Albedo measurements

4.2.1 Stationary tower measurements

Micrometeorological towers are usually equipped with an upward-facing pyranometer to measure downwelling (incoming) SW radiation. Some stations also measure upwelling (reflected) SW radiation. Thus, surface albedo can be directly calculated as the ratio of upwelling to downwelling radiation over any time interval. This is a robust way to determine site-level albedo. Continuous data from stationary towers include diurnal and seasonal variations in solar angle, atmospheric conditions and surface properties, and they often span multiple years. However, continuous measurements are expensive and only available at selected sites. Research stations belonging to national networks such as SITES (Swedish Infrastructure for Ecosystem Science), regional networks such as ICOS (Integrated Carbon Observation System) or global networks such as FLUXNET perform long-term monitoring that is representative of major ecosystem and climate types. In Sweden, only a few stations are located on agricultural land and some do not provide data on upwelling SW radiation. Even in FLUXNET, few albedo measurements are available for site pairs with similar environmental conditions, but differing management practices.

Measurements for willow used in Papers I and II were obtained from a research project on SRC willow run by the University of Gothenburg. The data covered a full three-year cutting cycle of the plantation. Albedo of the reference, long-term green fallow, was approximated by a mire vegetated with grasses and sedges, which was part of SITES. Paper I included a second case, comparing coniferous forest with clear-cut. Data for coniferous forest were obtained from SITES and for clear-cut from a research project on forest harvesting at Lund University.

Downwelling and upwelling SW irradiance were received as 30-min averages. Raw data were cleaned and corrected to remove instrumental noise and unrealistic values. Data gaps were usually a few hours due to instrument failure, but could be several weeks in the case of maintenance or harvesting.

Gaps were filled while preserving the diurnal and seasonal cycles of SW radiation. The corrected and gap-filled time series were used to calculate hourly, daily, monthly or annual albedo. The methods were developed and described in Paper I, and used again in Paper II.

4.2.2 Mobile mast measurements

A mobile system was developed for measuring albedo on multiple plots in Uppsala on the same day throughout one year. A total of 22 plots were regularly sampled, covering common crops and management practices in Northern Europe, some less common crops (e.g. winter barley, maize, peas) and different cultivation intensities. The plots were located on four commercially farmed fields and at an experimental site consisting of two fields with multiple plots each. Field operations and inputs followed normal practice in conventional cropping, except for trials with low, normal and high nitrogen fertilisation of three cereal crops. Paper IV included 14 plots, whereof two were replicates (plots not included were maize, undersown ley with minimal management, and low and high fertilisation levels on winter rye, winter wheat and spring barley).

The mobile system consisted of a portable tripod with a vertically extendable mast and 2-m long horizontal cross-arm, a pair of thermophile pyranometers, a data logger, a Bluetooth serial adapter and a mobile app for direct data transfer and display in the field. The plots were sampled every 1-2 weeks under stable conditions. Measurements were taken for 3-5 minutes within three hours of solar noon. The sampling design was based on a previous evaluation of continuous pyranometer data in Paper I, which showed that selectively measured albedo on clear days around solar noon can be used to approximate seasonal albedo for energy balance calculations.

Albedo for each plot and sampling day was calculated as the ratio of reflected to incoming average irradiance in a stable period. After replacing individual missing values, the sampled time series was interpolated to daily frequency.

Annual albedo was calculated as the weighted mean of daily albedo, using daily incoming radiation for weighting. Weighting is necessary to handle discontinuous measurements.

4.2.3 MODIS satellite products

MODIS BRDF/Albedo products utilise reflectance observations from the MODIS instruments aboard Terra and Aqua satellites, which travel in near-polar orbits at 705 km altitude. MODIS scans a wide swath of 2300 km, so both instruments sample almost the entire surface of the Earth every day.

MODIS BRDF/Albedo products are provided daily at 500 m nominal grid resolution. Because of their high temporal resolution (at the expense of spatial resolution), the products are commonly used to monitor land surface

properties over time and to characterise the albedo of contrasting land cover classes. However, the products use a spatial and temporal sampling procedure, so that the BRDF and albedo values assigned per pixel may derive from observations over a larger area and temporal interval than the product’s nominal resolution suggests.

The coarser effective resolution and lack of information about the actual observational footprint of MODIS BRDF/Albedo pixels make it difficult to match the product with surface conditions. In spatially heterogeneous agricultural landscapes, many fields are smaller and differently shaped and placed than the area contributing to a pixel’s signal. Thus, most BRDF and albedo values do not represent a single crop or management regime, but are composed of several crops, grassland and other landscape elements (e.g.

hedges, trees etc.). Furthermore, different annual crops are grown every year and accurate maps of yearly crop cultivation are not generally available.

In Paper III, methods were developed to identify homogeneous pixels and to obtain representative albedo values for common crops and unimproved permanent grassland in production region PO1. Geospatial analysis was used to determine the overlap between the observational footprint of MODIS BRDF/Albedo pixels, modelled by a Gaussian function, and agricultural land use in harvest years 2011-2020, represented by annual polygon layers of individual fields. Pixels whose signal originated to at least 80% from a single land use were selected to achieve high pixel purity while maintaining a representative sample of pixels per land use and year. Blue-sky albedo, i.e. albedo under actual illumination conditions with a combination of diffuse and direct radiation, was calculated in two steps.

First, the kernel-driven BRDF model was used with a pixel’s daily BRDF parameters for the SW band (300-5000 nm) to calculate daily white-sky albedo (WSA) under completely diffuse illumination and hourly black-sky albedo (BSA) under direct illumination as a function of solar zenith angle.

Second, blue-sky albedo was obtained as the average of WSA and BSA weighted by the fractions of diffuse and direct surface irradiance. As with other discontinuous albedo data, daily and annual albedo were calculated as the weighted mean, using incoming radiation for weighting.

Similar methods were employed in Paper IV to derive MODIS-based albedo for crops in production region PO4 (encompassing Uppsala). MODIS data were used for comparison with field-measured values and to assess variability across years and sites within a region.