Emerging Model-Data Requirements for Land-Use Information

Transcription

1 Emerging Model-Data Requirements for Land-Use Information George Hurtt University of Maryland NASA LCLUC Meeting Alexandria, VA October 6, 2011

2 The Scale of Land-use Land-use has substantially transformed the Earth s surface. >50% of the land surface has been affected by direct human action, >25% of forests cleared, >30% in agriculture now, million km 2 secondary recovering lands. 1 Land-use strongly affects carbon balance, climate, and biodiversity. Cumulative land-use emissions to date are comparable to fossil fuel emissions, there are regional-global biogeochemical and biophysical affects on climate, habitat destruction is the primary cause of species extinctions. 2 Future land-use decisions are important. The demand for food, fuel, and fiber are increasing; deforestation of Amazon, or logging of boreal forests, could lead to regional-global climate change; ~2000 Pg C in terrestrial biosphere. 3 (1) Vitousek et al. 1986; Turner et al. 1990; Daily 1995; Waring & Running 1998; Hurtt et al (2) Houghton et al. 1999; UNEP 2002; Sala et al. 2000; Caspersen et al. 2000; Pacala et al. 2001; Pielke et al. 2002; Hurtt et al. 2002; Bonan 2004; Roy et al. 2004; Brovkin et al. 2004; Shevliakova et al (3) Sigenthaler & Sarmiento 1991; Bonan et al. 1992; Cox et al. 2000; Steffen et al. 2003; D Almedia et al. 2006, 2007; Wise et al

3 Proposed AR5 Scheme (Land-use) LAND-USE HISTORY Reconstruction: Agriculture Wood harvest Transitions Gridded LAND-USE FUTURE IAM RCPs: Population Socioeconomic Energy Land-use Gridded/Regional LAND-USE HARMONIZATION Consistency Integration Gridding ESMs Climate C Stocks/Fluxes Biophysical effects Hurtt et al. (2009) 3

4 INPUT MODEL OUTPUT gridded (5 ) land-use states gridded (0.5 x0.5 ) potential biomass density and recovery rate national annual wood harvest Regional/gridded landuse states and wood harvest spatial pattern of crop and pasture residency time of agriculture inclusiveness of wood harvest statistics prioritization of land for conversion/logging spatial pattern of wood harvesting gridded (0.5 x0.5 F) land-use transitions gridded (0.5 x0.5 F) land-use states gridded (0.5 x0.5 F) secondary land area and age Hurtt et al. (2009, 2011) 4

5 latitude Mathematical Structure l(x,y,t+1) = A(x,y,t) l(x,y,t) l 1 l 2 a 11 a 12 a 13 a 21 a 22 a 23 l 1 l 2 l 3 a 31 a 32 a 33 t+1 t l 3 longitude Global, 0.5 x0.5, 600y, 4D: ~10 9 unknowns! after Hurtt et al. (2006) 5

6 Land-use transitions Definition: Land-use transitions describe annual changes in land-use, such as harvesting of trees or establishing/abandoning of agricultural or urban land. Relevance: Land-use transitions often directly alter land-surface characteristics. Knowledge of LUT is critical for tracking the spatial pattern of land-use activities, the effects of land-conversion events, and the legacy of recovering lands. 6

7 LUH-MESSAGE (8.5) LUH-GCAM (4.5) LUH-AIM (6) LUH-IMAGE (2.6) Hurtt et al. (2011) 7

8 Pereira et al. (2010) 8

9 Wise et al. (2009)

10 Thomson et al. (2010) 10

11 Model factor Number of Cases: Description H: Historical land-use 3: HYDE 3.0, No Data, None reconstruction T: Residence time of 2: Shifting cultivation, no shifting cultivation agricultural land L: Wood harvest history 3: No wood harvest, FAO wood harvest reconstruction, No Data reconstruction W: Land-conversion wood 2: Included 100%, not included clearing tallied as harvest to satisfy annual wood harvesting P: Priority for land-use 2: Primary, secondary transitions D: Historical start date 4: 1500, 1700, 1850, 2005 U: Urban land use included 2: Included, not included F: Future land-use projections 4: AIM, GCAM, IMAGE, MESSAGE Total number of simulations =

12 Gross T. Net T. Sec. Sec. Age Total C Net C --- Focal case --- Data-based --- All Hurtt et al. (2011) 12

13 Gross T. Net T. Wood harvest Sec. Sec. Age Total C Net C Hurtt et al. (2011) 13

14 Gross T. Net T. Shifting Cultivation Sec. Sec. Age Total C Net C Hurtt et al. (2011) 14

15 Gross T. Net T. Start Date Sec. Sec. Age Total C Net C Hurtt et al. (2011) 15

16 Gross T. Net T. Future Scenario Sec. Sec. Age Total C Net C Hurtt et al. (2011) 16

17 iesm Multi-phase Coupling Strategy iesm Control (RCP 4.5) Fossil fuel emissions IMAGE/GCAM GLM landuse CLM/CCSM C stocks, Productivity phase 1 Climate phase 2 Up/down Up/down scaling scaling (space (space and and time) time) Atm CO 2 phase 3 Edmonds et al. (2010) 17

18 Resolution Mearns (Pers. Comm.)

19 Hurtt et. al. (In prep) 19

20 Ecosystem Demography (ED) Model 1 90m (CMS) Howard County Anne Arundel County County boundary Aboveground Biomass (Mg/ha) Values <

21 Height (m) Discrete Return LiDAR 40 Cumulative Intensity km 0 25 m Intensity Dubayah et al (pers. comm.) 21

22 Fine scale forest change history mapped across the US (Huang, Goward, et al) Mississippi 22

23 ied G. Hurtt, S. Frolking, J. Fisk, L. Chini, J. Edmonds, A. Janetos, A. Thompson, B. Bond-Lamberty, R. Dubayah, J. Chambers, J. Collatz, D. Morton How can remote sensing and mechanistic ecosystem models be used to improve integrated assessments involving coupled humanforest dynamics? What are the opportunities for landuse strategies such as afforestation or woody bioenergy crop production to contribute to stabilization of atmospheric CO 2 concentrations? What are the linked remote sensing/ecosystem modeling requirements for improving integrated assessments of climate mitigation strategies? How could potentially altered disturbance rates affect vegetation, carbon stocks and fluxes, and the development of climate change mitigation strategies? GLM 23

24 Emerging Model-data Requirements 0.5x0.5 degree fractional coverage land-use maps and annual land-use transitions including wood harvest are now, or soon to be, in use by all ESMs. Next generation terrestrial models needing this information at 1 ha sub-annually in development for CMS+.

25 What should be done to move forward? Improve tracking of most sensitive factors, especially wood harvest, shifting cultivation, future scenarios Increase resolution and dependence on global land cover information, particularly satellite based and national inventories Prepare for next generation of fully coupled iesms Incorporate data on consequences of land-use transitions for LC Disaggregate land-use classes into multiple functional categories Develop harmonized land management information Monitor and develop parameterizations of future land-uses likely to be important e.g. biofuels, plantations, 25

26 Parking Lot

27 HYDE 3: Cropland & rangeland in 2000 (Klein Goldewijk et al. 2007) cropland Area (km 2 ) per 5 min. grid rangeland 0 < > 60 27

28 Global wood harvest by country, (Pg C y -1 ) HYDE 3 population Zon & Sparhawk (1923) per capita FAO Pg C y Hurtt et al. 2006; revised

29 Butler (1984) 29 Hurtt et al. (2011)

30 Hurtt et al. (2011) 30

31 Comparison with Results from Hurtt et al Diagnostic Reference data Hurtt 2006 LUHa Percentage of land surface impacted by human land-use activities (C+P+S) 42-68% 57.5% Total secondary land increase x 10 6 km 2 26 x 10 6 km 2 Percentage of secondary land increase that is forested 50% 51% Percentage of secondary land generated by wood harvest and shifting cultivation (permanent agriculture generated the rest) 70-90% 83.5% Wood harvest (including slash) Pg (Houghton 1999)*; 82 Pg (FAO) 77 Pg 82 Pg Wood clearing for agriculture Pg (Houghton 1999) Pg 170 Pg Area of forested land in shifting cultivation fallow (2000) 4.42 x 10 6 km 2 (FAO) x 10 6 km x 10 6 km 2 Rates of clearing land in shifting cultivation x 10 6 km 2 /yr (Rojstaczer et al. 2001) x 10 6 km 2 /yr 0.58 x 10 6 km 2 /yr Percentage of US Forests that are secondary (2000) 94-99% 100% Mean age of Eastern US Secondary Forests (2000) 38yrs * 71 yrs 63 yrs Total gross transitions (2000) 1.6 x 10 6 km 2 /yr 2.0 x 10 6 km 2 /yr Total net transitions (2000) 0.17 x 10 6 km 2 /yr 0.19 x 10 6 km 2 /yr Global Cropland Area (1990) 12.1 x 10 6 km x 10 6 km 2 Global Pasture Area (1990) 25.8 x 10 6 km x 10 6 km 2 Global Primary Land Area (1990) 57.7 x 10 6 km x 10 6 km 2 Hurtt et al. (2011) 31

32 Representative Concentration Pathways (RCPs) IAM RCP Radiative Forcing MESSAGE 8.5 >8.5 W/m 2 in 2100 AIM 6 ~6 W/m 2 at stabilization after 2100 MiniCAM 4.5 ~4.5 W/m 2 at stabilization after 2100 IMAGE 2.6 Peak at ~3 W/m 2 before 2100 and then declines to 2.6 W/m 2 by 2100 Concentration > ~1370 CO 2 -eq in 2100 ~850 CO 2 -eq at stabilization after 2100 ~650 CO 2 -eq at stabilization after 2100 Peak at ~490 CO 2 -eq before 2100 and then decline Pathway Shape Rising Stabilization without overshoot Stabilization without overshoot Peak and decline Moss et al. (2008) Towards New Scenarios for Analysis of Emissions, Climate Change, Impacts, and Response Strategies. IPCC, Geneva. 32

33 IAM Data IAM RCP Crop and Pasture Data Wood Harvest Data Includes Climate Feedbacks for Forest Regrowth? MESSAGE 8.5 Gridded Gridded No (only for allocation of crops) Biofuel treatment Included in wood harvest AIM 6 Gridded Gridded No Included in cropland MiniCAM 4.5 Regional Regional No Included in cropland IMAGE 2.6 Gridded Regional Yes Included in cropland 33

34 Harmonization Steps Develop consensus land-use history reconstruction Minimize differences between end of historical reconstruction and beginning of future projections Preserve as much information from IAMs on future as possible 34

35 35

36 36

37 Global area of crop, pasture, primary, and secondary land (10 6 km 2 ) primary non-forest revised wood harvest revised shifting cult. 0.5 x 0.5 HYDE primary forest PASTURE (HYDE 3) secondary non-forest secondary forest CROP (HYDE 3) URBAN (HYDE 3) Assume secondary = 0 Hurtt et al. (2011) 37

38 ΔF (LUH) crop pasture IMAGE AIM MESSAGE ΔF (IAM) 38

39 Simulated Atmospheric CO2 in GFDL ESM model Stouffer, pers. communication

40 Summary/Conclusions Virtually all ESMs either now track, or are developing capacity to track, gridded land-use changes. Modeling the effects of land-use in the Earth System requires the treatment of land-use effects consistently in the past, present, and future. The strategy described here is a nascent approach for harmonizing land-use transitions across multiple data sources, models, and scenarios. Future studies are needed to further constrain land-use reconstructions/models with data, address land mgt., and prepare for the next generation of fully integrated models (e.g. iesms+)