The aircraft utilized was the University of Tennessee Space Institute Piper Navajo (PA-31; N11UT) equipped with a forward looking infrared (FLIR) thermal imager, a Resonon Pika hyperspectral imager (400-1000 nm; 2.1 nm resolution), air temperature, dew point, and, downwelling infrared (IR), shortwave and photosynthetically active radiation (PAR). The flights were conducted during midday periods and covered the region in lawnmower-type patterns of straight north-south and slightly overlapping data transects. The sUAS’s were a DJI S-1000 octocopter equipped with a FLIR thermal imager and two iMet-XQ air temperature and relative humidity sensors, and a Microdrone MD4-1000 quadcopter, which was equipped solely with the two iMet-XQ sensors.
Sensible heat fluxes were estimated by a simple model using a correlation coefficient multiplied by the difference between the observed land surface temperature (from thermal imaging) to the sUAV/aircraft air temperature measurement near the top of the surface layer (~10% height of the Atmospheric Boundary Layer (ABL)) or from a higher altitude corrected with the near-surface lapse rates observed from the 1130 local daily balloon soundings. Latent heat fluxes over crops were modeled using the measured incoming PAR, the Normalized Difference Vegetation Index (NDVI) derived from hyperspectral red (630-690nm) and near-infrared (NIR; 770-900nm) bands, the water band index (ratio of the ~970 and ~900 nm bands), and adjusted with tower-based measurements of latent heat fluxes and soil moisture. NDVI relates vegetation health and coverage, as chlorophyll strongly absorbs red light while the cell structure of the leaves strongly reflect NIR light. The water band index is most sensitive to leaf water content. As the leaf water content increases, the absorption at ~970 nm increases relative to the absorption at ~900 nm. This water is found within the leaf mesophyll cells, and is strongly related to midday evapotranspiration rates during the growing season.
The coldest midday land covers were the deeper ponds and filled creeks, while the bare (tilled) soil and dry roads had the warmest observed surface temperatures. Surface temperatures of the crop covers, from coldest to warmest, were; mature soybeans, grasslands, immature soybeans, corn, wheat stubble and bare (tilled) soil. Some features showed opposite temperature trends depending on management. Older tilled soils were relatively warm while very recently tilled soil, observed during the tilling operations, were typically cooler than the surroundings. Wide trenches/ditches were very common in the low relief fields to reduce pooling water. Roughly half the farmers treated these runoff conduits with herbicide resulting in warmer bare soil, while the other half mowed or left the conduits covered with natural vegetation, resulting in surfaces were often cooler than the surrounding field.
While the trends among the aircraft, sUAV and tower-based measurements of surface temperature were very similar, we noted a distinct offset, where the aircraft and sUAV observed surface temperatures were, on average, ~2 deg. C higher than the concurrent tower based IR surface temperatures. We suggest that the aircraft and sUAV image coverages included the warmer field edges where herbivores, and water pooling next to roads, result in decreased crop cover categorized by a lower leaf area index. The fixed towers were located toward the centers of the respective fields to maintain a uniform eddy-covariance flux footprint area, and were not as subjected to edge effects.
Our measurement scales ranged from a pixel size of 2 mm for the UAV at 10 m altitude to a pixel of ~1 m for the aircraft at 2 km, with the mono-crop field size in the LAFE region averaging 0.12 square km (~29.5 acres). Our goal is to determine the grid spacing of land cover variables that approximate the important inhomogeneities in latent and sensible heat fluxes to the near surface air, which provide the local scales of air parcel buoyancies as a precursor to deep convection. Heterogeneity of land cover drives heterogeneity in these sensible and latent fluxes, alternately inhibiting or promoting deep convection. Finally, we note that due to limitations of flights in clouds, or high winds, or “bad” weather, these aircraft and sUAV data are typically biased towards “sunny” days.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner