R. Manzanas

Reassessing Statistical Downscaling Techniques for Their Robust Application under Climate Change Conditions

AbstractThe performance of statistical downscaling (SD) techniques is critically reassessed with respect to their robust applicability in climate change studies. To this end, in addition to standard accuracy measures and distributional similarity scores, the authors estimate the robustness of the methods under warming climate conditions working with anomalous warm historical periods. This validation framework is applied to intercompare the performances of 12 different SD methods (from the analog, weather typing, and regression families) for downscaling minimum and maximum temperatures in Spain. First, a calibration of these methods is performed in terms of both geographical domains and predictor sets; the results are highly dependent on the latter, with optimum predictor sets including near-surface temperature data (in particular 2-m temperature), which appropriately discriminate cold episodes related to temperature inversion in the lower troposphere.Although regression methods perform best in terms of co...

Precipitation variability and trends in Ghana: An intercomparison of observational and reanalysis products

Inter-annual variability and trends of annual/seasonal precipitation totals in Ghana are analyzed considering different gridded observational (gauge- and/or satellite-based) and reanalysis products. A quality-controlled dataset formed by fourteen gauges from the Ghana Meteorological Agency (GMet) is used as reference for the period 1961–2010. Firstly, a good agreement is found between GMet and all the observational products in terms of variability, with better results for the gauge-based products—correlations in the range of 0.7–1.0 and nearly null biases—than for the satellite-gauge merged and satellite-derived products. In contrast, reanalyses exhibit a very poor performance, with correlations below 0.4 and large biases in most of the cases. Secondly, a Mann-Kendall trend analysis is carried out. In most cases, GMet data reveal the existence of predominant decreasing (increasing) trends for the first (second) half of the period of study, 1961–1985 (1986–2010). Again, observational products are shown to reproduce well the observed trends—with worst results for purely satellite-derived data—whereas reanalyses lead in general to unrealistic stronger than observed trends, with contradictory results (opposite signs for different reanalyses) in some cases. Similar inconsistencies are also found when analyzing trends of extreme precipitation indicators. Therefore, this study provides a warning concerning the use of reanalysis data as pseudo-observations in Ghana.

Seasonal Predictability of Wintertime Precipitation in Europe Using the Snow Advance Index

AbstractThis study tests the applicability of Eurasian snow cover increase in October, as described by the recently published snow advance index (SAI), for forecasting December–February precipitation totals in Europe. On the basis of a classical correlation analysis, global significance was obtained and locally significant correlation coefficients of up to 0.89 and −0.78 were found for the Iberian Peninsula and southern Norway, respectively. For a more robust assessment of these results, a linear regression approach is followed to hindcast the precipitation sums in a 1-yr-out cross-validation framework, using the SAI as the only predictor variable. With this simple empirical approach, local-scale precipitation could be reproduced with a correlation of up to 0.84 and 0.71 for the Iberian Peninsula and southern Norway, respectively, while catchment aggregations on the Iberian Peninsula could be hindcast with a correlation of up to 0.73. These findings are confirmed when repeating the hindcast approach to a ...

Validation of 40 year multimodel seasonal precipitation forecasts: The role of ENSO on the global skill

The skill of seasonal precipitation forecasts is assessed worldwide—grid point by grid point—for the 40 year period 1961–2000, considering the ENSEMBLES multimodel hindcast and applying a tercile-based probabilistic approach in terms of the relative operating characteristic skill score (ROCSS). Although predictability varies with region, season, and lead time, results indicate that (1) significant skill is mainly located in the tropics—20 to 40% of the total land areas; (2) overall, September–October (March–May) is the most (least) skillful season; and (3) the skill weakens (with respect to the 1 month lead case) at 4 months lead—especially in June–August—although the ROCSS spatial patterns are broadly preserved—particularly in Northern South America and the Malay Archipelago. The contribution of El Nino–Southern Oscillation (ENSO) events to this 40 year skill is also analyzed, based on the idea that the seasonal predictability may be mainly driven by El Nino and La Nina precipitation teleconnections and, consequently, limited by the ability of the different seasonal forecasting models to accurately reproduce them. Results show that the ROCSS spatial patterns for (1) the full period 1961–2000 and (2) El Nino and La Nina events are highly correlated—over 0.85. Moreover, the observed teleconnection patterns are properly simulated (predicted)—with spatial correlations around 0.8—by most of the models at both 1 and 4 months lead time.The skill of seasonal precipitation forecasts is assessed worldwide —grid point by grid point— for the forty-years period 1961-2000. To this aim, the ENSEMBLES multimodel hindcast is considered. Although predictability varies with region, season and leadtime, results indicate that 1) significant skill is mainly located in the tropics —20 to 40% of the total land areas—, 2) overall, SON (MAM) is the most (less) skillful season and 3) predictability does not decrease noticeably from one to four months lead-time —this is so especially in northern south America and the Malay archipelago, which seem to be the most skillful regions of the world—. An analysis of teleconnections revealed that most of the skillful zones exhibit significant teleconnections with El Niño. Furthermore, models are shown to reproduce similar teleconnection patterns to those observed, especially in SON —with spatial correlations of around 0.6 in the tropics—. Moreover, these correlations are systematically higher for the skillful areas. Our results indicate that the skill found might be determined to a great extent by the models’ ability to properly reproduce the observed El Niño teleconnections, i.e., the better a model simulates the El Niño teleconnections, the higher its performance is.

Statistical Downscaling in the Tropics Can Be Sensitive to Reanalysis Choice: A Case Study for Precipitation in the Philippines

AbstractThis work shows that local-scale climate projections obtained by means of statistical downscaling are sensitive to the choice of reanalysis used for calibration. To this aim, a generalized linear model (GLM) approach is applied to downscale daily precipitation in the Philippines. First, the GLMs are trained and tested separately with two distinct reanalyses (ERA-Interim and JRA-25) using a cross-validation scheme over the period 1981–2000. When the observed and downscaled time series are compared, the attained performance is found to be sensitive to the reanalysis considered if climate change signal–bearing variables (temperature and/or specific humidity) are included in the predictor field. Moreover, performance differences are shown to be in correspondence with the disagreement found between the raw predictors from the two reanalyses. Second, the regression coefficients calibrated either with ERA-Interim or JRA-25 are subsequently applied to the output of a global climate model (MPI-ECHAM5) in o...

Reassessing Model Uncertainty for Regional Projections of Precipitation with an Ensemble of Statistical Downscaling Methods

AbstractThis is the second in a pair of papers in which the performance of statistical downscaling methods (SDMs) is critically reassessed with respect to their robust applicability in climate change studies. Whereas the companion paper focused on temperatures, the present manuscript deals with precipitation and considers an ensemble of 12 SDMs from the analog, weather typing, and regression families. First, the performance of the methods is cross-validated considering reanalysis predictors, screening different geographical domains and predictor sets. Standard accuracy and distributional similarity scores and a test for extrapolation capability are considered. The results are highly dependent on the predictor sets, with optimum configurations including information from midtropospheric humidity. Second, a reduced ensemble of well-performing SDMs is applied to four GCMs to properly assess the uncertainty of downscaled future climate projections. The results are compared with an ensemble of regional climate ...

An R package to visualize and communicate uncertainty in seasonal climate prediction

Abstract Interest in seasonal forecasting is growing fast in many environmental and socio-economic sectors due to the huge potential of these predictions to assist in decision making processes. The practical application of seasonal forecasts, however, is still hampered to some extent by the lack of tools for an effective communication of uncertainty to non-expert end users. visualizeR is aimed to fill this gap, implementing a set of advanced visualization tools for the communication of probabilistic forecasts together with different aspects of forecast quality, by means of perceptual multivariate graphical displays (geographical maps, time series and other graphs). These are illustrated in this work using the example of the strong El Nino 2015/16 event forecast. The package is part of the climate4R bundle providing transparent access to the ECOMS-UDG climate data service. This allows a flexible application of visualizeR to a wide variety of specific seasonal forecasting problems and datasets.

The R-based climate4R open framework for reproducible climate data access and post-processing

This work has been funded by the Spanish R+D Program of theMinistry of Economy and Competitiveness, through grants MULTI-SDM(CGL2015-66583-R) and INSIGNIA (CGL2016-79210-R), co-funded byERDF/FEDER. We would like to thank the two anonymous reviewersfor their valuable suggestions and comments.