Reader Beware Land Use Classification Nomenclature Issues Presented

Reader Beware! Land Use Classification Nomenclature Issues Presented at Sustainability & Land Use Change Workshop 27 September 2018 Sharon Bard Centrec Consulting Group

What prompted this exploration? • ILUC project • Which classification nomenclature to use? • How can our work help advance knowledge in this area? • As we dug into classification nomenclature and suggestions made by the Advisory Team, issues surfaced: • • Inconsistent land cover classification Inconsistent land cover categorization Census of Ag: cropland pasture versus permanent pasture Implications of these issues on research output / conclusions

Dimensions of Classification of Land Cover and Land Use PRIMARY • Various US: • USDA NASS C of Ag • USDA CDL • GCL 2000 • CORINE Data Collection Methods • Remote sensing • Surveys • On-ground • GTAP • GLOBIOM CATEGORIZED • FAO Modeling Land Cover &/or Land Use Data Sets Classification Nomenclature Output Analysis • IGBP • UN LCSS • Intergovernmental Panel on Climate Change (IPCC) Conclusions

Sources of Error in LC/LU Classification • Inappropriate dataset • Incorrect assumptions • Inconsistent categorizations • Missing data points • Classification error Modeling Data Collection Methods • Spatial & spectral resolution of satellite imagery • Temporal • Respondent / survey error Land Cover &/or Land Use Data Sets Classification Nomenclature Output Analysis • Inconsistent terminology • Vague definitions • Complicated or incomplete nomenclature Conclusions

Classification Nomenclature Key Issues • Missing data points • Accuracy of data • Proper accounting for double cropping / intensification • Complexity of delineating land cover vs land use vs land management • For example, CRP land • Aggregation of land cover types • Land use change methodology • Explicit categories omitted • Defining marginal land

Classification Nomenclature Key Issues: Examples Grassland/ pasture Permanent pasture Cropland Idle/fallow cropland Grassland Alfalfa Cultivated Hay ground Other Non cultivated (grass)

Focus: Respondent Misclassification PRIMARY • Various US: • USDA NASS C of Ag • USDA CDL • GCL 2000 • CORINE Data Collection Methods • Spatial & spectral resolution of satellite imagery • Temporal • Respondent / survey error • GTAP • GLOBIOM CATEGORIZED • FAO Modeling Output Land Cover &/or Land Use Data Sets Analysis Classification Nomenclature Conclusions

Bridging Data: Source to Use USDA Census of Ag Cropland harvested Crop failure Cultivated summer fallow FAO Cropland used for crops Idle cropland Total cropland Arable land & permanent crops GTAP Cropland Total Ag Area Cropland pasture Pasture excl. CLP & woodland pasture Pasture (rented by AUM or not in farms) Grassland pasture & range Permanent pasture & meadows Pasture land

Focus: Cropland Pasture vs Permanent Pasture • From footnote in ERS Table: “Between 2002 and 2007, total cropland decreased by 34 million acres to its lowest level since this series began in 1945, even though harvested cropland (which accounts for most land planted to crops) increased 5 million acres due to a recovery of failed cropland from severe droughts in 2002. A 26 -million acre decline in cropland pasture contributed to this trend, partly due to methodological changes in the 2007 Census of Agriculture that reclassified some cropland pasture to permanent grassland pasture and range. ”

Focus: Cropland Pasture vs Permanent Pasture Cropland Pasture – generally considered to be in a long-term rotation; includes crop acres that have been hogged or grazed but not harvested; some land used for pasture that could have been cropped without additional improvement Permanent Pasture and Rangeland – all open land used primarily for pasture and grazing, including shrub and brushland types of pasture, grazing land with sage brush…. .

Focus: Cropland Pasture vs Permanent Pasture • Potential causes of the change in acres between categories • Non-measured errors or reporting artifacts (USDA reporting error) • • • Respondent and enumerator error Processing error Item nonresponse Record matching error Model uncertainty error • Respondent misclassification of land cover in either Census year • Missing information – non-response or under-coverage • Actual land cover change

Change between 2002 and 2007 in: Cropland Pasture: Decrease Permanent Pasture: Increase Courtesy: Ploughman-Analytics

Change between 2002 and 2007 in: Cropland Pasture: Increase Permanent Pasture: Increase Courtesy: Ploughman-Analytics

Focus: Cropland Pasture vs Permanent Pasture Potential Implications • Problematic “fuzzy” data • Changes in the western states => large number of acres • Would use aggregate numbers of cropland pasture and permanent pasture and trends with caution • “Reclassification” impacts available cropland acres • What should be done about historical classification? • May impact modeling efforts that account for intensification • Intensification – increasing production on existing area through yield improvement per harvested area, greater use of double cropping and/or cultivating the existing unused marginal land

Focus: Inconsistent Nomenclature PRIMARY • Various US: • USDA NASS C of Ag • USDA CDL • GCL 2000 • CORINE Data Collection Methods • Remote sensing • Surveys • On-ground Modeling Land Cover &/or Land Use Data Sets Classification Nomenclature Output Analysis • IGBP • UN LCSS • Intergovernmental Panel on Climate Change (IPCC) Conclusions

Focus: Inconsistent Nomenclature

Focus: Inconsistent Categorization PRIMARY • Various US: • USDA NASS C of Ag • USDA CDL • GCL 2000 • CORINE Data Collection Methods • Remote sensing • Surveys • On-ground Modeling Land Cover &/or Land Use Data Sets Classification Nomenclature • Researcher categorization of land cover data • IGBP • UN LCSS • Intergovernmental Panel on Climate Change (IPCC) Output Analysis Conclusions

Focus: Categorization of CDL Data

How can the errors be minimized? • Increase the accuracy of land cover classification • Increase the application of standardized classification nomenclature for land cover across data sets • Increase the use of standardized methodology for land use change • Be aware of the issues / limitations of the data / methods and account for them in the research

Questions / Suggestions for further exploration? Thank you!