Agroecosystems /
Carbon Sequestration
Research Focus
Nebraska Phase I Progress Report
Phase I: Methodology
Phase I: Methodology
The following sections describe the development of databases needed for a Century analysis. Databases of climate, soils, irrigation and land management were compiled from various sources and enable the modeling of complex agriculture cropping systems across the state. The process provides the state with information, and integrates information at varying scales from the sub-county to regional. This provides reasonable estimates at the selected regional scale, but has limited use for the local land manager. Additional data being collected for the Phase II assessment will provide more complete information at the county level and enable the local land manager to make accurate estimates of C changes due to land management decisions.
Climate and Soil Databases
Data on climate, soils, landuse and management practices used in the analysis were assembled from a variety of sources. Individual counties are the spatial unit for representing climate factors. In other words, counties were assumed to be homogeneous with respect to temperature and precipitation.
Monthly temperature (mean monthly maximum and minimum) and precipitation (monthly total) values were obtained from the Parameter-Elevation Regressions on Independent Slopes Model (PRISM) climate dataset (Daly et al., 1994). PRISM uses point data from the U.S. network of weather stations and a digital elevation model (DEM) to orographically adjust climate variables for 4 km grid cells across the coterminous U.S. The data used in our analysis consisted of long-term (1961-1990) monthly averages (Figure 10 & 11). Area-weighted mean values of monthly temperature and precipitation variables were calculated for each county.
Figure 10: PRISM Average Annual Temperature (1961-1990)
Figure 11: PRISM Average Annual Precipitation (1961-1990)
Figure 12: STATSGO Sandy Soil Distribution
Figure 13: STATSGO Silty Soil Distribution
Figure 14: STATSGO Clay Soil Distribution
Figure 15: STATSGO Surface Soil Texture Distribution
Figure 16: STATSGO Hydric Soil Distribution
Figure 17: Adams County Modeled Soil Textures
Figure 18: Lincoln County Modeled Soil Textures
The Nebraska Statewide Pivot and Surface Irrigation geospatial database - (N)ebraska (I)rrigation from (M)ultiple (D)ata (S)ources (NIMDS) - is a combination of two existing Arc/Info (ESRI, 2001) datasets: The (CO)operative (HY)drology (ST)udy in the Central Plattte River Basin (COHYST) and the drought mitigation dataset (WILHELMI) developed by Wilhelmi, 1999. These data were provided by the University of Nebraska, Lincoln, Center for Advanced Land Management Information Technologies (CALMIT). The NIMDS irrigated acreage for the state is calculated at 7,859,000. Comparison of NIMDS irrigated areas by county to the 1997 National Resource Inventory (NRI) data and the 1995-1999 National Agriculture Statistics Service (NASS) data show a high correlation between the different data sources (NRI, 1997; NASS, 1995-1999). The RMS error for NRI is 0.92 and for NASS is 0.97.
The COHYST geospatial database is a pivot and surface irrigation inventory that covers 43 of the 93 Nebraska counties (4,136,472 acres). The data were developed from 1997 LandSat Thematic Mapper (TM) Imagery. Twenty-three of the forty-three counties were field checked for accuracy by local Nebraska Resource Districts (NRD's). However, only 20 counties have complete coverages. COHYST consists of two separate coverages - pivot features and surface irrigation features. These were processed into a single theme using the Arc/Info UNION command. A common attribute, "irrigation type", was added prior to merging to denote the original themes irrigation method - pivot or surface. This merged COHYST data was used as a starting point for the final statewide irrigation coverage NIMDS.
WILHELMI data was used to supplement irrigation information for those areas missing from the COHYST coverage. The WILHELMI irrigated layer was developed using 1991 through 1993 growing season LandSat Thematic Mapper (TM) Imagery at a 40 meter resolution. Irrigated areas were compiled using ArcView heads-up digitizing in the Albers Conic Equal Area Projection, NAD 27. The projection parameters are not documented, however, it is thought that the Central Meridian and Standard Parallels are referenced to the coterminous United States. Data error was calculated by comparing the digitized areas by county to state statistics. Overall, the difference between WILHELMI and state statistics was within 0 to 8 percent. The final shape file was then converted to an Arc/Info coverage.
According to the state experts, the WILHELMI data is believed to over-estimate irrigation for Nebraska (G. Henebry, personal comm). This data error may be due to digitizing methods or imagery misclassification due to differences in time of the Landsat data. Visual inspection of the WILHELMI data show that many of the irrigation polygons are distorted - pivots that should be circular are represented with odd shapes and angles. This error is likely from the conversion of the original shape file to a vector coverage. The tolerances that default to the bounding area of the shape file were set too large resulting in data generalization during the conversion. Additionally, some pivot and surface irrigated areas were digitized as large irregular shapes instead of detailed individual polygons as compared to the same areas in the COHYST layer. Attribute values that determine irrigated and non-irrigated polygons are missing in the WILHELMI dataset resulting in no distinction between surface, pivot, and no data areas in the coverage. For our use, a major problem with the WILHELMI data was the geographic feature positioning mis-alignment as compared with other data sources. Irrigation polygons are incorrectly positioned in relation to cropland areas from the GAP land cover dataset (GAP, 1993) and the same irrigated areas from the COHYST theme. This error is more apparent in the western half of the state where polygons are shifted to the southwest. The positioning mis-alignment was likely caused because the selected projection is problematic for states like Nebraska that lie in an east/west plane and the projection parameters were defined incorrectly compounding the data error as the features move out from the Central Meridian defined at 96 degrees.
Before the WILHELMI data was used, the data problems needed correction. Using the Arc/Info ARC EDIT module, a new attribute, "irr-code", was added to the coverage. The WILHELMI dataset pivot and surface polygons were assigned a single irrigation attribute by selecting all polygons smaller than the largest background polygon in the coverage. This procedure identified the majority of the irrigated polygons. However, small island polygons representing spaces between irrigated areas are still erroneously labeled as "irrigated" (Figure 19). These attribute errors were determined to be acceptable due to manpower limitations. After the WILHELMI data were attributed, polygon information inside the COHYST boundary were deleted using the Arc/Info ERASE command. This step removed 7,088,000 hectares (17,508,000 acres) in the WILHELMI coverage. Upon visual inspection of the remaining polygons outside of the COHYST boundary, it was determined that only the southwest corner of the WILHELMI dataset was significantly mis-aligned in relation to the GAP cropland data.
Figure 19: WILHELMI Irrigation Polygons
A boundary layer of this southwest region was generated and used as the clipping and erase area with the Arc/Info CLIP and ERASE commands. A total of 855,000 hectares (2,113,000 acres) were removed from the southwest region of the altered WILHELMI theme and saved to a new coverage (Figure 20). The remaining 12,089,000 hectares (29,861,000 acres) in the northern and eastern areas outside the COHYST boundary in the altered WILHELMI layer received no further adjustment. A statewide irrigation control data layer was developed by extracting cropland areas from the GAP land cover grid using the Arc/Info SELECT command. Visual analysis of the GAP cropland areas showed circular polygon features that could be easily identified as irrigation pivots. Additionally, these pivot features visually coincided with the COHYST pivot features. The resulting cropland grid data were filtered to remove data noise and then converted into an Arc/Info vector coverage. The southwest WILHELMI coverage was georeferenced using the GAP cropland coverage as the ground control. Tics were visually identified and digitized as TO and FROM positions for the Arc/Info ADJUST command. These tics were used to shift groups of polygon features in the southwest WILHELMI data subset to their correct geographic position (Figure 21). A total of 785 TO and FROM tics were manually generated to correct the positional errors found within the southwestern half of the WILHELMI theme.
Figure 20: Statewide Dataset Development Areas
Figure 21: Southwest WILHELMI Irrigaton Adjustment
Landuse/Land Cover Geospatial Database
The Nebraska Landuse/Land Cover raster dataset is a digital product based on the draft statewide Land Cover Map of Nebraska (GAP, 1993) and the Nebraska Pivot and Surface Irrigation (NIMDS) coverage. The original 20 class GAP raster dataset was generalized using the Arc/Info RECLASS command resulting in the following classifications: Cropland, Range, Fallow, Forest, Urban, Bottom Land, and Water (Table 1).
Table 1: Reclassification of GAP Land Cover Data.
|
Irrigated land cover classes were identified by masking the reclassed raster using the irrigation grid at 30 meters resolution developed from the Nebraska Pivot and Surface Irrigation (NIMDS) coverage. The agricultural landuse classifications inside the irrigated areas were assigned to: Irrigated Cropland, Irrigated Hayland, and Irrigated Fallow. Any irrigated areas classified as Forest, Bottom Land, Urban, or Water were dropped out of the irrigation area. This adjustment resulted in 182,000 acres being removed from the irrigation area. The difference in acreage between the NIMDS irrigation raster (7,858,757) and the final product's irrigated land acreage (7,676,000) can be attributed to an over estimate of Nebraska's irrigated lands in the source coverage.
The final product (Figure 22) was generated by merging the irrigation land cover raster with the reclassed GAP land cover raster. Nebraska landuse/land cover classification statistics are documented in Table 2.
Table 2: Landuse/Land Cover Areas.
|
Figure 22: Nebraska Landcover Distribution
The information on historical cropping practices needed for Century simulations was gathered from a variety of sources with differing scales of coverage, from the experiences of a single farmer (Miner 1998) to national level databases (NASS, 2000). Figure 23 illustrates how national databases provide state and county values and trends for crops grown over time.
Figure 23: Crop Changes From Plow Out To Present
In 1997, the state was divided in thirteen regions based on major land resource areas (MLRA's) as defined by USDA-NRCS. Figure 24 details the MLRA boundaries for Nebraska. Individual counties were associated to each region based on area weighting and communications with NE NRCS technical specialist (Figure 25). The modified MLRA regions provided the basis for the development of cropping and management systems within each region. MLRA 65 was subdivided into an east and west region to account for the landuse change from rangeland to irrigation in the western part of the Sand Hills. Histories have been constructed on modified MLRA regions for use with the Century model in an attempt to closely chronicle the opening of the agricultural lands, the changes in dominant crops, tillage practices, residue management and inputs to the soil (Table 3).
Table 3: Historic Crop Rotations by NE Modified MLRA Century Regions.
|
|||||||||||||||||||||||||||||||||||||||||||||
Figure 24: Major Land Resource Areas (MLRA)
Figure 25: Nebraska Modified MLRA Regions
Modern crop rotations (1970- present) were developed for each modified MLRA based on 1997 NRI data. These include non-irrigated and irrigated systems, which represent the dominant crop rotations in each region. Individual crop files were developed to represent higher grain yields, increases in plant bio-mass and nitrogen fertilizer additions (USDA-ERS, 1992). To represent the change in land management due to the Conservation Reserve Program (CRP), the non-irrigated crop rotations were also modeled showing the conversion to grass starting in 1986. Irrigation was represented in the model starting in 1970 and provided all the necessary water required by the plant to meet potential evapotranspiration. Region 65W is the area where rangeland was converted to irrigated cropland in the 1970 and is dominated by the continuos corn crop rotations with lesser amounts of other irrigated crop rotations. Table 4 details the crop rotations and if irrigation was modeled for each region.
Three tillage regimes (intensive tillage, moderate tillage and no tillage), were simulated for each rotation from 1986-2000 in each county. All prior cultivation is based on intensive tillage. Intensive tillage was defined as multiple tillage operations every year, including significant soil inversion (i.e., plowing, deep disking) and low surface residue coverage. This definition corresponds to the intensive tillage and ‘reduced’ tillage systems as defined by Conservation Technology Information Center (CTIC, 1998). No tillage was defined as not disturbing the soil except through the use of fertilizer and seed drills and where no-till is applied to all crops in the rotation. Moderate tillage made up the remainder of the cultivated area, including mulch tillage and ridge tillage as defined by CTIC (CTIC, 1998) and intermittent no-till. After year 2000, all crop rotations, except non-irrigated wheat-fallow, were no tilled so we can show the maximum effects due to change in management. The wheat-fallow rotation became a wheat-corn-fallow crop rotation under no tillage management system to reflect intensification of the cropping system.
Table 4: Modern Crop Rotations By NE Modified MLRA Century Regions.
|
|||||||||||||||||||||||||||||||||||||||||||||