Meso-scale Land Use/Land Cover Influences on Summer Precipitation Variations for the Midwest. U.S. Corn Belt, 2002–2011

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 9:15 AM
122BC (Phoenix Convention Center - West and North Buildings)
Andrew M. Carleton, Pennsylvania State University, University Park, PA; and B. Hall and B. Newman

Because agriculture in the Midwest U.S. “Corn Belt” is dominantly rain-fed, the monitoring and physical explanation of temporal variations (e.g., sub-seasonal, inter-annual) in growing-season precipitation is of critical importance. Although the synoptic-scale dynamical and thermo-dynamical features accompanying warm-season precipitation for the Corn Belt are now better understood (e.g., upwards vertical motion in the free atmosphere, spatial significance of the lifted condensation level metric), the contribution of meso-scale processes to the spatial and temporal variations of deep convection— particularly involving atmospheric feedbacks with land surface conditions, such as land use/land cover (LULC) and soil moisture— is still largely unknown. Improvements in local prediction of convective precipitation useful to the Corn Belt farming community and other stakeholders would be greatly facilitated by consideration of these meso-scale land surface-deep convection interactions.

Our previous work on this problem suggested that strong horizontal gradients in Corn Belt-region soil moisture and LULC— evident as long yet relatively narrow boundaries between extensive cropland and remnant forest— have associated statistically-significant increases in convective precipitation on days when the mid-tropospheric (“background”) flow is light (wind speed < ~10 m s-1), versus days when background winds exceeded that value; at least for the early-summer (15 June to 15 July) periods of 1996-99. However, unresolved questions from that study include the following: (1) Are the short-period results confirmed over the longer-term (i.e., climatologically)?; (2) Do different types of LULC boundary (e.g., urban-rural, water bodies, moist-dry soil, crops-forest) influence convective precipitation differently?; and (3) Is there evidence for a Corn Belt LULC boundary-deep convection feedback for the longer summer season, and how does it vary inter-annually?

To address the above questions, we undertook a climatological (10-year, 2002-2011) analysis of LULC and precipitation associations for the Corn Belt extending eastwards from Iowa to western Ohio, and bounded on the south by the Missouri and Ohio rivers and in the north by approximate latitudes 42o-44oN. Using image analysis of weekly Normalized Difference Vegetation Index (NDVI) data for the 2002-2011 early-to-mid-summer periods of 15 June to 31 July— when boundaries are most evident— the strong reflectance gradients comprising different LULC boundary types were identified and mapped in a GIS. This procedure permits the generation of separate map layers portraying the distribution, density and other attributes of individual LULC boundary types, their region-wide statistics (e.g., number of boundaries, areal coverage, etc.), and associations with station precipitation data.

To determine convective precipitation spatial and statistical associations with Corn Belt LULC boundaries, we used the high spatial resolution cooperative observer precipitation network (COOP) of stations. For the study period, these were available only as image copies of the original hand-written records via pdf from http://www7.ncdc.noaa.gov/IPS/coop/coop.html. We quality-controlled the COOP data to eliminate short-period or occasional (e.g., weekend) stations, duplicate records, and stations whose data availability did not exceed at least 90% for the 6-week early-summer period. Because we were most interested in the potential role of LULC boundaries in enhancing convective precipitation during the peak solar heating time of local afternoon through early evening, we determined these “convective days” from analysis of hourly NEXRAD imagery, and eliminated from consideration those stations recording precipitation daily totals outside the 05-09 local time on each following day. This culling of precipitation records left a total of 679 stations across the Corn Belt, suitable for subsequent analysis. A synoptic analysis of daily background flow led us to stratify station precipitation data separately for 33 “weak flow” convective days and 37 “strong flow” convective days in the 10-year study period.

The GIS analysis reveals that, far from being a homogeneous or amorphous landscape, the Corn Belt exhibits much meso-scale heterogeneity as LULC “permanent” boundaries comprising cropland-forest, urban-rural, and water bodies. An initial analysis of precipitation data from a 172 COOP station sample reveals strong suppression of convective development associated with water bodies that is confirmed in MODIS case images from the Terra/Aqua satellites. Surprisingly, precipitation was also suppressed along and near the crop-forest boundaries on weak-flow days but was enhanced at these same locations on strong flow days, for this smaller sample of stations. Analysis of the full suitable COOP station record possessing a more complete spatial coverage for the Corn Belt, is ongoing; both in terms of clarifying the climatology, and for evaluating summers of strong precipitation anomalies, in the 2002-2011 period. The results of those analyses should help yield a more complete picture of deep convection and its temporal (sub-seasonal and inter-annual) and spatial variations for the Corn Belt, particularly the role of meso-scale LULC forcing in context of the larger (synoptic) scale atmospheric mechanisms.