The portion of total solar radiation that is reflected by Earth’s surface, or surface albedo, is a critical parameter for determining radiative energy fluxes on land. Increases in albedo can lead to decreases in precipitation, and small changes in albedo on the order of 10-1 can produce temperature changes of 10˚C. Since albedo calculations present a major radiative uncertainty in climate modeling, high resolution (1 km) remotely sensed aerosol optical depths and Ross-Thick Li-Sparse Reciprocal (RTLS) kernel parameters were obtained to estimate the albedo at the surface. These data were developed with the Multi-angle Implementation of Atmospheric Correction (MAIAC) algorithm, which uses the time series of measurements from the Moderate Resolution Imaging Spectroradiometer (MODIS) to collect aerosol optical thickness and surface bi-directional reflection factor (BRF) retrievals simultaneously. Along with MODIS land surface temperature data, Global 30 Arc-Second Elevation (GTOPO30) digital elevation data, and Famine Early Warning Systems Network (FEWS NET) Land Data Assimilation System (FLDAS) surface pressure products, the aerosol optical depths and RTLS kernel parameters were used to calculate hourly spectral narrowband and shortwave broadband albedo data sets over the span of 20 years (1998-2018).
To evaluate changes in plant photosynthetic rates, data such as the MODIS leaf area index (LAI), vegetation fraction (VF), and solar-induced chlorophyll fluorescence (SIF) are integrated in a multi-dimensional database with albedo. These data are jointly analyzed to understand the long-term subgrid scale variability and diurnal land-atmosphere dynamics of the miombo ecosystem. Further, we characterize the relationship between albedo, LULC types, and precipitation fluctuations at multiple spatial resolutions. With fractal methods and machine learning, we assess this joint behavior with current climate model simulations to predict agricultural productivity and uncertainty in regional climate change projections.