and through newly installed field plots. Climatic, ecological, edaphic, geologic, and topographic data were collected from field observations or derived from digital elevation models. Three statistical models were developed to estimate AT and two models for predicting Douglas-fir SI. The modelled relationships were spatially displayed using a geographicl information system (GIS). Statistical model 1 was developed to estimate AT using ecologically based plant associations as an explanatory variable. Climax plant associations moister than grand fir-queen cup beadlily (Abies grandis/Clintonia uniflora) had consistently thick ash mantles and soils that would be classified as either Andisols or andic subgroups using Soil Taxonomy. Multiple linear regression (MLR) results showed that topographic factors accounted for ~ 29 percent of the explained variance, with elevation accounting for ~ 18 percent alone. Plant associations accounted for ~ 71 percent of the explained variation in AT. The statistical model error of 10.7 cm was significantly lower than the >20 cm variation often found in local soil series. Overall model fit produced an R2 of 0.6. Results suggest that plant associations integrate many of the local factors that influence volcanic ash distribution. Statistical model 2 was developed to determine AT for situations when climax plant associations are not physically observable. This model was based solely upon topographic factors. Predictive models were developed using MLR and geographically weighted regression (GWR). Results show that elevation, slope, plan curvature, and the wetness index all significantly influence volcanic ash distribution. Similar to model 1, elevation explained the most variation in AT. GWR improved model fit and precision by 36 percent and 30 percent, respectively, over the MLR model (R2A = 0.64). The localized approach of GWR modelling showed that elevation, slope curvature, and the wetness index behaved differently depending on geographic location. This suggests that the global approach of MLR modelling is masking conditional landscape relationships that are unique and highly localized. Volcanic ash and other climatic, edaphic, geologic, and topographic factors were used to predict Douglas-fir SI using both MLR and GWR in statistical model 3. Elevation, ash depth, slope, and aspect were significantly correlated with Douglas-fir SI. Elevation was found to be nonstationary, indicating that parameter estimates associated with elevation significantly change depending on the geographic location of an observation. SI showed a positive logarithmic response to increasing AT. GWR significantly improved MLR model fit by 28 percent and reduced the sum of square errors by 54 percent (R2A = 0.5). Residuals analysis of MLR SI estimates showed extreme SI overestimation in the south and west portion of our study area. This overestimation was significantly reduced in the GWR model. The significant fit and precision improvements of GWR models suggest that the complex interactions affecting Douglas-fir SI are better analyzed at the local level.](https://objects.lib.uidaho.edu/iftnc/small/iftnc3878_sm.jpg)
- Title:
- Soil Volcanic Ash and Douglas-Fir (Pseudotsuga menziesii [Mirb.] Franco var. glauca) Productivity in North and Central Idaho
- Creator:
- Kimsey, Jr., M.J.
- Date Created (ISO Standard):
- 2006-05-01
- Description:
- Volcanic ash distribution, thickness, and its role in Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var. glauca) site index (SI) were determined for a forested region of north central Idaho. Ash thickness (AT) and Douglas-fir SI measurements were collected from local Natural Resource Conservation Service soil surveys and through newly installed field plots. Climatic, ecological, edaphic, geologic, and topographic data were collected from field observations or derived from digital elevation models. Three statistical models were developed to estimate AT and two models for predicting Douglas-fir SI. The modelled relationships were spatially displayed using a geographicl information system (GIS). Statistical model 1 was developed to estimate AT using ecologically based plant associations as an explanatory variable. Climax plant associations moister than grand fir-queen cup beadlily (Abies grandis/Clintonia uniflora) had consistently thick ash mantles and soils that would be classified as either Andisols or andic subgroups using Soil Taxonomy. Multiple linear regression (MLR) results showed that topographic factors accounted for ~ 29 percent of the explained variance, with elevation accounting for ~ 18 percent alone. Plant associations accounted for ~ 71 percent of the explained variation in AT. The statistical model error of 10.7 cm was significantly lower than the >20 cm variation often found in local soil series. Overall model fit produced an R2 of 0.6. Results suggest that plant associations integrate many of the local factors that influence volcanic ash distribution. Statistical model 2 was developed to determine AT for situations when climax plant associations are not physically observable. This model was based solely upon topographic factors. Predictive models were developed using MLR and geographically weighted regression (GWR). Results show that elevation, slope, plan curvature, and the wetness index all significantly influence volcanic ash distribution. Similar to model 1, elevation explained the most variation in AT. GWR improved model fit and precision by 36 percent and 30 percent, respectively, over the MLR model (R2A = 0.64). The localized approach of GWR modelling showed that elevation, slope curvature, and the wetness index behaved differently depending on geographic location. This suggests that the global approach of MLR modelling is masking conditional landscape relationships that are unique and highly localized. Volcanic ash and other climatic, edaphic, geologic, and topographic factors were used to predict Douglas-fir SI using both MLR and GWR in statistical model 3. Elevation, ash depth, slope, and aspect were significantly correlated with Douglas-fir SI. Elevation was found to be nonstationary, indicating that parameter estimates associated with elevation significantly change depending on the geographic location of an observation. SI showed a positive logarithmic response to increasing AT. GWR significantly improved MLR model fit by 28 percent and reduced the sum of square errors by 54 percent (R2A = 0.5). Residuals analysis of MLR SI estimates showed extreme SI overestimation in the south and west portion of our study area. This overestimation was significantly reduced in the GWR model. The significant fit and precision improvements of GWR models suggest that the complex interactions affecting Douglas-fir SI are better analyzed at the local level.
- Subjects:
- research soil science data modeling statistics
- Location:
- North and Central Idaho Eastern Washington Western Montana Northeastern Oregon
- Source:
- Kimsey, M.J. Doctorial Dissertation.2006. Soil Volcanic Ash and Douglas-Fir (Pseudotsuga mensiesii [Mirb.] Franco var glauca)Productivity in North Central Idaho. University of Idaho.
- Source Identifier:
- Soil_Volcanic_Ash_and_DF_Productivity_in_NC_Idaho_UIGradStudies_2006
- Type:
- Text
- Format:
- application/pdf
Source
- Preferred Citation:
- "Soil Volcanic Ash and Douglas-Fir (Pseudotsuga menziesii [Mirb.] Franco var. glauca) Productivity in North and Central Idaho", Intermountain Forestry Cooperative, University of Idaho Library Digital Collections, https://www.lib.uidaho.edu/digital/iftnc/items/iftnc3878.html