GRID NAMES: AGE and AGENO

AGE/AGENO represents the age of the dominant vegetation, as derived from a variety of sources, and can be interpreted as the time since the last stand-replacing disturbance (or time since stand origin). Note: some cover types do not have ages because they only have one seral stage (CONDITION) and are assigned an age value of 99999. This includes the COVER types road, barren, water, mountain grassland, riparian, greasewood, sparse pinyon-juniper, agriculture and urban. AGE includes roads while AGENO does not.

REQUIREMENT: Required

DERIVED FROM:

Using the size class information in the UNC_CVU_TM coverage supplied by Region 2, USDAFS with the COVER, COVERNO, and RANDOM grids and based on Bill Romme and Amanda Clement's "Age and Percent Estimates for Low Elevation Cover Types on the Uncompahgre Plateau" document (low elevation types) and Kevin McGarigal's estimates for age ranges based on CVU size classes for each COVER type (high elevation), we created an AGE and AGENO grid. The RMACT_A_UNC and GMUG_FIRES02 coverages supplied by Region 2, USDAFS were used to mask out any age value since these areas have been altered

PROCESS USED TO CREATE:

The process of creating the AGE grid was somewhat complicated by the fact that age data did not exist for all vegetation polygons and was strongly biased toward the high elevation forested cover types. In particular, no age data existed for the buffer area surrounding the core project area (i.e., the extent of the CVU data), so we were forced to make assumptions about what ages should correspond to which cover/size class combinations. Consequently, we used a randomization process to not only assign size classes to patches outside of the CVU area that had size class information (i.e., Forest Service land) but also to then randomly assign an age value within a predefined age range. We used relative proportions of cover types and size classes derived from the CVU coverage for the FS land (where available) and replicated these proportions for all of the non-FS land. Essentially, the many patches (polygons) with age data were used to interpolate and/or extrapolate age to the remaining project and buffer areas without age data based on similarities in cover and size class and distance from patches with known ages.

The UNC_CVU_TM coverage has size class values for some polygons as well as class names for most of the polygons. The class names were used to crosswalk this coverage to create the COVER grid as is discussed below. We used this information to then correlate these cover classes to the size classes. This enabled us to create a spreadsheet of cover types, size classes, area, and percent of cover/size combinations. These percentages became our target for later steps.

We tried to fit the age information from UNC_SE_POINTS into this process but the results were meaningless because the spatial accuracy of these points led to contradictory conclusions. However, it is important to note the process that was used since it was instrumental in making the decisions that we did. After consulting with staff from Region 2, USDAFS, we decided to use the "rmris_year_origin", "fsv_age" and "survey_year" items to extract age for 3,298 points. The process is to use the "rmris_year_origin" data where available (2,901 records) and subtract this number from 2003. Where the "rmris_year_origin" data is not available (397 records), calculate age equal "fsv_age" plus the difference between 2003 and the "survey_year". In order to decide how well the UNC_SE_POINTS cover the various cover types and size classes, we took the UNC_SE_POINTS coverage and the UNC_CVU_TM grid and produced a series of grids and coverages to assess what the relative proportions of cover and size classes. A text file is produced (covpoint.txt) that is used to assess these relationships so that this information can be used in other age processes. The text file is brought into Excel. We used the covpoint.aml for this process which is similar to the SJNF aml agesize.aml. Covpoint.aml also produces a coverage called COVPOINT which has size class information in it.

The general process was to use the CVU-based size class attributes for the small area where this was available and randomly assign size classes based on real percentages for all the other areas for each high elevation COVER type. Once these were assigned, ages were randomly assigned to each COVER and size patch based on a predefined age range. The low elevation cover type ages were assigned based on the RANDOM grid to determine appropriate age range values (using the age and percent guidelines from Bill Romme and Amanda Clement's document). All of these grids were merged together and updated with the RMACT_A_UNC coverage to take into account recent management activities (AGE = 2) and with the GMUG_FIRES02 coverage for fires in 2002 (AGE = 1).

Therefore, there are two pathways that were utilized. The first pathway is for the high elevation cover types utilizing the CVU size class information and extrapolated to all high elevation cover types throughout the UPL. The second pathway is for the low elevation cover types.

HIGH ELEVATION PATHWAY (PPO, PPOA, WMC, WMCA, CMC, CMCA, ASP, SF, SFA, MS)

All of the processes in this paragraph are combined into the covsizebuf1.aml and covsizebuf2.aml. For most of the high elevation processes, there are two distinct tasks. This is due to the fact that cover types that are not in the CVU core area (which translates to mean not in the Forest Service area) do not have any size class information (this is different from the SJNF where every polygon within the core area had size class information). Therefore, these two areas needed to be processed differently. In order to assign a size class (1 - large, 2 - medium, or 3 - small) for these cover types, we first selected each cover type from the COVERNO grid for just the non FS area or the FS area that was lacking size class information. These cells were regiongrouped (with the 8 neighbor rule) and we then used the RANDOMINT grid to assign a random number to each region (or cluster) using the zonalmajority function. However, since zonalmajority has a software limit of around 20,000 zones and some of our cover types exceeded this limitation, they were split into smaller sections so that they could be processed and were then merged back together. A series of amls were run iteratively to fine tune the break points so that we could assign a size class to these cover types based upon the proportion found in the CVU core area. For example, we wanted to approximate the 82.5%/13.1%/4.5% large/medium/small ratio found in ponderosa pine - oak. By changing reclass values in the corresponding aml, we were able to get a ratio very close to this target. This was done with the amls z14.aml - z24.aml and they were run repeatedly outside of covsizebuf2.aml until we got the right numbers. Finally, we created a grid for each cover and size class combination called SIZEGRID which has the size class values for the entire project area, SIZE1GRID for just the FS area with size classes and SIZE2GRID which is SIZEGRID minus SIZE1GRID (necessary for final processing). The covsizebuf1.aml was a lot simpler since we were using real data (size classes) and didn't have to worry about meeting any cover/size proportions since these already existed. Covsizebuf1.aml also creates the grids ZZ14 - ZZ24 while covsizebuf2.aml creates grids ZM14 - ZM24 which are very important for later processes.

The next process was to take the RANDOMINT grid and merge with the ZM and ZZ grids mentioned above. This created grids called RANDOMZONE (1 and 2) that correspond to the 2 processes discussed above. The reason for these grids is that we wanted to assign patch-based ages based on the COVER and size class patches (8 neighbor-based patches). This was a way to insure that each COVER/size patch got the same age value since this was done randomly in the next step. The aml called randomzone.aml was used for this.

For cover and size patches within the CVU area with size information, we ran highage1.aml. This identifies the 30 COVER/size combinations (10 x 3) and assigns an age number using the RANDOMZONE1 grid. Formulas were developed that insured that age values were assigned randomly but covered the full range of predefined age ranges. Highage2.aml is more complicated because even though it assigns age values randomly (based on a random number between 0 and 10,000), it had to use the random value break points that were determined after running the z14.aml - z24.aml processes. Therefore, there are different formulas for each COVER/size combination. For example,where the COVER type is PPO (14) and the size class is medium (2), the formula is:

      age = ((randomzone2 - 5501) x 189/1849 + 60

            randomzone2 is the random number

            5501 is the lowest random number that is found in size class 2

            189 is the age range difference for this COVER/size class

            1849 is the random number difference for this COVER/size class

            60 is the lowest age within the age range for this COVER/size class

The output grids from these two processes are merged together in the highage3.aml. This creates the AGEHIGH grid.

LOW ELEVATION PATHWAY (SDG, SDS, PJ, PJS, PJOS, MTS)

Each low elevation cover type has its own aml. This was done because they needed to be run multiple times to adjust random numbers to try to match the percentages specified by Bill Romme and Amanda Clement. Once these CONDITION targets were met, age values were assigned randomly within each broad CONDITION class. The following amls were run:

      agesdgupl.aml       for Semi-Desert Grassland

      agesdsupl.aml       for Semi-Desert Savannah

      agepjwupl.aml      for Pinyon-Juniper Woodland (the %s are off due to many cells with one value because of the regiongrouping process)

      agepjsupl.aml       for Pinyon-Juniper Sagebrush Woodland

      agepjoupl.aml       for Pinyon-Juniper Oak Serviceberry Woodland

      agemtsupl.aml      for Mountain Shrubland

These amls produced six grids (SDGAGE, SDSAGE, PJWAGE, PJSAGE, PJOAGE and MTSAGE), without the effect of roads. Later on, roads are patched into the AGE grid but left out of the AGENO grid. They were all merged together to form the AGELOW grid with the lowmergeupl.aml.

FINAL PROCESSING

Using the agefinalupl.aml, the high and low elevation grids are merged together. To insure compatibility with the COVERNO grid, an age of 99999 was given to the road, barren, water, mountain grassland, riparian, greasewood, sparse pinyon-juniper, agriculture, and urban types (this grid is called AGENOPRE). The next step merges in the CROADS grid that has been reclassified as value 1 to create an age grid with the effect of roads called AGEPRE. In order to integrate recent activities and fires, the RMACT_A_UNC and GMUG_FIRES02 coverages were used to mask out any age value since these areas have been altered. Any cell that is covered by one of these areas, has the age changed to 2 (RMACT_A_UNC) or 1 (GMUG_FIRES02). This aml creates two grids - AGE and AGENO.

ATTRIBUTES:

      Description                                                           Value

age of vegetated, seral COVER types0 to 99998

non-vegetated or non-seral COVER types99999

RMLANDS SIGNIFICANCE:

AGE effects succession probabilities and various anthropogenic disturbance processes.

AMLS USED:

UPL: : covpoint.aml,covsizebuf1.aml, covsizebuf2.aml, randomzone.aml, highage1.aml, highage2.aml, highage3.aml, agesdgupl.aml, agesdsupl.aml, agepjwupl.aml, agepjsupl.aml, agepjoupl.aml, agemtsupl.aml, lowmergeupl.aml, and agefinalupl.aml

NOTES:

The 2002 fires were incorporated into the UPL process from the beginning.