Forest Carbon Baseline for Wisconsin - Semantic Scholar

Baseline for forest lands of Wisconsin

Report Submitted to the Wisconsin Department of Natural Resources

Forest Carbon Baseline for Wisconsin By Sandra Brown, Sean Grimland, Tim Pearson and Nancy Harris

January 2008

Submitted by: Sandra Brown, Project Coordinator Ecosystem Services Unit 1621 N. Kent St, Suite 1200 Arlington, VA 22209

Winrock International

1


Table of Contents Table of Contents ...................................................................................................................................... 2 List of Figures ............................................................................................................................................ 3 List of Tables ............................................................................................................................................. 4 Executive summary....................................................................................................................................... 5 1. BACKGROUND ............................................................................................................................... 7 1.1. General Approach........................................................................................................................ 7 1.2. Datasets used in the analysis ...................................................................................................... 7 1.3. Geographical subdivision of the state.......................................................................................... 7 2. METHODS FOR BASELINE DEVELOPMENT ............................................................................... 8 2.1. General Approach........................................................................................................................ 8 2.2. Forest area classification ............................................................................................................. 9 2.3. Calculating change in forest area .............................................................................................. 12 2.4. Estimating forest carbon stocks................................................................................................. 12 2.4.1. Carbon stocks for forest loss................................................................................................. 13 2.4.2. Carbon stocks for forest gain ................................................................................................ 15 2.4.3. Carbon accumulation in forests remaining as forests ........................................................... 16 3. RESULTS....................................................................................................................................... 17 3.1 Forest extent in 1992 ..................................................................................................................... 17 3.2 Forest area gains and losses, 1992-2001...................................................................................... 18 3.2.1 Change in forest area at the state level ................................................................................ 18 3.2.2 Change in forest area at the county level.............................................................................. 19 3.3 Existing carbon stocks in 1992 ...................................................................................................... 23 3.4 CO2 emissions and removals, 1992-2001 ..................................................................................... 24 3.4.1 CO2 emissions and removals from forests at the state level................................................. 24 3.4.2 CO2 emissions and removals due to land use change at the county level ........................... 25 3.4.3 CO2 emissions and removals at the county level .................................................................. 25 3.5 Comparison with the USFS analysis for Wisconsin....................................................................... 31 4. CONCLUSIONS............................................................................................................................. 34 5. REFERENCES............................................................................................................................... 35


2


List of Figures Figure 1. Wisconsin counties. Source: Digital Map Store. http://county-map.digital-topomaps.com/wisconsin ..................................................................................................................................... 8 Figure 2. WISCLAND 1992 (left) and NLCD 2001 (right) land cover maps, reclassified into the same eight land cover categories .................................................................................................................................... 9 Figure 3. Forest (displayed in green, including woody wetlands) and non-forest (displayed in grey) for WISCLAND 1992 (left) and NLCD 2001 (right)........................................................................................... 12 Figure 4. The percentage of forest land cover for each county in 1992 ..................................................... 18 Figure 5. Change in forest cover in Wisconsin between the years 1992 and 2001. Forest gain is shown in green and forest loss is shown in red. No change refers to the persistence of different land cover classes over the time period. ................................................................................................................................... 18 Figure 6. Percentage of forest land cover lost in each county from 1992 to 2001 based on the difference in forest cover of the Wiscland (1992) and NLCD (2001) maps. ................................................................ 20 Figure 7. Percentage of forest land cover gained in each county from 1992 to 2001 ................................ 20 Figure 8. Net percent forest change in forest cover in Wisconsin by county from 1992 to 2001. A negative sign indicates a net loss in forest area and a positive sign indicates a net gain in forest area. ................. 21 Figure 9. Area-weighted average tons of CO2 per acre for each county (t CO2/acre)................................ 23 Figure 10. The total quantity of CO2 for each county in forests for 1992 (Mt CO2)..................................... 24 Figure 11. The annual estimated emission of CO2, in tons, for each county caused by loss of forest cover between 1992-2001. ................................................................................................................................... 27 Figure 12. The annual estimated removal of CO2 for each county that was caused by gain in forest cover between 1992-2001 in tons CO2. ................................................................................................................ 28 Figure 13. The annual estimated removal of tons CO2 for each county caused by forest remaining forests between 1992-2001. ................................................................................................................................... 28 Figure 14. The net change in total tons of CO2 in Wisconsin counties. Negative sign indicates a removal from the atmosphere and positive sign is an emission. .............................................................................. 29 Figure 15. Change in forest cover from 1990-2007 based on the FIA data base...................................... 33 Figure 16. Removals of CO2, in million t/yr, in live trees only (above and below ground) and in all live and dead organic matter pools (live trees plus dead wood and forest floor; does not include soil pool) from 1991 to 2007 (from Linda Heath, 2007, USFS, personal communication). ................................................ 34


3


List of Tables Table 1. 1992 WISCLAND reclassified land cover categories and their respective areas......................... 10 Table 2. 2001 NLCD reclassified land cover categories and their respective areas. ................................. 11 Table 3. Comparison of reclassified land cover categories. ....................................................................... 11 Table 4. FIA forest types cross-walked into Wisconsin forest types........................................................... 13 Table 5. Mean forest carbon stocks as derived from FIA Mapmaker and used for forest loss (adjusted to account for wood products as described above). ....................................................................................... 15 Table 6. Carbon stocks in live tree biomass after five years of growth for six forest types in Wisconsin... 16 Table 7. Forest type, mean carbon stock, age, interpolated annual growth rate........................................ 17 Table 8. Gross and net changes in land and forest cover (forest and woody wetland classes) between 1992 and 2001. ........................................................................................................................................... 19 Table 9. Land use change matrix for 1992 and 2001 land cover categories (woody wetland have been combined with forests into one class). ........................................................................................................ 19 Table 10. Area of forest gain and loss in based on the Wiscland 1992 and NLCD 2001 data bases. Net loss is indicated by a negative sign, net gain is indicated by a positive sign.............................................. 21 Table 11. Carbon emissions and removals in Wisconsin forests (from changes in the live tree biomass only). A positive sign indicates a CO2 emission (addition to the atmosphere)and a negative sign indicates a CO2 removal (subtraction from the atmosphere). .................................................................................... 25 Table 12. County summary of annual emissions and removals of CO2 from forest cover change. A positive sign indicates a CO2 emission (addition to the atmosphere) and a negative sign indicates a CO2 removal (subtraction from the atmosphere). ............................................................................................... 25 Table 13. Total CO2 sequestered in forests gained between 1992-2001 and extended to 2024. .............. 29 Table 14. Source of forest inventory and average year of field survey used to estimate Wisconsin-wide carbon stocks and changes in carbon stocks in live trees.......................................................................... 32


4


EXECUTIVE SUMMARY The main goal of work presented in this report was to establish the baseline emissions and/or removals of CO2 as measured by changes in carbon stocks for the forest sector in the state of Wisconsin during the most recent 10-year period, in this case the period 1992-2001 (9-year period for which the data are available). Two approaches were used to obtain estimates of the change in carbon stocks and thus emissions or removals of CO2 to and from the atmosphere. One approach (RS Approach) was based on the use of two remote sensing map products (Wiscland 1992 and NLCD 2001) to estimate change in forest cover that was then combined with estimates of forest carbon stocks in above and below ground live tree biomass based on FIA plots. This resulted in estimates of CO2 emissions caused by forest loss, CO2 removals caused by gain in new forests, and CO2 removals caused by forests remaining as forests. During the 9 year period, about 13.6 million acres remained as forest, about 2.2 million acres of forest were converted to a non forest use, and about 3.2 million acres of non-forest land became forest, for a net increase of about 0.98 million acres of forest. The dynamics of forest change resulted in a net annual removal of CO2 from the atmosphere of about 8.45 million tons or 76 million tons for the entire 9 year period. Heavily forested counties in the north east and north of the state had the highest net removal of CO2—up to more than 150,000 tons over the 9 year period, whereas several counties in the south-west part of the state exhibited a net emissions of CO2 (up to 100,000 ton; Figure E-1).

-200,000 Figure E-1. The net change in total tons of CO2 in Wisconsin counties. Negative sign indicates a removal from the atmosphere and positive sign is an emission The second approach (USFS Heath) was based on a combination of models developed by the USFS and applied to the FIA data for the state of Wisconsin. This approach resulted in estimates of net change in forest area and net removals of CO2 only. According to this second approach, the net change in forest area during the 9 year period was an increase of about 0.21 million acres, and an estimated net annual removal of CO2 of 7.9 million tons or 71 million tons for the entire period. The USFS Heath approach also projected further changes in forest area and CO2 emissions until 2007. Between 2001 and 2007, the


5


forest area was projected to increase by another 0.45 million acres; net annual removals CO2 were projected to be 7.0 million tons. A comparison of the two sets of estimates results in the following conclusions: 1. Neither approach was really designed to estimate emissions and removals of CO2 from changes in forest lands and consequently both have sources of uncertainty. a. The RS approach (use of remote sensing products with high accuracy combined with field data for carbon estimation), theoretically, is capable of producing the best estimates of the dynamics of land cover change and resulting emissions and removals of CO2. Not only can such a system track changes in forest lands but it is able to do a complete land based accounting using one system. This approach is being used by some countries to develop a GHG accounting system for all land use changes. For example, Australia and New Zealand (and several developing countries are heading this way too) did not have a national forest inventory and so developed their system from scratch designed to monitor all lands in the country. The most accurate way to obtain data on area change is to use “change detection” method in which first a benchmark map is produced and then this is overlain with subsequent RS images to determine which “pixels” have changed and how. It does not require a complete wall to wall interpretation each time. Also to use this method more frequent acquisition of imagery would be needed (annual to biannual) to better detect the dynamics of land conversions. In the analysis presented here it was not possible to use the change detection method for determining change and the two RS products were not developed by the same organization and were 9 years apart. b. The USFS Heath approach focuses only on forests and was originally designed to estimate the growing stock volume of timber from a statistically designed state wide inventory of timberlands. The re-measurement intervals varied by state (from anywhere from 5 to 14 years). The design of the system has changed through time and more frequent measurements are now made in a consistent manner across all states. The focus is still on measurements of trees with measurements for other carbon pools made on a subset of the FIA plots. Given the way the data are reported, it is difficult to estimate gross changes in forest cover and thus emissions and removals of CO2 (Linda Heath, 2007, personal communication). Models are used to interpolate between inventories to obtain estimates of annual net rates of removals of CO2. 2. Given these fundamental differences in the two approaches, it is difficult to conclude what the real change in forest area has been. The RS approach provides estimates of the dynamics of forest change while the USFS Heath approach provide an estimate of the net change. Both approaches conclude that the area of forests in Wisconsin increased during the 1992-2001 period, with this increase occurring more rapidly through 2007 based on the USFS Heath approach. However, there is a five-fold difference between what the area has increased by: 0.98 million acres for the RS approach and 0.21 million acres for the USFS Heath approach. The RS approach gives the highest total area of forests in 2001—16.8 million versus 16 million for the USFS approach. However, given the sources of uncertainty in both analyses, it is difficult to conclude that there is a real difference in the area of forests in 2001. 3. The two approaches provide estimates of the annual net CO2 removals by live trees that are practically the equal for the same 9 year period, given the various sources of uncertainty in both analyses—8.5 million tons for the RS approach and 7.9 million tons for the USFS Heath approach. Thus we conclude that the baseline net emission for this period falls within this range. 4. Given the attention that the land use and forestry sector has in Wisconsin and its potential role to mitigate GHG emissions, it behooves Wisconsin to consider allocating the resources to develop and implement a complete land based GHG accounting system based on a combination of remote sensing imagery, collected at a minimum of a 2 year cycle using the change detection technique, combined with FIA plots where appropriate or collection of targeted data in areas where most change occurs.


6


1. BACKGROUND 1.1.

General Approach

The goal of work presented in this report is to establish the baseline carbon stocks and changes in carbon stocks for the forest sector in the state of Wisconsin during the most recent 10-year period, in this case the period 1992-2001 (9-year period for which the required data are available). Baselines provide an estimate of the emissions and removals of greenhouse gases, in this case carbon dioxide, due to changes in the use and management of forest land. Such baselines are useful as (1) a reference against which future emissions and removals from changes in the use and management of forest lands can be monitored and (2) for identifying where, within the landscape of Wisconsin, major opportunities could exist for enhancing carbon stocks and/or reducing carbon emissions to potentially mitigate greenhouse gas emissions.

1.2.

Datasets used in the analysis

To develop the baseline for a specified time period, two types of data are needed: (1) the area of forests undergoing a change and (2) the change in carbon stocks in the same areas. To develop a trend in the baseline, data for three points in time are needed to cover two time intervals. However, the areas of change in forest were available only for the year 1992 from the Wisconsin Land Cover (WISCLAND) and for the year 2001 from the National Land Cover Database, so only one time interval (two points) was used in the analysis. Both land cover/land use maps used different land classification systems that were harmonized as much as possible with a focus on the forest classes. The 1992 map had more classes of forest types than the 2001 map. Forest carbon stocks across Wisconsin were estimated using the U.S. Forest Service Forest Inventory and Analysis (FIA) database. Following Acts of Congress in 1928 and 1974, the USFS has been collecting data systematically on US forests via the FIA. Rates of carbon accumulation and carbon stocks were derived from Smith et al (2006).

1.3.

Geographical subdivision of the state

The forest baseline analysis is performed at the county-level; counties in Wisconsin are shown in Figure 1.


7


Figure 1. Wisconsin counties. Source: Digital Map Store. http://county-map.digital-topomaps.com/wisconsin

2. METHODS FOR BASELINE DEVELOPMENT 2.1.

General Approach

The goal of this section is to outline methods used to develop a baseline of carbon dioxide (CO2) emissions and/or removals in the forest sector of Wisconsin for the period of time between 1992 and 2001. The focus of this work is carbon, as carbon dioxide, and does not include other greenhouse gases such as methane (CH4) and nitrous oxide (N2O). The emissions and removals of carbon on forest lands are caused by: 1. Forests converted to non-forests—emissions of CO2 2. Non-forests (e.g. cropland and pasture) converted to forests—removals of CO2 Forests remaining as forests, both protected forests and those managed for timber—could be 3. either emissions or removals of CO2 The general approach was to determine the change in forest area due to conversion, both forest area loss and gain as well as to determine the area of forests that remained as forests. These areas of forests were estimated from state-wide land cover maps of 1992 and 2001.


8


Carbon estimates for various forest types were derived from U.S. Forest Service Forest Inventory and Analysis (FIA) data base and from Smith et al. (2006). The details of all steps are given in the following sections.

2.2.

Forest area classification

The 1992 land cover data, WISCLAND, was downloaded from Wisconsin DNR. WISCLAND is a data development partnership of public and private organizations in Wisconsin. The 2001 land cover data is the National Land Cover Dataset, NLCD, which was downloaded from the MRLC website. The 2001 NLCD is an update to a 1992 dataset and despite the name was only recently made available to the public (December 2006). A vector file for the state boundary and another for the counties was also downloaded from the Wisconsin DNR website. Both land cover data sets were clipped to the state boundary to ensure that both land cover datasets covered the same area. The total state area from this analysis was 35,88 million acres (Table 1 and 2). The 2001 NLCD database is classified into 15 categories and the WISCLAND land cover map is classified into 38 different categories. The land cover categories from each respective dataset were reclassified into the same eight aggregate categories to facilitate comparison between maps (Figure 2). Reclassification schemes for WISCLAND and NLCD land cover maps are shown in Tables 1 and 2, and a summary comparison of reclassified area between WISCLAND 1992 and NLCD 2001 land cover categories are shown in Table 3.

Cropland Developed Forest Other Pasture Water Wetland Woody Wetland

Figure 2. WISCLAND 1992 (left) and NLCD 2001 (right) land cover maps, reclassified into the same eight land cover categories .


9


Table 1. 1992 WISCLAND reclassified land cover categories and their respective areas. WISCLAND

ACRES

RECLASS

High Intensity

271,760

Developed

Low Intensity

261,402

Developed

Golf Course

26,993

Developed

AGRICULTURE

24,942

Cropland

Herbaceous/Field Crops

348,206

Cropland

Row Crops

1,003,551

Cropland

Corn

3,621,700

Cropland

Other Row Crops

1,376,714

Cropland

Forage Crops

4,679,439

Pasture

GRASSLAND

3,842,615

Pasture

Coniferous

932

Forest

Jack Pine

404,968

Forest

Red Pine

452,572

Forest

White Spruce

6,298

Forest

424,781

Forest

7,780

Forest

Aspen

2,292,055

Forest

Oak

Mixed/Other Coniferous Broad-Leaved Deciduous

1,268,906

Forest

Northern Pin Oak

22,745

Forest

Red Oak

27,484

Forest

Maple

622,726

Forest

Sugar Maple

248,726

Forest

Mixed/Other Broad-Leaved Deciduous

6,376,276

Forest

Mixed Deciduous/Coniferous

1,288,150

Forest

OPEN WATER

1,203,115

Water

Emergent/Wet Meadow

1,146,726

Wetland

Floating Aquatic Herbaceous Vegetation

12,891

Wetland

Lowland Shrub

527,729

Wetland

Lowland Shrub Broad-Leaved Deciduous

851,083

Wetland

Lowland Shrub Broad-Leaved Evergreen

82,033

Wetland

Lowland Shrub Needle-Leaved

24,790

Wetland

Forested Broad-Leaved Deciduous

1,143,823

Woody Wetland

Forested Coniferous

756,662

Woody Wetland

Forested Mixed Deciduous/Coniferous

518,442

Woody Wetland

Cranberry bog, Shrubland, Barren

711,617

Other

3,844

Deleted

CLOUD COVER TOTAL AREA


35,880,633

10


Table 2. 2001 NLCD reclassified land cover categories and their respective areas. NLCD

ACRES

Open Water Developed, Open Space Developed, Low Intensity Developed, Medium Intensity Developed, High Intensity Deciduous Forest Evergreen Forest Mixed Forest Shrub/Scrub, Barren Land (Rock/Sand/Clay Grassland/Herbaceous Pasture/Hay Cultivated Crops Woody Wetlands Emergent Herbaceous Wetlands TOTAL AREA

RECLASS

1,225,228 1,394,300 725,106 204,358 72,513 11,272,930 1,295,751 1,197,651

Water Developed Developed Developed Developed Forest Forest Forest

382,848 607,157 3,828,497 9,437,168 3,076,457 1,160,670 35,880,633

Other Pasture Pasture Cropland Woody Wetland Wetland

Table 3. Comparison of reclassified land cover categories. ACRES Cropland Developed Forest Other Pasture Water Non-woody Wetland Woody Wetland TOTAL

WISCLAND 1992

NLCD 2001

6,375,114 560,155 13,444,400 711,617 8,522,053 1,203,115

9,437,168 2,396,276 13,766,332 382,847 4,435,654 1,225,228

2,645,252 2,418,927 35,880,633

1,160,670 3,076,457 35,880,633

Some of the differences in areas of land use between the two mapping products are likely caused by attempts to harmonize different classification schemes of remote sensing imagery, the interpretation and classification done by different groups and for different purposes. This is particularly the case for agricultural lands, which have different reflectances depending upon the time of the year and the specific ground cover present when the imagery was analyzed. For example, in Wiscland classification, forage crops were identified as a separate classification but when re-classified by us forage crops were combined with grassland to generate one class of “pasture”. The reasons we re-classified Wiscland in this manner was because in the NLCD map, hay (a forage crop) was grouped with pasture. Combining the cropland and pasture classes for Wiscland in Table 3 results in a total area of 14.9 million acres, a value slightly higher than the 14.6 million acres of crop and pasture land reported by the 1992 National Resources Inventory data. The combined cropland and pasture classes of the NLCD map gives a total of 13.9 million acres or about a 1 million acre difference. Some of this difference is likely a real change whereas some is likely attributed to the respective classification methods. The differences in areas reported for forests in the two map products most likely reflects real differences as there are fewer problems in identifying forests.


11


After both land cover maps were harmonized to reflect identical classification schemes, maps were again reclassified into categories of forest and non-forest to calculate areas of forest gain and forest loss. Woody wetlands were grouped into the forest category. The resultant forest/non-forest maps for the years 1992 and 2001 are shown in Figure 3.

Figure 3. Forest (displayed in green, including woody wetlands) and non-forest (displayed in grey) for WISCLAND 1992 (left) and NLCD 2001 (right).

2.3.

Calculating change in forest area

Change analyses are conducted by comparing the land cover maps from time 1 with that for the same location at time 2. A Boolean dataset for 1992 and another for 2001 of forest, non-forest was created using the reclassification technique. All non-forest categories in both land cover data sets were given a new value of 0. For forest and woody wetland categories, in 1992 a value of 1 was used while for the 2001 data a value of -1 was used. By combining the two Boolean images using a map algebra technique of addition, areas of change and no change were detected. New dataset, now referred to as change data, were developed that contained three values: forest gain, no change, forest loss. The change data were combined with the county vector file and the original land cover data from both years. It was then possible to see how the original land cover categories within each county and across the whole state changed, if at all, and into what new category the change occurred—often referred to as a land cover change matrix. The data on changes in forest area between specified dates for each pixel are summarized by the use of pivot tables in Excel and a table of the areas of each land use type that changed is produced. The number of acres with an increase, decrease, or no change in forest cover is then summed.

2.4.

Estimating forest carbon stocks

All estimates of forest carbon stock and change in carbon stock are for live tree above and below ground biomass only. The data from FIA Mapmaker program only report carbon in these pools whereas the tables from Smith et al. (2006) do contain estimates for the other pools—dead wood, forest floor,


12


understory, and soil. For consistency across all analyses of the forest carbon emissions/removals, we only include the live tree above and below ground carbon pools.

2.4.1. Carbon stocks for forest loss 1

The specific data source for summary data on forest carbon stocks was the FIA Mapmaker Program . In the program, all Wisconsin counties were selected and data were extracted on areas of forestlands and all live tree oven-dried biomass of forestlands, by the different types. When the data set is reclassed into different forest types, the number of FIA plots representing a given forest type can be small and thus the level of confidence around the mean estimate is potentially reduced. To overcome this potential problem and increase the number of FIA plots representing a given forest class, each of the counties immediately bordering Wisconsin in Minnesota, Iowa, Illinois and Michigan were also selected and extracted. In total, forty-five FIA forest types were cross-walked into the 14 Wisconsin forest types (Table 4). The mean stocks representing the biomass present in live trees above- and below-ground only in each of the Wisconsin/NLCD forest types are listed in Table 5. These estimates derived from Mapmaker are not based on an average per se of a given number of plots, but rather represent the average stock based on dividing the total carbon stock reported for a given forest type in the analytical domain by the total area covered by the given forest type. Thus the values for carbon represent the average carbon stock across all age classes within a given forest type. Where an area of forest was lost between the two land cover images (1992 and 2001), carbon loss was calculated as the area of loss multiplied by the mean forest carbon stock as given in Table 5. More detailed analysis was possible for the forest loss component than the forest gain component as more forest types were mapped in the 1992 Wisconsin map and FIA data exist for many of those forest types. It is not correct to assume that the entire carbon stock present prior to the deforestation event will be immediately oxidized. A proportion will be extracted and converted to long term wood products, and a proportion of these products will remain in this sequestered state even 50 or 100 years after the initial deforestation. Here it is assumed that 20% of the hardwood standing volume and 30% of the softwood standing volume is destined to form long-term wood products. Assumptions on the rate of retirement and subsequent oxidation are derived from the USDOE 1605b voluntary reporting guidelines (http:www.pi.energy.gov/enhancing GHGregistry/documents/PartIForestryAppendix.pdf). As wood products are retired and oxidized every year a decision has to be made on what constitutes a permanent sequestration. Here we use the biomass remaining in wood products 50 years after initial deforestation. Table 4. FIA forest types cross-walked into Wisconsin forest types.

FIA Forest Type Jack pine Red pine White spruce Eastern white pine White pine / hemlock Eastern hemlock Balsam fir Eastern red cedar Blue spruce Scotch pine 1

Wisconsin Forest Types Jack Pine General Coniferous Red Pine White Spruce Mixed/Other Coniferous

Version 3.0: http://www.ncrs2.fs.fed.us/4801/fiadb/fim30/wcfim30.asp


13

Wisconsin Forest Category Forest


FIA Forest Type Other exotic softwoods White pine / red oak / white ash Aspen Post oak / blackjack oak White oak Northern red oak Bur oak Chestnut oak / black oak / scarlet oak Red maple / upland White oak / red oak / hickory Black walnut Black locust Red maple / oak Mixed upland hardwoods Black ash / American elm / red maple Sugarberry / hackberry / elm / green ash Sugar maple / beech / yellow birch Black cherry Cherry / ash / yellow-poplar Hard maple / basswood Elm / ash / locust Paper birch Balsam poplar Eastern redcedar / hardwood Other pine / hardwood White pine / red oak / white ash River birch / sycamore Cottonwood

Wisconsin Forest Types

Aspen Oak

Mixed DeciduousConiferous Broad-leaved Deciduous

Sycamore / pecan / American elm Silver maple / American elm Red maple / lowland Cottonwood / willow Coniferous

Tamarack Northern white-cedar


General Broadleaved Deciduous

Maple Mixed/Other Broadleaved Deciduous

Willow

Black spruce

Wisconsin Forest Category

14

Mixed DeciduousConiferous

Wetland


Table 5. Mean forest carbon stocks as derived from FIA Mapmaker and used for forest loss (adjusted to account for wood products as described above). Mean Forest Carbon Stock

Forest Type FOREST

Coniferous

Broad-leaved deciduous

Jack pine Red pine White spruce Mixed/Other Coniferous General Coniferous Aspen Oak Maple Mixed/Other Broad-leaved Deciduous General Broad-leaved Deciduous

Mixed Deciduous-Coniferous WETLAND

Broad-leaved Deciduous Coniferous Mixed Deciduous-Coniferous

t C/ha 20 43 29 39 35 28 54 43 46 42 36 45 26 29

t CO2/acre 30 64 43 58 53 42 80 64 68 62 54 67 38 43

2.4.2. Carbon stocks for forest gain Where additional areas of forest were detected, applying a mean carbon stock value from across the state would be unreasonable, because a forest attaining this carbon stock value would require many years of growth. Because the date that the new forest area was planted or regenerated naturally is unknown, an assumption of 5 years of growth was applied, with 5 years representing the likely average age of all new forest areas during the interval between the two map products. Carbon values from five years of growth were derived from Smith et al (2006; which contains the 1605b tables for afforestation based on FORCARB 2 model and FIA data). The 1605b ‘look-up’ tables give estimates of carbon sequestration after afforestation for forest types across the whole United States. For Wisconsin (Northern Lake States), tables are given for six forest types (Table 6): (1) aspen-birch; (2) elmash-cottonwood; (3) maple-beech-birch; (4) oak-hickory; (5) spruce-balsam fir; and (6) white-red-jack 2 pine. These data are available as Tables B7 to B12 in the Forestry Appendix or Smith et al. (2006) These tables can be reclassed as follows: • Aspen-Birch, Maple-Beech-Birch, Oak-Hickory = Broad-leaved deciduous forests • White-Red-Jack Pine = Coniferous forests • Elm-Ash-Cottonwood = Wetland broad-leaved deciduous • Spruce-Balsam Fir = Wetland coniferous forests

2

http://www.pi.energy.gov/enhancingGHGregistry/documents/PartIForestryAppendix.pdf


15


Table 6. Carbon stocks in live tree biomass after five years of growth for six forest types in Wisconsin. Forest Type (1605b)

Forest type (Wisconsin)

Aspen-Birch Elm-Ash-Cottonwood

Broad-leaved deciduous forests Wetland broad-leaved deciduous forests Broad-leaved deciduous forests Broad-leaved deciduous forests Wetland coniferous forests Coniferous forests

Maple-Beech-Birch Oak-Hickory Spruce-Balsam Fir White-Red-Jack pine

Carbon value (5 yrs of growth, t CO2/acre) 10.8 5.8 7.6 9.9 5.0 0.6

2.4.3. Carbon accumulation in forests remaining as forests The majority of the forests in Wisconsin remained as forest between 1992 and 2001, representing neither an increase nor decrease in forest area. However, it is not valid to assume that these forests were not removing carbon dioxide from the atmosphere and sequestering net carbon during the process of growth and management. For the ‘forests remaining as forests’ category, a carbon accumulation rate was calculated in three steps. First, a mean carbon stock was estimated by forest type based on Table 5 above. Second, based on this mean stock value, forest age was estimated from the appropriate 1605b ‘look-up’ table by matching the mean stock against the stock versus age data in the 1605b tables. Third, the annual growth rate was interpolated linearly from the look-up table and applied for the 9-year period between 1992 and 2001 (Table 7).


16


Table 7. Forest type, mean carbon stock, age, interpolated annual growth rate. Forest Type FOREST

Coniferous

Broad-leaved deciduous

t CO2/acre Jack pine

30

Red pine

66

White spruce Mixed/Other Coniferous

43

General Coniferous

54

Aspen Oak

43 81

Maple Mixed/Other Broadleaved Deciduous General Broadleaved Deciduous

65

60

69 63

Mixed DeciduousConiferous WETLAND

55 Broad-leaved Deciduous

68

Coniferous Mixed DeciduousConiferous

39 44

Look-up Table White-Red-Jack Pine Clearcut White-Red-Jack Pine Clearcut White-Red-Jack Pine Clearcut White-Red-Jack Pine Clearcut White-Red-Jack Pine Clearcut Aspen-Birch Oak-Hickory Maple-BeechBirch Maple-BeechBirch Maple-BeechBirch Maple-BeechBirch Elm-AshCottonwood Spruce-Balsam Fir Elm-AshCottonwood

Equivalent Age (yr)

Accumulation (t CO2/ac.yr)

20

4.07

30

4.08

25

4.07

25

4.07

25

4.07

25 40

1.91 2.43

35

2.58

35

2.58

35

2.58

30

2.58

55

1.25

25

2.23

35

1.44

3. RESULTS 3.1

Forest extent in 1992

In 1992, northern counties in Wisconsin were more forested (on a percentage basis) than southern counties (Figure 4). Menominee County had the highest percentage of forest area (>90% forested). The counties in the south and southeast have few forests remaining, ranging from 90%

Figure 4. The percentage of forest land cover for each county in 1992

3.2

Forest area gains and losses, 1992-2001

3.2.1 Change in forest area at the state level Differences in forest cover between 1992 and 2001 in Wisconsin are shown in Figure 5. “No change’ refers to the persistence of forest land (and other land cover classes) over the time period. From this analysis, there appears to be a net gain of 979,462 acres of forest between 1992 and 2001 (Table 8).

Forest gain No change Forest loss

Figure 5. Change in forest cover in Wisconsin between the years 1992 and 2001. Forest gain is shown in green and forest loss is shown in red. No change refers to the persistence of different land cover classes over the time period.


18


Table 8. Gross and net changes in land and forest cover (forest and woody wetland classes) between 1992 and 2001. ACRES No change (all lands) No change in forest cover Forest gain Forest loss Net gain in forest area

30,416,118 13,620,801 3,221,989 2,242,527 979,462

The value of using remote sensing products to analyze land use change is that the origin and destination of the changes can be determined as shown in Table 9. However, when the two map products are from different sources, some of the land use changes may not necessarily be “real”, but reflect instead the problems of comparing two different land cover maps (see above section 2.2). For example, non-woody wetlands appear to have lost about 1.5 million acres between 1992 and 2001, but this may be a reflection of mis-classification rather than actual loss; the appearance of flooding on the land cover maps are affected by time of year and amount of precipitation in the two years compared. Despite these difficulties, most of the gain in forest area (forest and woody wetland classes) by 2001 appears to have arisen from pasture (about 1.2 million acres), non-woody wetlands (1.3 million acres) and cropland (0.39 million acres). Loss of forest area was caused mainly by conversion to cropland (0.67 million acres), developed (0.53 million acres) and pasture/non-woody wetland (0.75 million acres). The area of forest in 2001 shown by the NLCD map is 16.8 million acres. According to the state FIA data base (Mapmaker vers. 2.1), the area of forest is 16.1 million acres. An estimate of the 95% confidence interval (based on data provided by Wisconsin DNR) is ±8.3% of the mean area or ±1.3 million acres. Thus we conclude that the estimate of forest cover from the NLCD 2001 map is comparable to the estimate from the FIA data base. Table 9. Land use change matrix for 1992 and 2001 land cover categories (woody wetland have been combined with forests into one class).

2001 acres Cropland Developed Cropland

Forest

1992 acres Other Pasture

Water

Wetland 2001 Total

4,078,814

20,774

665,960

209,761

3,999,020

17,241

445,597

9,437,168

Developed

434,083

493,509

572,197

111,479

697,880

14,188

72,941

2,396,276

Forest

385,720

24,713 13,620,801

259,344

1,173,357

112,343

Other

60,716

2,034

152,567

17,150

113,700

2,605

34,075

382,847

1,305,559

14,165

481,922

84,955

2,422,319

6,027

120,707

4,435,654

Water

11,157

2,823

98,865

4,286

9,875

1,011,746

86,477

1,225,228

Wetland

99,065

2,137

271,017

24,643

105,902

38,966

618,941

1,160,670

560,155 15,863,327

711,617

8,522,053

1,203,115

Pasture

1992 Total

6,375,114

1,266,512 16,842,789

2,645,252 35,880,633

3.2.2 Change in forest area at the county level 3.2.2.1

Forest loss

Counties in the south and southeast of the state appear to have lost the largest percentage of their forests (Figure 6). This relatively high percent loss for the southern counties is caused by the fact that


19


they had little forest cover to begin with—less than 10 to 20% (Figure 4). The counties in the north with large areas of forest lost the least in percentage terms.

< 10% 11% - 15% 16% - 20% 21% - 25% 26% - 30% 31% - 35% > 35%

Figure 6. Percentage of forest land cover lost in each county from 1992 to 2001 based on the difference in forest cover of the Wiscland (1992) and NLCD (2001) maps.

3.2.2.2

Forest Gain

Counties that had the highest percent loss of forest cover located in the south and south-central-east were also the same counties that had the highest percent gain (Figure 7). But like the trend for the forest loss, the relatively high percent gain for these counties is caused by the fact that they had little forest cover to begin with so even a small absolute gain could represent a relatively large percent gain. The counties in the northern part of the state gained more in terms of percent than they lost—many gaining more than 15% compared to a loss of 35%

Figure 7. Percentage of forest land cover gained in each county from 1992 to 2001


20


3.2.2.3

Net change in forest area

Across the state, 70% of counties showed a net gain of forest area during the 9-year period and 30% showed a net loss of forest area (Figure 8). Large net gains (>50,000 acres) occurred in five counties (Ashland, Douglas, Oneida, Price, and Sawyer); these counties gained a combined 380,116 acres of forest (Table 10) and accounted for roughly 38% of the total net gain in forest area across the state.

< -30% -30% to -15% -15% to 0% 0% to 15% 15% to 30% > 30%

Figure 8. Net percent forest change in forest cover in Wisconsin by county from 1992 to 2001. A negative sign indicates a net loss in forest area and a positive sign indicates a net gain in forest area. Table 10. Area of forest gain and loss in based on the Wiscland 1992 and NLCD 2001 data bases. Net loss is indicated by a negative sign, net gain is indicated by a positive sign.

County Adams Ashland Barron Bayfield Brown Buffalo Burnett Calumet Chippewa Clark Columbia Crawford Dane Dodge Door Douglas Dunn Eau Claire


Forest area gain Acres

Forest area loss Acres

Net Gain/Loss Acres

44,459 84,591 40,585 96,630 23,516 25,840 69,031 15,183 57,276 61,640 26,998 34,642 30,366 23,562 21,031 153,144 33,869 33,655

49,733 28,437 45,023 60,255 9,670 44,300 54,557 3,614 45,201 43,729 31,634 19,413 45,810 12,844 23,977 76,053 56,426 38,515

-5,273 56,154 -4,438 36,375 13,846 -18,460 14,473 11,569 12,075 17,911 -4,636 15,229 -15,444 10,719 -2,946 77,091 -22,557 -4,860

21


County Florence Fond du Lac Forest Grant Green Green Lake Iowa Iron Jackson Jefferson Juneau Kenosha Kewaunee La Crosse Lafayette Langlade Lincoln Manitowoc Marathon Marinette Marquette Menominee Milwaukee Monroe Oconto Oneida Outagamie Ozaukee Pepin Pierce Polk Portage Price Racine Richland Rock Rusk Saint Croix Sauk Sawyer Shawano Sheboygan Taylor Trempealeau Vernon Vilas


Forest area gain 20,283 28,133 54,485 37,808 12,236 12,155 23,984 66,271 61,144 22,548 55,868 16,824 16,900 22,467 8,337 49,071 58,624 32,015 85,071 95,192 26,281 9,819 8,809 47,214 63,721 126,689 29,099 14,994 7,161 24,165 67,274 46,711 149,086 18,697 40,943 17,215 83,012 36,361 32,881 101,251 66,876 27,537 66,584 29,472 39,206 62,416

Forest area loss 14,748 10,127 27,086 53,217 31,563 15,414 42,271 23,537 72,174 14,264 65,879 9,352 5,474 16,717 30,450 27,433 37,265 9,033 55,053 53,283 25,816 13,003 10,128 48,965 36,707 46,230 11,800 7,177 15,512 25,492 45,441 32,265 39,079 12,758 21,258 12,647 34,408 25,269 38,577 44,846 20,028 14,328 39,450 56,994 27,945 47,490

22

Net Gain/Loss 5,534 18,006 27,399 -15,409 -19,327 -3,259 -18,287 42,733 -11,030 8,284 -10,011 7,472 11,426 5,750 -22,113 21,638 21,358 22,982 30,019 41,909 465 -3,184 -1,319 -1,751 27,015 80,459 17,299 7,817 -8,350 -1,327 21,833 14,446 110,007 5,939 19,685 4,568 48,604 11,092 -5,695 56,405 46,848 13,209 27,134 -27,522 11,260 14,926


County Walworth Washburn Washington Waukesha Waupaca Waushara Winnebago Wood Total

3.3

Forest area gain 26,921 61,927 28,165 39,442 54,530 45,162 13,153 53,783 3,221,989

Forest area loss 15,102 48,203 13,993 20,399 19,552 20,808 7,204 40,124 2,242,527

Net Gain/Loss 11,819 13,724 14,171 19,043 34,978 24,354 5,949 13,659 979,462

Existing carbon stocks in 1992

Counties with forests with high carbon stocks are located in the central and southern counties as expected as this is where the mixed hardwood forests tend to be located (Figure 9). The forests in the northern counties are dominated by conifer species which tend to grow more slowly and accumulate less carbon.

< 55 56 - 60 61 - 65 66 - 70 71 - 75 > 75

Figure 9. Area-weighted average tons of CO2 per acre for each county (t CO2/acre)

Even though the forests in the more central and southern counties have the highest carbon stocks they also have less area of forests, resulting in these counties storing less forest carbon overall (Figure 10). There is almost a 10-fold difference in the total quantity of carbon stored in the northern counties compared to the south-eastern and east-central counties. The northern counties contain more than 20 million metric tons of CO2 compared to less than 6 million t in the south-eastern and east-central counties.


23


26

Figure 10. The total quantity of CO2 for each county in forests for 1992 (Mt CO2)

3.4

CO2 emissions and removals, 1992-2001

3.4.1 CO2 emissions and removals from forests at the state level Across Wisconsin, over the 9 year period, a total of 189 million t CO2 are removed from the atmosphere in forests remaining as forests or 21 million t CO2 per year. In addition, 27 million t CO2 were sequestered by new forests as a result of forest area gain; an equivalent of 3 million t CO2 per year. This total gain of 24 million t CO2 per year compares to about 15.5 million t CO2 per year emitted to the atmosphere as a result of a loss in forest area between 1992 and 2001 for a total of about 140 million t CO2. Forests in the state of Wisconsin were therefore a net carbon sink, sequestering approximately 8.5 Mt CO2 per year during the period 1992-2001 (Table 11). The net change in forest area in Wisconsin resulted in a net emission to the atmosphere of about 12.5 million t CO2/year. The reason for this is that the forests that were converted to another land use contained more biomass than the forests that were gained. For example, the forest area lost contained about 54-82 t CO2/acre, and only a small proportion of this biomass was effectively permanently sequestered in long term wood products (permanent sequestration here assumed to be equal to biomass remaining in wood products after 50 years), resulting in an emission estimate of 15.5 million t CO2. For the areas that gained forests, we assumed that their biomass stock would be equivalent of that of 5-year old forests (rounded up mid point of the 9 year time interval). These forests contained on average between 0.6-10.8 t CO2 per acre at age 5 yr. The estimated emissions and removals of CO2 from the changes in forest cover include only those emissions and removals occurring in live tree biomass; they do not include any emissions or removals in the soil carbon pool. Given that forest area increases during the 9-year period there is a possibility that an additional quantity of CO2 would be sequestered in the soil. However, most of the gain in forest area came from lands classed as pasture and non-woody wetlands in the 1992 image (more than 76% of the total gain in area), and, typically, conversion of grassland to forests does not result in any gain in soil carbon; it can range from a small loss to a small gain. Only about 12% of the gain in forest cover


24


originated from cropland, where it is possible that over an average period of 5 years a small amount of soil carbon may have accumulated. For example, soil carbon gain under no til agriculture is about 0.400.55 t CO2/acre.year (see Section on Soil C sequestration by Neil Sampson). Using this value for the 0.39 million acres of cropland converted to forests over an average of 5 years results in an estimated gain of about 926,000 t CO2 over the whole period or about 3% of the CO2 gained in the live trees. On the loss side of the equation, about 56% of the forest land lost in 1992 was converted to crop land or urban land. Assuming that soil CO2 is lost from forest lands at the same rate as it is gained and using the same calculations as above, an estimated 2.9 million t CO2 may have been emitted from the forest soils converted to crop and urban land, accounting for about 2% of the emissions lost from the live tree biomass. Table 11. Carbon emissions and removals in Wisconsin forests (from changes in the live tree biomass only). A positive sign indicates a CO2 emission (addition to the atmosphere)and a negative sign indicates a CO2 removal (subtraction from the atmosphere). Component Forest Emissions Forest Removals Net (emission) Removals by forests remaining as forests Net (removal)

Carbon emissions/removals (t CO2/year) + 15,543,727 - 3,000,206 + 12,543,521 - 20,999,595 8-8,456,074

3.4.2 CO2 emissions and removals due to land use change at the county level 3.4.3 CO2 emissions and removals at the county level Details of county by county CO2 emissions and removal and net change caused by change in forest cover are given in Table 12, with county scale maps in Figures 11-14. Table 12. County summary of annual emissions and removals of CO2 from forest cover change. A positive sign indicates a CO2 emission (addition to the atmosphere) and a negative sign indicates a CO2 removal (subtraction from the atmosphere). County

Gain in CO2

Loss in CO2

Growth in Forests Remaining as Forests in CO2 Total tons

Net Change

Adams

-34,719

331,063

-341,859

-45,514

Ashland

-76,716

170,770

-398,063

-304,009

Barron

-39,840

326,220

-365,768

-79,387

Bayfield

-84,164

353,516

-426,604

-157,252

Brown

-22,955

71,183

-250,229

-202,001

Buffalo

-25,034

334,580

-239,965

69,581

Burnett

-57,665

340,798

-426,604

-143,470

Calumet

-14,317

26,280

-144,830

-132,867

Chippewa

-55,803

328,427

-365,768

-93,143

Clark

-64,636

308,197

-333,472

-89,911

Columbia

-25,256

237,203

-309,563

-97,616


25


Gain in CO2

Loss in CO2

Growth in Forests Remaining as Forests in CO2

Net Change

Crawford

-35,846

162,738

-129,308

Dane

-31,377

357,046

-226,320

99,348

Dodge

-22,531

94,478

-175,248

-103,302

Door

-2,415

-15,855

173,304

-260,368

-102,919

-137,081

408,307

-426,604

-155,377

Dunn

-31,479

426,113

-239,965

154,669

Eau Claire

-32,728

283,276

-314,821

-64,273

Florence

-19,111

87,636

-416,715

-348,190

Fond du Lac

-26,068

74,737

-175,248

-126,579

Forest

-46,762

168,144

-416,715

-295,332

Grant

-39,635

433,022

-180,380

213,007

Green

-12,761

251,538

-180,380

58,397

Green Lake

-11,245

113,811

-341,859

-239,292

Iowa

-25,134

334,488

-208,295

101,059

Douglas

Iron

-60,062

144,535

-398,063

-313,590

Jackson

-56,503

450,815

-365,768

28,544

Jefferson

-20,120

106,188

-175,248

-89,180

Juneau

-51,193

453,458

-365,768

36,497

Kenosha

-18,334

70,302

-144,830

-92,862

Kewaunee

-14,112

39,370

-144,830

-119,572

La Crosse

-22,179

130,667

-365,768

-257,280

Lafayette

-8,933

248,215

-129,308

109,974

Langlade

-45,975

172,934

-479,429

-352,469

Lincoln

-51,470

220,751

-398,063

-228,782

Manitowoc

-27,377

64,588

-144,830

-107,619

Marathon

-84,552

399,384

-365,768

-50,935

Marinette

-87,897

307,415

-447,133

-227,615

Marquette

-22,830

194,142

-341,859

-170,546

Menominee

-8,858

85,379

-314,821

-238,300

Milwaukee

-8,953

75,695

-126,804

-60,062

Monroe

-46,294

341,022

-333,472

-38,744

Oconto

-57,762

253,549

-479,429

-283,641

Oneida

-104,889

259,813

-335,350

-180,425

Outagamie

-27,019

87,366

-250,229

-189,882

Ozaukee

-14,350

53,226

-144,830

-105,954

Pepin

-7,126

117,377

-225,694

-115,444

Pierce

-25,695

193,270

-314,821

-147,245

Polk

-67,271

325,315

-426,604

-168,560

-44,383

227,744

-341,859

-158,498

-128,151

227,196

-367,645

-268,600

Portage Price Racine

-19,816

95,567

-144,830

-69,079

Richland

-44,201

164,932

-180,380

-59,649

Rock

-17,932

95,971

-226,320

-148,282

Rusk

-77,706

222,862

-314,821

-169,665


26


Gain in CO2

Loss in CO2

Growth in Forests Remaining as Forests in CO2

Net Change

Saint Croix

-33,909

190,665

-314,821

Sauk

-33,737

293,806

-309,563

-158,064 -49,495

Sawyer

-93,173

292,397

-314,821

-115,596 -238,986

Shawano

-63,631

139,466

-314,821

Sheboygan

-25,694

106,682

-175,248

-94,260

Taylor

-62,798

259,381

-398,063

-201,480

Trempealeau

-30,311

430,658

-290,912

109,435

Vernon

-42,347

232,906

-180,380

10,179

Vilas

-52,276

268,808

-335,350

-118,818

Walworth

-28,560

113,891

-226,320

-140,989

Washburn

-57,928

306,292

-426,604

-178,239

Washington

-26,873

104,258

-175,248

-97,863

Waukesha

-38,847

151,923

-175,248

-62,172

Waupaca

-51,008

142,858

-314,821

-222,971

Waushara

-38,615

154,366

-341,859

-226,108

Winnebago

-12,543

53,448

-250,229

-209,324

Wood

-47,298

275,996

-365,768

-137,070

-3,000,206

15,543,727

-20,999,595

-8,456,075

TOTAL

< (-400,000) (-399,999) - (-300,000) (-299,999) - (-200,000) (-199,999) - (-150,000) (-149,999) - (-100,000) (-99,999) - (-75,000) (-74,999) - (-50,000) > -50,000

Figure 11. The annual estimated emission of CO2, in tons, for each county caused by loss of forest cover between 1992-2001.


27


< - 10,000 10,001 - 20,000 20,001 - 30,000 30,001 - 40,000 40,001 - 50,000 50,001 - 60,000 60,001 - 70,000 > 70,000

Figure 12. The annual estimated removal of CO2 for each county that was caused by gain in forest cover between 1992-2001 in tons CO2.

< 150,000 150,001 - 200,000 200,001 - 250,000 250,001 - 300,000 300,001 - 350,000 350,001 - 400,000 400,001 - 450,000 > 450,000

Figure 13. The annual estimated removal of tons CO2 for each county caused by forest remaining forests between 1992-2001.


28


-200,000

Figure 14. The net change in total tons of CO2 in Wisconsin counties. Negative sign indicates a removal from the atmosphere and positive sign is an emission.

3.4.1.1 Projected CO2 removals between 2001-2024 by forests gained during 1992-2001 In future years an additional sequestration will result from the additional CO2 that will be removed from the atmosphere assuming that the forests gained between 1992-2001 continued to grow and sequester carbon. The projection of this additional carbon sequestration up until 2024 is 954 million t CO2 (Table 13), a significant quantity. Assuming the annual rate of carbon accumulation for the period 1992-2001 in forests remaining as forests is the same for these forests during the period 2001-2024, an additional 483 million t CO2 could be sequestered. Table 13. Total CO2 sequestered in forests gained between 1992-2001 and extended to 2024. County

Gain in CO2 Total tons

Adams Ashland Barron Bayfield Brown Buffalo Burnett Calumet Chippewa Clark Columbia


29

13,359,346 34,972,867 11,018,488 48,804,291 2,487,594 11,922,148 18,891,273 1,141,808 14,260,800 18,685,545 5,789,333


County


Crawford Dane Dodge Door Douglas Dunn Eau Claire Florence Fond du Lac Forest Grant Green Green Lake Iowa Iron Jackson Jefferson Juneau Kenosha Kewaunee La Crosse Lafayette Langlade Lincoln Manitowoc Marathon Marinette Marquette Menominee Milwaukee Monroe Oconto Oneida Outagamie Ozaukee Pepin Pierce Polk Portage Price Racine Richland


30

10,722,983 6,295,161 2,125,706 4,990,651 34,842,459 10,716,788 10,667,870 16,809,092 2,484,163 34,154,096 11,068,168 2,353,837 1,881,358 8,242,686 27,671,864 21,651,319 2,459,646 12,580,639 1,501,258 1,724,361 7,539,259 2,084,167 24,256,337 25,330,217 3,009,889 23,738,796 38,783,481 5,970,674 12,907,740 804,974 14,880,286 20,711,347 37,895,822 3,390,099 1,341,307 3,226,264 5,768,830 15,030,185 11,193,414 38,716,898 1,773,302 9,286,083


County


Rock Rusk Saint Croix Sauk Sawyer Shawano Sheboygan Taylor Trempealeau Vernon Vilas Walworth Washburn Washington Waukesha Waupaca Waushara Winnebago Wood TOTAL

3.5

2,517,481 21,340,570 5,654,526 11,168,995 40,299,118 14,485,346 3,339,604 21,991,386 10,257,510 11,869,081 33,451,633 3,276,564 21,465,252 3,056,498 4,123,756 10,620,675 8,586,360 1,072,328 11,498,854 953,992,504

Comparison with the USFS analysis for Wisconsin

The US Forest Service prepares an annual inventory of GHG emissions and removals from all forest lands in the United States. The analysis is based on the FIA database, and the FORCARB2 and ATLAS 3 models . The FORCARB2 model uses forest inventory data on growing stock volume, forest area and harvests or projections of such information from the ATLAS model to derive estimates of the carbon in live and dead trees. It also estimates carbon in other forest ecosystem pools such as dead wood and forest floor (fine litter) and in forest soil. Here we present the results of the application of this analysis for the 4 change in carbon stocks for the state of Wisconsin for the period 1990-2007 . The results for the removals in the live trees (above and below ground) for all forest lands are presented to be consistent with the values presented by our analysis above presented in Section 3.4 . The results for the other pools, except soil, are also presented for comparison. We did not include the change in carbon in the soil pool because of the way it is accounted for currently in the USFS analysis—as the forest area increases so does the carbon stock---i.e. for every new acre that comes in as forest (presumably from agricultural land) then it is assigned the carbon stock to 1 m depth for that forest type in that region with no adjustment for the original C stock that was in the agricultural land. Thus the change in soil C is greatly overestimated. The overall approach for determining forest carbon stocks and change in stock using the USFS approach above is to estimate forest C stocks based on data from two or more forest surveys conducted several 3

Woodbury et al., 2007 Linda Heath, USFS, Northern Research Station, FIA, [email protected], November 20, 2007; personal communication 4


31


years apart for each state or sub-state. There are generally two to four surveys per state available electronically, beginning with a pre-1990 survey. Carbon stocks are calculated separately for each state based on available inventories conducted since 1990 and for the inventory closest to, but prior to, 1990 (the base year). This approach ensures that the period 1990 to present can be adequately represented. Surveys conducted prior to and in the early to mid 1990s focused on land capable of supporting timber production (timberland). As a result, information on less productive forest land or lands reserved from harvest was limited. However, for Wisconsin, most forestland is timberland so there is little with limited data. Prior to the mid- to late-1990s, inventory field crews periodically measured all the plots in a state at a frequency of every 5 to 14 years. The more recent data (in this case labeled 2004 and 2005 in Table 14) are from annualized surveys. In the annualized surveys, a subset of the data (20% of the plots) from each state is measured each year. Using only these data result in a very high sampling error; thus, the full dataset measured over a period of years (called a cycle) is used as the estimate because the entire sample size is estimated to produce the target sampling error for the survey. In Wisconsin, the data labeled 2004 are actually data collected over the years 2000-2004. The data labeled 2005 are the same data from 2001-2004, but include new data collected in 2005. (This is a rolling average that FIA is currently using as an estimate.) Table 14 provides a list of the specific surveys used for Wisconsin. In essence, the results presented here from the USFS analysis are based on four sets of inventory data with models interpolating trends between the inventory data and extrapolating the trend from 2005 to 2007. Further details to produce the carbon estimates from FIA data are given in Smith and Heath (2007, specifically the Annex in this source). Table 14. Source of forest inventory and average year of field survey used to estimate Wisconsin-wide carbon stocks and changes in carbon stocks in live trees. Average Year Assigned b State to Inventory Wisconsin FISDB 2.1, 1983 1982.44 FISDB 2.1, 1996 1995.11 FISDB 2.1, 2004 2002.38 c FISDB 2.1, 2005 2003.24 a FISDB 2.1 is the snapshot version of FIADB 2.1 as available on Internet April 1, 2007 (USDA Forest Service 2007). b Average year is based on average measurement year of forest land survey plots and rounded to the nearest integer year. c The 2004 data are the complete cycle for the state, with average year 2002. The 2005 data are a complete cycle for the state with the 2005 year measurements added from cycle 7 and the previous 4 years of measurement from cycle 6. (This is the way the data are loaded in the database.) Source of Inventory Data, a Report/Inventory Year

The net change in area of forests in Wisconsin shows a gradual increase over the whole 17 yr period, from about 15.7 million acres in 1990 to 16.5 million acres in 2007 (Figure 15). The forest area from the 1992 Wiscland map is estimated to be the same as that from the FIA data base (15.8 million acres). However, whereas the forest area estimate for 2001 from the NLCD map is 16.8 million acres, the USFS analysis estimates the area at about 16 million acres for this year. After 2002, the area of forest increases to about 16.5 million acres by 2007. According to Mapmaker vers. 2.1 (from DNR), the area of Wisconsin’s forests in 2005 is 16.1 million acres (± 95% confidence interval of 1.4 million acres). As the focus of the FIA is on forests it could be argued that the area values from this data base are more accurate than those from the NLCD based on satellite imagery. On the other hand, the forest area obtained from the NLCD map is based on complete coverage and not a sample, and depending on the accuracy assessment (about 91% accuracy for most of the state and 76% for a small area in the northwest corner) one could argue that the forest area from this product is more accurate. However, taking into account the potential sources of error, the forest area based on the NLCD map is within the error bound


32


Area in million acres

of the FIA data for 2005. Even the projected increase in forest area by the USFS (Figure 15) to 16.5 million acres by 2007 is still within the 95% confidence interval of the 2005 estimate.

16.50 16.40 16.30 16.20 16.10 16.00 15.90 15.80 15.70 15.60 15.50 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 Year

Figure 15. Change in forest cover from 1990-2007 based on the FIA data base. Although in general the difference in the change in forest area between the FIA and the satellite imagery approach was small and certainly within statistical agreement, the removals of CO2 from change in forest carbon stocks in live trees is quite different (Figure 16). Based on the USFS Heath analysis, there is a declining trend in the removals of CO2 in live trees from Wisconsin forests—that is although the area is increasing, the rate of removals in live trees is decreasing. By 1991 the rate of removal was 8.4 million t CO2/yr with a gradual decline to 6.8 million t/ yr by 2007. Based on the analysis done for this report above (section 3.4), the total net removal, including emissions due to forest loss, removals due to forest gain, and removals in forests remaining as forests, was 8.5 million t CO2/yr. The total for the 9 year period 1992-2001 was 71 million t for the USFS Heath analysis and 76 million t for the imagery based analysis, or a less than 10% difference. Given that both the USFS Heath study and the one presented here (Section 3.4) include several sources of error (sampling error, error associated with converting inventory data to biomass carbon stock estimates, image interpretation error, and the like), we conclude that a reasonable estimate for the net removals from Wisconsin’s forest lands during the period 19922001 is 7.8 to 8.5 million tons of CO2 per year. The addition of other pools included in the analysis by Heath shows that the total CO2 removals increased in the recent 4-5 year period mostly caused by a large increase in the litter pool (25 million t CO2 in 4 years). Inclusion of the other pools more than double the removals of CO2 in Wisconsin forests for the most recent 4-yr period (Figure 16). It is likely that the decrease in C stocks in the live trees in the recent period is caused not only by the forest cover dynamics (loss of high C stock forests and gain by low C stock forests—cf. Section 3.4) but also by an increase in harvest of timber. An increase in the harvest of timber is the only possible explanation for the trend in the USFS Heath results—an increase in forest area, but yet a decrease in carbon stocks in live trees accompanied by a very large increase in the litter pool. The litter pool includes organic material on the forest floor that is composed of fine woody debris up to 7.5 cm in diameter (likely slash), tree litter, humus, and any fine roots in the organic forest floor layer above mineral soil). According to Wisconsin DNR (personal communication from A. Hellman), the rate of timber harvest has been relatively stable, a trend that appears to be inconsistent with the most plausible explanation for USFS Heath results. However, it must be remembered that the USFS Heath analysis is an estimation of the net removals of CO2 based on the a models that use a combination of FIA data and harvest projections (from the ATLAS model).


33


14.00

Live and dead organic matter

12.00 10.00 8.00 6.00 4.00 2.00

20 07

20 05

20 03

20 01

19 99

19 97

19 95

19 93

0.00 19 91

Removals of CO2, million t/yr

Live trees

Figure 16. Removals of CO2, in million t/yr, in live trees only (above and below ground) and in all live and dead organic matter pools (live trees plus dead wood and forest floor; does not include soil pool) from 1991 to 2007 (from Linda Heath, 2007, USFS, personal communication).

4. CONCLUSIONS The main goal of work presented in this report was to establish the baseline carbon stocks and changes in carbon stocks for the forest sector in the state of Wisconsin during the most recent 10-year period, in this case the period 1992-2001 (9-year period for which the data are available). Two approaches were used to obtain estimates of the change in carbon stocks and thus emissions or removals of CO2 to and from the atmosphere—the RS Approach (Section 3. 2-3.4) and the USFS Heath approach (Section 3.5). The RS approach resulted in estimates of CO2 emissions caused by forest loss, CO2 removals caused by gain in new forests, and CO2 removals caused by forests remaining as forests. During the 9 year period, about 13.6 million acres remained as forest, about 2.2 million acres of forest were converted to a non forest use, and about 3.2 million acres of non-forest land became forest, for a net increase of about 0.98 million acres of forest. The dynamics of forest change resulted in a net annual removal of CO2 from the atmosphere of about 8.45 million tons or 76 million tons for the entire 9 year period. The USFS Heath approach resulted in estimates of net change in forest area and net removals of CO2 only. According to this second approach, the net change in forest area during the 9 year period was an increase of about 0.21 million acres, and an estimated net annual removal of CO2 of 7.9 million tons or 71 million tons for the entire period. The USFS Heath approach also projected further changes in forest area and CO2 emissions until 2007. Between 2001 and 2007, the forest area was projected to increase by another 0.45 million acres; net annual removals CO2 were projected to be 7.0 million tons. A comparison of the two sets of estimates results in the following conclusions:


34


1. Neither approach was really designed to estimate emissions and removals of CO2 from changes in forest lands and consequently both have sources of uncertainty. a. The RS approach (use of remote sensing products with high accuracy combined with field data for carbon estimation), theoretically, is capable of producing the best estimates of the dynamics of land cover change and resulting emissions and removals of CO2. Not only can such a system track changes in forest lands but it is able to do a complete land based accounting using one system. This approach is being used by some countries to develop a GHG accounting system for all land use changes. For example, Australia and New Zealand (and several developing countries are heading this way too) did not have a national forest inventory and so developed their system from scratch designed to monitor all lands in the country. The most accurate way to obtain data on area change is to use “change detection” method in which first a benchmark map is produced and then this is overlain with subsequent RS images to determine which “pixels” have changed and how. It does not require a complete wall to wall interpretation each time. Also to use this method more frequent acquisition of imagery would be needed (annual to biannual) to better detect the dynamics of land conversions. In the analysis presented here it was not possible to use the change detection method for determining change and the two RS products were not developed by the same organization and were 9 years apart. b. The USFS Heath approach focuses only on forests and was originally designed to estimate the growing stock volume of timber from a statistically designed state wide inventory of timberlands. The re-measurement intervals varied by state (from anywhere from 5 to 14 years). The design of the system has changed through time and more frequent measurements are now made in a consistent manner across all states. The focus is still on measurements of trees with measurements for other carbon pools made on a subset of the FIA plots. Given the way the data are reported, it is difficult to estimate gross changes in forest cover and thus emissions and removals of CO2 (Linda Heath, 2007, personal communication). Models are used to interpolate between inventories to obtain estimates of annual net rates of removals of CO2. 2. Given these fundamental differences in the two approaches, it is difficult to conclude what the real change in forest area has been. The RS approach provides estimates of the dynamics of forest change while the USFS Heath approach provide an estimate of the net change. Both approaches conclude that the area of forests in Wisconsin increased during the 1992-2001 period, with this increase occurring more rapidly through 2007 based on the USFS Heath approach. However, there is a five-fold difference between what the area has increased by: 0.98 million acres for the RS approach and 0.21 million acres for the USFS Heath approach. The RS approach gives the highest total area of forests in 2001—16.8 million versus 16 million for the USFS approach. However, given the sources of uncertainty in both analyses, it is difficult to conclude that there is a real difference in the area of forests in 2001. 3. The two approaches provide estimates of the annual net CO2 removals by live trees that are practically the equal for the same 9 year period, given the various sources of uncertainty in both analyses—8.5 million tons for the RS approach and 7.9 million tons for the USFS Heath approach. Thus we conclude that the baseline net emission for this period falls within this range. 4. Given the attention that the land use and forestry sector has in Wisconsin and its potential role to mitigate GHG emissions, it behooves Wisconsin to consider allocating the resources to develop and implement a complete land based GHG accounting system based on a combination of remote sensing imagery, collected at a minimum of a 2 year cycle using the change detection technique, combined with FIA plots where appropriate or collection of targeted data in areas where most change occurs.

5. REFERENCES Smith, J. E., L. S. Heath, K. E. Skog, R. A. Birdsey. 2006. Methods for Calculating Forest Ecosystem and Harvested Carbon with Standard Estimates for Forest Types of the United States. General Technical Report NE-343. USDA Forest Service, Northeastern Research Station.


35


Smith, J. E., and L. S. Heath. 2007. Land use change and forestry and Annex 3.12. (Excerpted.) In: US EPA. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990 - 2005. EPA 430-R-07-002. Washington, DC. Available at http://epa.gov/climatechange/emissions/usinventoryreport.html (9 May 2007). Woodbury, P. B., J. E. Smith, and L. S. Heath. 2007. Carbon sequestration in the U.S. forest sector from 1990-2010. Forest Ecology and Management 241: 14-27.


36