1 / 35

Rapid Carbon Assessment (RCA) Sampling Groups

Rapid Carbon Assessment (RCA) Sampling Groups. Sampling: Where and Why Part #1. Who am I?. “Statistician” and Soil Scientist Coordinated the Sampling Design Email me: skye.wills@lin.usda.gov This is the BEST way to get the BEST answer Call me: Blackberry 402-314-5659

dick
Download Presentation

Rapid Carbon Assessment (RCA) Sampling Groups

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Rapid Carbon Assessment (RCA) Sampling Groups Sampling: Where and Why Part #1

  2. Who am I? • “Statistician” and Soil Scientist • Coordinated the Sampling Design • Email me: skye.wills@lin.usda.gov • This is the BEST way to get the BEST answer • Call me: Blackberry 402-314-5659 • I don’t mind at all, but I might have to call you back with the BEST answer

  3. The objectives of this project are: • to develop a scientifically-based and statistically valid inventory of soil carbon stocksfor the conterminous U.S • to evaluate differences in soil carbon associated with differing soil properties, agricultural management systems, ecosystems, and land uses

  4. Logistics • Resources are allocated by MO region: • 1 person (time – wise, this may actually be more than one person) • 1 VNIR • Support from “local” soil scientists

  5. Philosophy Without extraordinary effort, soil carbon stocks cannot be evaluated for all soils, ecosystems, land uses, and agricultural management systems that occur in the U.S. Thus, reasonable groupings of soils, ecosystems, and management systems expected to result in similar soil carbon stocks will be developed. The soil series or map unit component is the most efficient means to stratify soil sampling and data collection since the map unit component incorporates soil properties, landscape characteristics, and climate. Other variables affecting soil carbon stocks including agricultural management systems, ecosystem types, and land uses, will be stratified within the soil groups.

  6. Stuff we decided: • Sampling will be done by MO • Samples will be selected to represent a range of soil carbon stocks • Use soil series characteristics to stratify samples

  7. Why I think you should care • So you will know: • why we’re asking you to do certain things • why groups do not line up with typical taxonomic or geographic groups • the importance of following the protocols we give you • I put a lot of trust in your professional judgment, I want you to be as informed as possible.

  8. Statistical Design • Goal: stratify by MO and soil properties • MO • Map units and their dominant component/series were extracted from each physical MO boundary using GIS (relational database would yield different results) • Soil properties • Group soil series using similarity index of soil properties so that samples are collected from soils with a range of soil carbon stocks • Randomly select points from within each MO and soil group

  9. Why sample randomly? • Unbiased, statistically defensible results

  10. Soil Series Properties: Similarity/Dissimilarity • Soil Correlation • Norfleet and Epinette, 2009 • used for component composition

  11. RCA Grouping Procedure • Dominant soil series components were compiled for all map units in each MO* • The properties for all soil series were queried from the official series description (OSeD) database maintained by the soil survey lab (SSL) • Scores were assigned to properties thought to be important to soil carbon stocks • Scores were used to perform a hierarchical cluster for each MO* • Soils were assigned to a group for sampling *MO1 used a custom human-developed clustering system. MO2 used component data from Soil DataMart and a hybrid clustering approach

  12. RCA similarity scores • Properties or characteristics • Taxonomy • Moisture Regime • Temperature Regime • Flooding (queried and compiled from NASIS series components) • Family Particle Size Class • Drainage Class • Depth to Restrictive layer

  13. RCA similarity scores • Score were assigned • Initially using SSL pedon carbon stocks • Modified by scientists on the standards, interpretations and research staffs at the NSSC • Reviewed by academic researchers, some modifications made • Reviewed by MO leaders and senior soil scientists; further revisions made

  14. Example: Moisture Regime • Moisture regime was queried from taxonomy

  15. Hierarchical Cluster • Statistical Algorithm that assigns clusters based on n-dimensional distance • So that observations within a cluster are more similar than observations between more than one cluster • Represented in a tree structure called a dendogram

  16. 2-D example • From: Comp449 Speech Recognition, Steve Cassidy, Macquire University, Sydney Australia, http://web.science.mq.edu.au/~cassidy/comp449/html/comp449.html • Figure 2.1. An example clustering of the first two columns…

  17. Hierarchical

  18. 40 Clusters 20 Clusters 10 Clusters

  19. Things to keep in mind: • This hierarchy works much differently that Soil Taxonomy – this really bothers some people • Clusters are hierarchical in that you can split and recombine groups – it does not mean that it is taxonomically hierarchical • The clustering algorithm uses n-dimensions in its calculations; this means all properties are used in the calculation. Two soils could have identical moisture regimes and drainage classes, but occur in different clusters because of other factors.

  20. Soil Taxonomy vs. Hierarchical Clustering HC: All properties/scores Soil taxonomy: Moisture Regime HC: All properties/scores Soil taxonomy: Epipedon thickness HC: All properties/scores HC: All properties/scores Soil taxonomy: Organic vs. Mineral

  21. Things to Remember • Each MO has its own groups • Soils that occur in multiple MOs (most of them) will be grouped separately for each MO • The group numbers are meaningless outside of an MO • The groups were chosen to get a range of soil carbon stock values • The data is not appropriate for producing spatially explicit information about soil carbon stocks (no detailed, mu-level maps) • The data only make sense when aggregated

  22. How will you use the RCA Sampling Groups? • RCA sample points are selected using the groups (next topic) • They will be used to verify/reject points • You’ll get a list of soils in each group as well as general properties of each group • They will be used to aggregate and analyze the data when the project is complete

  23. My final thoughts on RCA Sampling Groups • I am happy to answer any questions about the grouping methodology • But…… • I did the best I could………….. • And…… • The groups are done.

  24. Questions

  25. Rapid Carbon Assessment (RCA) Sampling Locations Sampling: Where and Why Part #2

  26. Sample Selection • Predetermined random points form the basis of our sample pool (NRI points) • From that pool, locations will be selected by MO, Group and Land Cover/Use

  27. Idealized Sample Locations Distribution

  28. Sample Allocation • All MO’s will have ~ 400 locations • With 5 pedons for most locations • Distribution between soil groups and land cover/use combinations base on acreage • Most Group/LC combinations will have 5 locations (25 pedons)

  29. Sample Allocation • X,Y and Z vary by MO

  30. Actual Sample Locations • RCA sample points – randomly selected from within group and lc/lu • Excessive list of points will be selected and distributed to DC’s other official to obtain access • Accessible point lists will be distributed to each MO coordinator • Each group will have its own lu/lc point lists

  31. Actual Sample Locations • RCA point lists – generally sample in order of random selection (in chunks), not convenience alone • If you a group/lc is allocated 6 locations, plan to visit the 1st 6 locations on the list • Supplement the list of locations with extra points when a location is rejected • Ideally select #7 on the list, this can be modified somewhat for logistical constraints

  32. MO5 – Group 10 / Rangeland – 5 samples Allocated

  33. MO5 – Group 10 / Rangeland – 5 samples Allocated

More Related