Вы находитесь на странице: 1из 222

Analyzing Pebble Count Data Collected By Size Classes

Prepared by John Potyondy and Kristin Bunte Version 1.0a February 6, 2002 (Based upon a similar spreadsheet developed by Greg Bevenger and Rudy King)

Introduction
We have developed this workbook to assist you with the statistical analysis of Wolman pebble count data tallied by size class. Sampling should be geomorphically stratified based on the natural sorting of grain sizes into distinct channel features to sample homogeneous populations from the river bed. The intermediate axis of individual particles may be measured with a ruler or by using a gravel template (gravel-o-meter). We strongly recommend the use of templates because they avoid incorrectly identifying the intermediate axis and have been shown to reduce measurement errors among observers. Classifying particle sizes on the basis of square openings in the template has the further advantage of providing a measure of size that is compatible with To discourage serial correlation in the sample, the distance between successively sampled particles should be chosen so that successive particles are at least several grain diameters apart. This can typically be achieved by setting the spacing between sampled particles to a distance larger than the intermediate axis of the largest particles in the reach or the population being sampled. When conducting a pebble count, we recommend sampling geomorphic features or habitat units (riffles, pools, or combinations of pools and riffles) depending on study objectives. For example, pebble counting done to monitor changes in the particle-size distribution of a habitat feature of biological significance usually collect particles from riffles alone. This approach represents a spatially segregated, stratified sample that provides a size distribution for a specific geomorphic feature or habitat unit. Pebble counts done for general stream classification usually collect particles from an entire stream reach. Rosgen suggests sampling a reach two meander wavelengths long (four riffle-pool sequences). Transects are placed such that the percent stream length occupied by riffles (incl. rapids and runs) and by pools matches the percent of the transect located in riffles and pools (e.g., if riffles occupy 70% of the stream length, 70 % of all pebbles are collected on riffles). Since particles were collected on a spatially proportional basis, all data may be combined to obtain a reach-averaged particle-size distribution. Worksheets are provided to let you 1) estimate the required sample size, 2) enter data, 3) perform statistical analyses and produce graphs, and 4) take notes. Also provided are case studies that may assist you in developing your study plan and interpreting your analysis. Sample-size estimation and statistical analysis are explained in greater detail and were derived from: Bevenger, Gregory S. and Rudy M. King. 1995. A Pebble Count Procedure for Assessing Watershed Cumulative Effects. Rocky Mountain Forest and Range Experiment Station Research Paper RM-RP-319, 17 pages. We encourage you to become thoroughly familiar with the statistical discussions in RM-RP-319 before proceeding. Additional information about pebble counts is also available in Bunte, Kristin and Steven R. Abt. 2001. Sampling Surface and Subsurface Particle-Size Distributions in Wadeable Gravel- and Cobble-Bed Streams for Analyses in Sediment Transport, Hydraulics, and Streambed Monitoring. Rocky Mountain Research Station RMRS-GTR-74, 428 pages. You may order these documents from the Rocky Mountain Research Station, 240 West Prospect Road, Fort Collins, Colorado 80526.

How to Use this Workbook


Step 1.

Before initiating field work, you should ask "How many pebbles need to be counted to detect a real difference between reference and study conditions?" To answer this question, the nature of the reference condition, the size of change that might occur from a management action, the biological or physical significance (if any) of the induced change, and acceptable levels of risk must be specified. The necessary information may be obtained during a reconnaissance field trip and/or from information available in the office. The Sample-Size Worksheet is When in the field, the question to ask is, Where exactly should the pebble count(s) be done? Assume, for example, the geomorphic feature or habitat units to be sampled are riffles. Depending on the homogeneity of riffle surfaces encountered in the reach you might: (A) Find that all riffles in the reach look pretty much the same. In this case, select one riffle that is most representative and do a pebble count of the specified sample size; (B) If one riffle is too small for collecting the required sample size, or if one riffle alone does not cover the slight variability encountered between the riffles of the reach, select several riffles. They need to be similar enough, however, to characterize one population. Do one pebble count and collect particles from all riffles, distributing the specified sample size equally between the riffles; (C) If individual riffles depart in their appearance from each other more than you consider appropriate for members of one population, do an individual pebble count of the specified sample size on each of the selected riffles. In this complex situation you may need to employ a statistical test to evaluate whether the degree of statistical dissimilarity among the riffles is acceptable and perhaps re-evaluate your entire sampling strategy.

Step 2.
After you have completed your reference and study reach field work enter your pebble count data into the Data Input Worksheet. Direction on how to enter your data is provided in that worksheet.

Step 3.
Navigate to the Analysis Worksheet to see your output. Guidance on data interpretation is provided in that worksheet. If necessary, review the case studies for further thoughts on data interpretation. Add notes to the

Step 4.
If you need printed copies of your analysis, highlight the appropriate area, then select File, Print Area, Set Print Area. Next, select Print Preview and make adjustments accordingly (for example, switch from portrait to landscape view or add headers and footers). Once you are satisified with the print preview, select Print. Then, if necessary, save your work for future reference by selecting File, Save As and specifying a new file name.

Spreadsheet Assistance:
If you have problems using this spreadsheet contact: John Potyondy, Rocky Mountain Research Station, Stream Systems Technology Center (970) 295-5986;

Before initiating fieldwork, you should ask, "How many particles need to be counted to detect a real difference between percent fines in the reference (control or unimpacted) and the study (impacted) reach?" To answer this question, the nature of the reference condition, the size of change that might occur from a management action, and acceptable levels of risk must be specified. The potential risk(s) or cost(s) associated with possible testing errors should be considered. Type I error is the risk of falsely concluding that there is a difference between reference and study conditoins. Type II error is the risk of falsely concluding that there is no difference between reference and study conditions. In typical situations, Type II error is allowed to be larger than Type I error, but the reverse might be the case if the situation is highly sensitive. Typical values are Type I error = 0.05 and Type II error = 0.20. If reference and study sizes will be equal, then set the sample size factor to 1. This will be the situation in most applications using stratified sampling, e.g., riffles as habitat units. If the reference sample size will be greater than the study sample size, then the sample size factor should be less than 1. For instance, if double the number of reference samples will be taken, the sample size factor should be set to 0.5. Conversely, if double the number of study samples will be taken, then the sample size factor should be set to 2. Reference and study proportions of fines should be set to values expected to occur. The critical element is that the difference between reference and study proportions should reflect a "real difference" usually interpreted as one that has some biological significance. The hypothesis tested in this framework is that the study proportion is larger than the reference proportion. For example, suppose you have data indicating that the proportion of particles <8 mm for unimpacted riffles is about 10%. Based on this you would set the reference proportion to 0.10. Further assume that an increase in the proportion of particles <8 mm to 25% in an impacted study reach is considered biologically significant. Based on this you would set the study proportion to 0.25. To illustrate this example, if Type I error is specified as 0.05, Type II error is specified as 0.2, the number of particles collected in the study reach is the same as the number in the reference reach so the factor relating study to reference sample size is 1, reference and study proportions are 0.10 and 0.25 respectively, the estimated sample size for both the reference and study sample size is 92 particles each. It is generally a good idea to collect more samples than indicated because input values are frequently based on estimated values. In this To estimate the sample size you need, enter the respective values in the white cells. Your estimated sample sizes will appear automatically in the yellow cells.

Data Input
Enter size class pebble count data in this
Particle Size Class (mm)
<2 2 - 2.8 2.8 - 4 4 - 5.6 5.6 - 8 8 - 11.3 11.3 - 16 16 - 22.6 22.6 - 32 32 - 45.3 45.3 - 64 64 - 90.5 90.5 - 128 128 - 181 181 - 256 256 - 362 362 - 512 512 - 1024 1024 - 2048 2048 - 4096 >4096 Totals 101

Class Name
Sand VF Gravel VF Gravel Fine Gravel Fine Gravel Med. Gravel Med. Gravel Coarse Gravel Coarse Gravel VC Gravel VC Gravel Sm. Cobble Sm. Cobble Lg. Cobble Lg. Cobble Sm. Boulder Sm. Boulder Med. Boulder Lg. Boulder VL Boulder Bedrock

Reference Total
0 1 1 0 1 3 3 10 15 11 16 19 13 8 0 0 0 0 0 0 0

Study Total
0 4 6 2 2 5 6 6 8 10 15 10 13 8 0 0 0 0 0 0 0 95

Reference Cumulative %
0.0 1.0 2.0 2.0 3.0 5.9 8.9 18.8 33.7 44.6 60.4 79.2 92.1 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Study Cumulative %
0.0 4.2 10.5 12.6 14.7 20.0 26.3 32.6 41.1 51.6 67.4 77.9 91.6 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

ta Input
count data in this worksheet.

STEPS
1. Clear BOTH the reference and study reach data columns by pressing Ctrl-b; OR clear ONLY the reference reach data column by pressing Ctrl-r; OR clear ONLY the study reach data column by pressing Ctrl-s. 2. Enter or paste data in the white columns; output appears in yellow cells on other sheets. 3. Navigate to the Analysis worksheet to view the output. Note:
The data entry table is set up for 1/2 phi-size intervals. If you have phi-size interval data, leave the intervening intervals blank.

Contingency Tables
The following is a summary of the data you entered in the Data Input Worksheet. Summary data are presented as (1) total number of pebbles counted that were less than and greater than the designated particle size criterion and (2) percentage of pebbles counted that were less than the designated particle size criterion. Also presented is a p-value for each particle size criterion. A small p-value indicates the proportion of particles less than the criterion is statistically different between your reference and study reaches. For example, p-values <0.10 indicate a significant difference at the 10% level of significance (90% confidence level); p-values <0.05 indicate a significant difference at the 5% level of significance (95% confidence level); and p-values <0.01 indicate a significant difference at the 1% level of significance (99% confidence level).

User-selected size criterion: <ctrl-u> 2mm


Reference Study Total Reference < 0.0% < 0 0 0 Study < 0.0% > or = 101 95 196 Average < 0.0% p-value Total 101 95 196 Average >= 100.0% 1.0000 Reference Study Total Reference < 2.0% < 2 10 12 Study < 10.5%

4 mm
> or = 99 85 184 Average < 6.1% p-value Total 101 95 196 Average >= 93.9% 0.0140 Reference Study Total Reference < 3.0% < 3 14 17 Study < 14.7%

8 mm
> or = 98 81 179 Average < 8.7% p-value Total 101 95 196 Average >= 91.3% 0.0038

Analysis Notes
Use this worksheet to document notes relative to your analysis. The rows are set to wrap text.

to wrap text.

Case Study 1 - Significant Difference between Sulphur Creek and its Reference

Case Study 2 - Inconclusive if there is a Difference between Sulphur Creek and its Referenc

Case Study 3 - No Significant Difference between Sulphur Creek and its Reference

Case Study Situation


A wildfire has occurred on tributaries of the Middle Fork Salmon River on the Boise National Forest. The lightening-ignited fire began on August 4 and gradually expanded to an area of 51,000 acres by late October. Because the fire burned within the boundaries of the Frank Church River of No Return Wilderness and mostly burned decadent stands of lodgepole pine, minimal suppression activities were undertaken before the fire Following the fire, the Forest Supervisor responding to public concern over post-fire increases in fine sediment and possible effects on salmon, decides to initiate a post-fire monitoring program of aquatic and other resources. Due to the remoteness of the field site, financial constraints, and the onset of winter, resource specialists will be provided with a one-time, one day helicopter access to monitoring sites. These measurements, made as soon as possible after the fire before any fire-generated sediment can enter the streams, will document pre-fire, reference The forest hydrologist and forest fisheries biologist quickly review fire maps and select two drainages for monitoring: Boundary Creek, a small 6.4 square mile watershed representing the most severely burned watershed within the fires perimeter, and Sulphur Creek, a larger 44 square mile watershed with fisheries resource value, representing low intensity, mosaic burn patterns characteristic of much of the burn. Pebble counts are selected as one tool to evaluate sediment impacts using a basic before-after design strategy. They Before they shut down their computers and head for the field, the hydrologist and the fisheries biologist meet with Forest Staff to discuss the monitoring plan. They decide, because of the sensitivity of the salmon fisheries issue and possible ramifications for the let-burn policy in other Wilderness areas, that they need a high degree of certainty in the results. They decide to set their Type I error at 0.05 and their Type II error at 0.20. They recognize that more sampling sites and other sampling designs would be desirable, but given the time and Relying on prior monitoring data from similar stream types in unimpacted conditions in the Idaho Batholith, the fish biologist expects a reference particle size distribution of the <8 mm size class in Sulphur Creek to be about 10% fines. He also believes that an increase of 15% to around 25% fines < 8 mm in size has undesirable effects on salmon. Using the Sample Size Worksheet, they decide to collect about 100 particles across riffles at the mouth of each watershed at this time and for three post-fire years. They grab their field gear and head to the woods. After thoroughly enjoying themselves on this outing and after returning to the sites the following three field seasons, they begin to analyze their data. Using the workbook, they first enter the pebble count data for the reference reach (pre-fire condition) and each year for the study reach (post-fire years). They navigate to the Analysis Worksheet and

Case Study Data Analysis and Interpretation


CAUTION: If you have already entered your own data in the Data Input Worksheet and you activate the macros discussed in the case studies below, your data will be replaced with the case study data. To temporarily store your data in a safe place press ctrl-m. Your data will appear on the Misc Worksheet. Navigate to the Case Study Situation Worksheet and review the case studies of interest. When you are finished reviewing the case study information press ctrl-n to re-run your data. After re-running your data you will see the Analysis Worksheet, with your data populating the cells.

Case Study 1 - Significant Difference between Sulphur Creek and its Reference
Press ctrl-e to run the data set for Case Study 1. When the macro is done running you will see the Analysis Worksheet. Observe the contingency table and the graphs. Then return to this Case Study Situation Worksheet

This data represents a comparison between the pre-fire reference condition and the first post-fire year. Contingency table p-values for all three particle size criterion are well less than 0.05 indicating there is a significant difference in fine sediment between Sulphur Creek and its reference site for the first year after the fire. The graphs provide visual confirmation of this difference.

Case Study 2 - Inconclusive if there is a Difference between Sulphur Creek and its Reference
Press ctrl-f to run the data set for Case Study 2. When the macro is done running you will see the Analysis Worksheet. Observe the contingency table and the graphs. Then return to this Case Study Situation Worksheet This data represents a comparison between the pre-fire reference condition and the second post-fire year. Contingency table p-values for two of the particle size criteria (<2 mm and <4 mm) are much greater than 0.05 indicating there is no significant difference in fine sediment in these size classes between Sulphur Creek and its reference. The contingency table values for the 8 mm size class is however much less than 0.05 indicating that fines less than 8 mm remain significantly different during the second post-fire year. Results are somewhat inconclusive. The smaller size classes appear to have returned to near pre-fire conditions, but the 8 mm sizes

Case Study 3 - No Significant Difference between Sulphur Creek and its Reference
Press ctrl-g to run the data set for Case Study 3. When the macro is done running you will see the Analysis Worksheet. Observe the contingency table and the graphs. Then return to this Case Study Situation Worksheet This data represents a comparison between the pre-fire reference condition and the third post-fire year. Contingency table p-values for all three particle size criterion are much greater than 0.05 indicating there is no significant difference in fine sediment between Sulphur Creek and its reference three years after the fire. The

Reference

and its Reference

s Reference

6 1 1 0 1 3 3 10 15 11 16 19 13 8 0 0 0 0 0 0 0

22 4 6 2 2 5 6 6 8 10 15 10 13 8 0 0 0 0 0 0 0

Вам также может понравиться