In south central Washington, sunny days and clear nights in early spring often mean growers need to protect their crops from potential frost injury. In April and May 2006, we flew nighttime thermal infrared imagery over a collaborator’s orchards in order to see patterns of air temperature distributed by wind machines. We were surprised to see areas in two blocks that were noticeably colder than the surrounding trees. These colder areas revealed in the thermal infrared imagery turned out to be rows with Extenday reflective ground cover installed. As an example, we can see this effect in the thermal infrared image (see "Infrared image shows orchard temperatures"), which was collected about 4:30 a.m. on May 2.

The block in the center of the image is planted to Granny Smith apples on a V-trellis system. You can see that the wind machine was operating (the bright spot near the center), and see the warm air forced through the block. Near the top of the block are six rows where Extenday was installed on the ground. Based on the imagery, the average temperature of the entire block is about 35.6°F. A subset in the Extenday rows (labeled AR02 in the image) is colder, averaging 33.5°F. The area labeled AR04 just below the Extenday rows has an average temperature of 36.6°F. When first presented the imagery in April, the grower thought we were looking at some sort of artifact, not a measurable difference from the use of Extenday.

To evaluate any real differences, we instrumented two sites in the Granny Smith block, one in the Extenday area and the other in a row without the fabric. At both sites, we took air measurements at several heights in the trees, and we installed a temperature sensor one inch below the surface of the soil (between the tree rows). Flower blossom temperatures were taken by inserting a thermistor into a blossom. These blossom measurements were only used for the nighttime temperatures as sunlight on the sensors bias the daytime measurements. All of the measurements were recorded at one-minute intervals from April 30 to May 3.

To try to make sense of all of this data, we graphed temperature differences between the two sites, through time. We included an example graph (see "Temperatures with Extenday and without"), showing the blossom and soil temperature differences for the night of May 2. The measurements were subtracted so that positive differences correspond to the Extenday site being warmer. In the graph, we see that the soil temperatures at one inch below the surface are consistently 3°F colder for the Extenday. By early morning, flower temperatures were generally colder for the Extenday site. Note that if we had measured 15-minute average temperatures (see the line with open circles), we would have missed many of the biggest differences in temperature. The grower is employing a WSU AgFrostNet system, and has a nearby temperature monitoring site in a cold drainage area near the block. But we can see from the dotted line that the flower temperatures at the Extenday site actually get colder than this reference point.

The temperature responses using a reflective ground cover will undoubtedly vary, depending on many factors, including location, time of year (including seasonal variations), crop, trellis system, topography, and soil moisture levels. Additional influence on the temperature response will come from how the reflective ground cover is installed and the percentage of the surface area covered. We are planning more tests with Extenday to evaluate these effects, as well as to better understand the requirements for sensor deployment to characterize changes both spatially (i.e., throughout a block) and through time.

Decision making

To stay competitive, growers are continuously making changes in crop management. Based on our ongoing research in sensor networks, we believe that the location and timing of measurements may impact decision making (e.g., whether to turn on the wind machine). In this case, the grower needs to be aware that the AgFrostNet location may no longer be colder than the flower or canopy temperatures. In situ (in this case in the canopy) measurements may well give different answers than previous approaches.

As another example, the thermal infrared imagery suggests that the ground surface temperature of the road outside the block may be much warmer than inside the block. So, an air temperature measured from a truck parked outside the block might be much warmer than the actual temperatures in the block.

Timing of the measurements may also have an impact on decision making. One-minute data may give more accurate representation than standard 15-minute averaged data. For critical, dynamic decision making such as frost protection, 15-minute data may not be adequate. We need to perform more experiments to verify the effects we saw in April and early May of 2006, but these initial results suggest that your orchard could be even colder than you think it is.