Spatial predictions of near-surface air temperature ( $T_{air}$ ) in Antarctica are required as baseline information for a variety of research disciplines. Since the network of weather stations in Antarctica is sparse, remote sensing methods have large potential due to their capabilities and accessibility. Based on the MODIS land surface temperature (LST) data, $T_{air}$ at the exact time of satellite overpass was modelled at a spatial resolution of 1 km using data from 32 weather stations. The performance of a simple linear regression model to predict $T_{air}$ from LST was compared to the performance of three machine learning algorithms: Random Forest (RF), generalized boosted regression models (GBM) and Cubist. In addition to LST, auxiliary predictor variables were tested in these models. Their relevance was evaluated by a Cubist-based forward feature selection in conjunction with leave-one-station-out cross-validation to reduce the impact of spatial overfitting. GBM performed best to predict $T_{air}$ using LST and the month of the year as predictor variables. Using the trained model, $T_{air}$ could be estimated with a leave-one-station-out cross-validated $R^2$ of 0.71 and a RMSE of 10.51 ${}^{\circ }$ C. However, the machine learning approaches only slightly outperformed the simple linear estimation of $T_{air}$ from LST ( $R^2$ of 0.64, RMSE of 11.02 ${}^{\circ }$ C). Using the trained model allowed creating time series of $T_{air}$ over Antarctica for 2013. Extending the training data by including more years will allow developing time series of $T_{air}$ from 2000 on. Full article

While satellite-based monitoring of vegetation activity at the earth’s surface is of vital importance for many eco-climatological applications, the degree of agreement among certain sensors and products providing estimates of the Normalized Difference Vegetation Index (NDVI) has been found to vary considerably. In order to assess the extent of such differences in highly heterogeneous terrain, we analyze and compare intra-annual seasonal fluctuations and long-term monotonic trends (2003–2012) in the Kilimanjaro region, Tanzania. The considered NDVI datasets include the Moderate Resolution Imaging Spectroradiometer (MODIS) products from Terra and Aqua, Collections 5 and 6, and the 3rd Generation Global Inventory Modeling and Mapping Studies (GIMMS) product. The degree of agreement in seasonal fluctuations is assessed by calculating a pairwise Index of Association (IOA_s), whereas long-term trends are derived from the trend-free pre-whitened Mann–Kendall test. On the seasonal scale, the two Terra-MODIS products (and, accordingly, the two Aqua-MODIS products) are best associated with each other, indicating that the seasonal signal remained largely unaffected by the new Collection 6 calibration approach. On the long-term scale, we find that the negative impacts of band ageing on Terra-MODIS NDVI have been accounted for in Collection 6, which now distinctly outweighs Aqua-MODIS in terms of greening trends. GIMMS NDVI, by contrast, fails to capture small-scale seasonal and trend patterns that are characteristic for the highly fragmented landscape which is likely owing to the coarse spatial resolution. As a short digression, we also demonstrate that the amount of false discoveries in the determined trend fraction is distinctly higher for p < 0.05 ( 52.6 % ) than for p < 0.001 ( 2.2 % ) which should point the way for any future studies focusing on the reliable deduction of long-term monotonic trends. Full article