|
|
Effect of NOAA satellite orbital drift on AVHRR-derived phenological metrics
The U.S. Geological Survey, Earth Resources Observation and Science Center routinely produces a remote sensing phenology (RSP) dataset derived from a 1-km Advanced Very High Resolution Radiometer (AVHRR) product compiled from six National Oceanic and Atmospheric Administration (NOAA) satellites (NOAA-11, -14, -16, -17, -18, and -19). Each NOAA satellite experienced orbital drift that influences the AVHRR reflectance measurements. To understand the effect of the orbital drift on the RSP dataset, we analyzed the impact of solar zenith angle (SZA) on RSP metrics in the conterminous United States. The AVHRR weekly composites were used to calculate the growing-season median SZA from 1989 – 2014. The results showed that the SZA values in 1992, 1993, 1994, 1999, and 2000 are substantially higher than those in other years. The high SZA (44° – 60°) in those years caused negative trends in the SZA time-series that were statistically significant (α = 0.05) in 76.9% of the study area. Temporal correlation analysis showed that the phenological metrics and SZA were significantly correlated in 4.2 – 20.3% of the study area. After eliminating the five years with high SZA (>40°), the temporal SZA trend was largely reduced, significantly affecting only 1.9% of the area. The correlation coefficients between the phenological metrics and SZA were significant in 1.0 – 6.2% of the area. Our study concluded that the NOAA satellite orbital drift increased the SZA, and in turn, influenced the phenological metrics. Eliminating the years with high SZA greatly reduced the influence of orbital drift on the RSP time-series.
Users of these data sets should cite this DOI: https://doi.org/10.5066/F7G73BVV.
Data set name |
Download (data and metadata) |
Data Layer |
Description |
AVHRR growing season median SZA images (1989-2014)
|
avhrr_sza_gs_
median_1989to2014.tif |
26 data layers corresponding to 26 years from 1989 to 2014) |
The SZA weekly composites from AVHRR dataset were used to calculate the growing season SZA for each year from 1989 to 2014. The growing-season SZA is the median SZA value of all available weekly composites within the growing season of each year. The growing season is defined as the time period between the spring equinox and the autumn equinox (around March 20 and September 23, respectively). |
AVHRR SZA trend slope (1989-2014)
|
avhrr_sza_trend
_slope_1989to2014.tif |
2 data layers: (1) temporal trend from 1989 to 2014; (2) temporal trend from 1989 to 2014 (excluding 5 high SZA years) |
The trend analysis was performed by applying the linear regression to 26 SZA time series images. The dependent variable SZA was regressed on the independent variable year (yr): SZA = a + b(yr) + e where, a is the intercept, b is the slope, and e is the random error. The slope b indicates the rate of the SZA change through time (year). The second trend analysis was undertaken for the SZA images after the 5 years (1992, 1993, 1994, 1999, and 2000) with high-SZA were eliminated from the time series. |
AVHRR SZA trend significance (1989-2014)
|
avhrr_sza_trend
_sig_1989to2014.tif |
2 data layers: (1) temporal trend from 1989 to 2014; (2) temporal trend from 1989 to 2014 (excluding 5 high SZA years) |
The significance of the slope was tested using the F-statistic with H0: b = 0. Based on the F-values and p-values, The significance level (alpha = 0.05) of the regression was determined and mapped. The second trend analysis was undertaken for the SZA images after the 5 years (1992, 1993, 1994, 1999, and 2000) with high-SZA. were eliminated from the time series. |
AVHRR SZA - metrics correlation coefficient (1989-2014)
|
avhrr_sza_metrics
_corr_coef_1989to2014.tif |
9 Data layers corresponding 9 phenological metrics: SOST, SOSN, EOST, EOSN, MAXT, MAXN, DUR, Amp, and TIN |
The Pearson correlation analysis was performed for growing-season median SZA images and phenological metrics images from 1989 to 2014. The output images are the correlation coefficients through SZA and phenological metrics time series at pixel level. |
AVHRR SZA-metrics correlation coefficient (1989-2014, excluding 5 high-SZA years)
|
avhrr_sza_metrics
_corr_coef_1989to2014
_excl5years.tif |
9 Data layers corresponding 9 phenological metrics: SOST, SOSN, EOST, EOSN, MAXT, MAXN, DUR, Amp, and TIN |
The Pearson correlation analysis was performed for growing-season median SZA images and phenological metrics images from 1989 to 2014 (excluding 5 high-SZA years). The output images are the correlation coefficients through SZA and phenological metrics time series at pixel level. |
AVHRR SZA-metrics correlation significance (1989-2014, excluding 5 high-SZA years)
|
avhrr_sza_metrics
_corr_sig_1989to2014
_excl5years.tif |
9 Data layers corresponding 9 phenological metrics: SOST, SOSN, EOST, EOSN, MAXT, MAXN, DUR, Amp, and TIN |
The significance of the correlation coefficient between SZA and phenological metrics (1989 – 2014, excluding 5 high-SZA years) was tested using the t-test with H0: r = 0, which resulted in the t-values and the associated p-values. Based on the p-value, the significance of the correlation was determined at alpha = 0.05 level. |
|