**3. Results**

#### *3.1. Targeted UAV Measurements over Mixed Hardwood Forest*

We examined the temporal consistency of albedo estimates across flights and years in a series of flights over a mixed hardwood forest in Tully, NY. The series of five flights spaced hourly around local solar noon measured a summer forest albedo of 0.145 ± 0.005 SD, *n* = 5, and ranged from 0.140 to 0.146. The mixed hardwood albedo at solar noon, 0.145, was the same as the mean and was consistent with albedo values recorded over the course of the day.

#### *3.2. Comparative Tower and Satellite Measurements over Mixed Hardwood Forest*

UAV albedo values were compared to tower measurements from three similar mixed northern hardwood forests (Figure 4).

**Figure 4.** Boxplots showing albedo over mixed temperate forest as measured by UAV (red) and MODIS satellite data (blue) in (**a**) Tully, NY and by fixed towers (red, 0.145 ± 0.003 SD, *n* = 5) and MODIS satellite data (blue, 0.152 ± 0.005 SD, *n* = 70) at (**b**) Durham (red, 0.121 ± 0.015 SD, *n* = 12; blue, 0.151 ± 0.006 SD, *n* = 23), (**c**) Bartlett (red, 0.155 ± 0.008 SD, *n* = 13; blue, 0.148 ± 0.008 SD, *n* = 18), and (**d**) Petersham (red, 0.133 ± 0.009 SD, *n* = 12; blue, 0.145 ± 0.008 SD, *n* = 18). Box and whisker plots show medians, data quartiles, and outliers. Letters over tower and satellites represent significant differences within UAV and tower measurements at Tully, Durham, Bartlett, and Petersham (calculated by ANOVA, Tukey HSD), and within satellite measurements at each of the same (calculated by ANOVA, Tukey HSD).

The average summer albedo at the Durham, NH tower was lower than UAV measurements by 0.02, while albedo at the Bartlett tower was higher by 0.02. Albedo at the Petersham, MA tower was lower by 0.01. While differences between each of the four sites were small, they were statistically significant, showing clear across-site heterogeneity (Type II Anova, df = 3, *F* value = 271, *p* < 0.001). At all sites, albedo at solar noon was within 0.01 units of the five hour albedo, and was not significantly different from the full albedo. Overall, all sites fell within the needed accuracy of 0.02–0.05 albedo units of each other. In comparison, satellite albedo data varied little across sites, with only Tully and Petersham showing a significant difference of less than 0.01 units, showing very little across site heterogeneity.

We then compared the in-situ UAV and tower measurements to albedo measurements made by satellite. At the Tully site, MODIS average albedo was slightly higher than in-situ, UAV-measured albedo (Student *t*-test: df = 8.05, *t* = −6.34, *p* < 0.001). In Durham, satellite albedo was also significantly greater than in-situ, tower-measured albedo values (Student *t*-test: df = −15.3, *t* = 23.1, *p* < 0.001). Likewise, Petersham satellite albedo was significantly greater than the in-situ measurements by tower (Student *t*-test: df = 21.2, *t* = −6.03, *p* < 0.001). However, at Bartlett, the average albedo value measured by satellite was significantly less than the average tower albedo (Student *t*-test: df = 18.0, *t* = −4.60, *p* < 0.001). Both UAV showed similar consistency (0.01 albedo units lower) with satellite measurements as tower measurements had with respective satellite measurements (Durham: 0.03 lower, Bartlett: 0.01 higher, Petersham: 0.01 higher). Overall, the coefficient of variation across all summer UAV albedo measurements made over mixed hardwood forest at Tully (2.1%) was similar to the variability observed at the in-situ estimates made by towers (Durham: 1.9%, Bartlett: 2.7%, Petersham: 1.9%).

#### *3.3. Validation Measurements comparing Simultaneous UAV and Tower Data*

We compared albedo measurements taken by tower and UAV approaches, over the same field of willow biofuels. Initial side by side flights by both methods produced closely matched albedo estimates (Figure 5).

**Figure 5.** Albedo was measured by UAV over a willow biofuel field at three sites: immediately adjacent to a fixed tower (0.208 ± 0.001 SD, *n* = 19), 24 meters east of the tower (0.208 ± 0.002 SD, *n* = 23), and 24 m west of the tower (0.202 ± 0.002 SD, *n* = 21). Tower measurements were taken simultaneously with each flight from a fixed location at center (0.209 ± 0.001 SD, *n* = 19; 0.210 ± 0.001 SD, *n* = 23; 0.211 ± 0.001 SD, *n* = 21). UAV measurements are depicted in red, while paired, simultaneous fixed tower measurements are depicted in blue. Box and whisker plots show medians, data quartiles, and outliers.

Measurements were repeated over the same willow field, using the UAV in flights 24 m west and then 29 m east of the tower. All UAV-derived willow albedo measurements were well within ±0.01 of each other, but there was slightly greater correspondence between the side-by-side measurements and measurements made over willow a distance from the tower. UAV albedo tended to have greater variability than tower albedo, regardless of the sub-site, although overall variance was low.

#### *3.4. UAV Measurements Demonstrating Albedo across Multiple Scenarios of Land Use*

Three land use types were examined in parallel: surveys of monoculture of Norway spruce (Figure 6) measured an albedo of 0.0743 and 0.0824. Two flights resurveyed the same forest site and an adjacent point, as described above, with albedo of 0.149 and 0.154. Three flights over a cropped willow field, as described above, were measured at 0.208, 0.208, and 0.202 albedo. The variation across sites was an order of magnitude less than the variation across the different land uses.

**Figure 6.** Albedo measured by UAV for three different land use types: two adjacent coniferous forest sites at Tully, NY (0.074 ± 0.002, *n* = 24; 0.082 ± 0.002, *n* = 23) (red), two adjacent mixed northern hardwood forest sites at Tully, NY (0.149 ± 0.002, *n* = 22; 0.154 ± 0.002, *n* = 23) (blue), and three adjacent cropped willow sites at Geneva, NY (0.208 ± 0.001, *n* = 19; 0.208 ± 0.002, *n* = 23; 0.202 ± 0.002, *n* = 21) (gray).Errorbarsrepresentthevariance,reportedasstandarderror,withineachflight.
