Present and future Köppen-Geiger climate classification maps at 1-km resolution — TEI-tagged view

Hylke E. BeckpersName, Niklaus E. ZimmermannpersName, Tim R. McVicarpersName, Noemi VergopolanpersName, Alexis BergpersName, Eric F. WoodpersName (2018date). “Present and future Köppen-Geiger climate classification maps at 1-km resolution.” Scientific Datatitle·j.

Object of study — global present-dayterm·dom and projected future Köppen-Geiger climate classificationsterm·dom at 1-km resolutionterm·mth

Research question — Can multi-source, topographically corrected climateterm·dom ensembles produce more accurate, uncertainty-qualified present and future global climate-classification maps?

Contribution — new freely available 1-km present and RCP8.5 future Köppen-Geiger maps, confidence layers, classification code, and station-based validation

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

Present and future Köppen-Geiger climate classificationterm·dom maps at 1-km resolutionterm·mth Abstract We present new global mapsterm·dom of the Köppen-Geiger climate classificationterm·dom at an unprecedented 1-km resolutionterm·mth for the present-dayterm·dom (1980–2016) and for projected future conditionsterm·dom (2071–2100) under climateterm·dom change. The present-dayterm·dom map is derived from an ensemble of four high-resolution, topographically-corrected climatic mapsterm·mth. The future map is derived from an ensemble of 32 climate model projectionsterm·mth (scenario RCP8.5term·mth), by superimposing the projected climate change anomalyterm·mth on the baseline high-resolution climatic mapsterm·mth. For both time periods we calculate confidence levelsterm·mth from the ensemble spreadterm·mth, providing valuable indications of the reliability of the classifications. The new maps exhibit a higher classification accuracyterm·mth and substantially more detail than previous maps, particularly in regions with sharp spatial or elevation gradientsterm·dom. We anticipate the new maps will be useful for numerous applications, including species and vegetation distribution modelingterm·dom. The new maps including the associated confidence mapsterm·mth are freely available via www.gloh2o.org/koppen .

Background & Summary The Köppen-Geiger system classifies climateterm·dom into five main classesterm·dom and 30 sub-typesterm·dom. The classification is based on threshold valuesterm·mth and seasonality of monthly air temperature and precipitationterm·mth. Considering vegetation as “crystallized, visible climateterm·dom” 1 , this classification aims to empirically map biome distributionsterm·dom around the world: different regions in a similar class share common vegetation characteristics. The first version of this classification was developed in the late 19th century 2 ; it is still widely used today, for many applications and studies conditioned on differences in climatic regimes, such as ecological modeling or climateterm·dom change impact assessments 3–8 . This wide use reflects the fact that climateterm·dom has since long been recognized as the major driver of global vegetation distributionterm·dom 9–11 . In species distribution models 12 , climate variablesterm·dom are considered the primary driver to explain species rangesterm·dom at larger spatial extents, while habitat and topographyterm·dom are considered to only be modifiers of plant species distributions at smaller extents 13–15 . The Köppen-Geiger climate classificationterm·dom is a highly suitable means to aggregate complex climateterm·dom gradients into a simple but ecologically meaningful classification scheme. It is therefore often used as input when analyzing the distribution 4 , 16 , 17 or growth behavior 18 of species, or to set-up dynamic global vegetation models 19 .

Three recent versions of the world maps of the Köppen-Geiger climate classificationterm·dom exist 20–22 . Kottek et al. 20 produced a map (0.5° resolution) based on CRU TS 2.1 23 for temperature and VASClimO V1.1 24 for precipitation. CRU was based on approximately 7000–17,000 stations (depending on the year) and VASClimO on 9343 stations. The Peel et al. map 21 (0.1° resolution) was derived from 4844 air temperature stations and 12,396 precipitation stations. Kriticos et al. 22 produced a map (0.083° resolution) based on WorldClim V1 temperature and precipitation datasets 25 , which are based on 24,542 and 47,554 stations, respectively.

All maps have a relatively low resolution (≥0.1°) and the map of Peel et al. 21 has not been explicitly corrected for topographic effectsterm·dom, which influences air temperature 26 and precipitation 27 in mountainous regions. In addition, the maps of Kottek et al.term·dom 20 and Peel et al. 21 are based on a relatively small number of stations. This can lead to widespread misclassifications, particularly in regions with a low station densityterm·mth and/or strong climatic gradients such as mountain ranges 28 . Moreover, since these maps do not include corresponding uncertainty estimatesterm·mth, they may provide users a false sense of confidence.

Here, we present a new and improved Köppen-Geiger climate classificationterm·dom map for the present (1980–2016) with an unprecedented 0.0083° resolution (approximately 1 km at the equator), providing more accurate representation of highly heterogeneous regions ( Fig. 1a ). To maximize the accuracy and assess uncertainties in map classifications, we combine climatic air temperature and precipitation data from multiple independent sources, including WorldClim V1 and V2term·mth, CHELSA V1.2term·mth, and CHPclim V1term·mth ( Table 1 ). These datasets have all been explicitly corrected for topographic effectsterm·dom and, with the exception of the CHELSA V1.2term·mth temperature dataset, been based on a large number of stations (≥34,542 for precipitation and ≥20,268 for temperature). The use of multiple data sources allows us to provide an estimate of uncertainty in the derived classes. Further, we combine climateterm·dom change projections from 32 Coupled Model Intercomparison Project phase 5 (CMIP5 29 ) models to map future (2071–2100) climateterm·dom classes at the same spatial resolution ( Fig. 1b ).

Methods Köppen-Geiger climate classificationterm·dom We follow the Köppen-Geiger climate classificationterm·dom as described in Peel et al. 21 , which was also used by Kriticos et al. 22 ( Table 2 ). This classification is identical to that presented by Köppen in 1936 1 with three differences. First, temperate (C) and cold (D) climatesterm·dom are distinguished using a 0 °C thresholdterm·mth instead of a 3 °C thresholdterm·mth, following the suggestion of Russell 30 . Second, the arid (B) sub-climates W (desert) and S (steppe) were identified depending on whether 70% of precipitation occurred in summer or winter. Third, the sub-climates s (dry summer)term·dom and w (dry winter)term·dom within the C and D climatesterm·dom were made mutually exclusive by assigning s when more precipitation falls in winter than in summer and assigning w otherwise. Note that the tropical (A), temperate (C), cold (D), and polar (E) climatesterm·dom are mutually exclusive but may intersect with the arid (B) class. To account for this, climateterm·dom type B was given precedence over the other classes.

Climateterm·dom data The present Köppen-Geiger classification map was derived from three climatic datasets for air temperature (WorldClim V1 and V2term·mth, and CHELSA V1.2term·mth) and four climatic datasets for precipitation (WorldClim V1 and V2term·mth, CHELSA V1.2term·mth, and CHPclim V1term·mth; Table 1 ). All datasets have a 0.0083° resolution with the exception of CHPclim V1term·mth.2, which has a 0.05° resolution. For consistency CHPclim V1term·mth.2 was downscaled to 0.0083° using bilinear interpolationterm·mth.

The future Köppen-Geiger classificationterm·dom was produced using monthly historical and future air temperature and precipitation data from the CMIP5 archiveterm·mth 29 . For the future scenario, we used Representative Concentration Pathway 8.5 (RCP8.5 31 ). All climateterm·dom models with data during the 1980–2016 and 2071–2100 periods were used. Data for 1980–2016 was derived by concatenating historical runs (which end in 2005) and future runs (which begin in 2006). For each model, we only considered a single initialization ensemble. In total 32 models had sufficient data and hence were used for deriving the future map ( Table 1 ).

Present-dayterm·dom Köppen-Geiger map The present-dayterm·dom Köppen-Geiger map ( Fig. 1a ) was derived from an ensemble of high-resolution climatic datasets ( Table 1 ) using the criteria listed in Table 2 . Since the climatic datasets have inconsistent temporal coverages, we first adjusted them to reflect the period 1980–2016. To this end, we calculated, for each climatic dataset, monthly 0.5° climatologies for temperature using CRU TS V4.01 and for precipitation using GPCC FDR V7, for both the 1980–2016 period and the temporal span of the climatic dataset. Next, for each month we calculated climateterm·dom change offsets (for temperature) or factors (for precipitation) between the two periods, and resampled these offsets or factors from 0.5° to 0.0083° resolution using bilinear interpolationterm·mth, and adjusted the climatic maps by addition (for temperature) or multiplication (for precipitation).

For each adjusted temperature and precipitation climatic dataset combination, we derived a Köppen-Geiger map at 0.0083° resolution. From this ensemble of 4×3=12 maps we derived a final map by selecting, for each grid-cell, the most common classterm·mth ( Fig. 1a ). A corresponding confidence mapterm·mth was derived by dividing the frequency of occurrence of the most common classterm·mth by the ensemble size and converting these fractions to percentages ( Fig. 2a ). For example, if Csa is the most common classterm·mth for a particular grid-cell, and it has been assigned eight times out of 12, the resulting confidence levelterm·mth is 100 × 8 12 = 66.6 %. This confidence levelterm·mth should be interpreted as the degree of trustterm·dom we place in our final present-dayterm·dom classification. Confidence levelsterm·mth are generally lower in the vicinity of borders between climateterm·dom zones, in particular at high latitudes where the climatic data show more uncertainty.

Future Köppen-Geiger map The future Köppen-Geiger map ( Fig. 1b ) was derived by the so-called “anomaly methodterm·mth” 32 based on an ensemble of climateterm·dom projections from the 32 CMIP5 models ( Table 1 ). First, observed monthly present-dayterm·dom reference temperature and precipitation climatologies (0.0083° resolution) were derived, by simple averaging of the ensemble of temporally-homogenized, high-resolution climatic maps. Then, for each climateterm·dom model and each month, we subsequently calculated climateterm·dom change offsets (for temperature) or factors (for precipitation) between 1980–2016 and 2071–2100 and resampled these offsets or factors from the native model resolution to 0.0083° using bilinear interpolationterm·mth ( Fig. 3 ). Finally, future high-resolution climatic temperature and precipitation maps were derived from the present-dayterm·dom, observed reference climatologies by addition of the offsets (for temperature) or multiplication by the factors (for precipitation). We want to emphasize that the change factors are never excessively high (i.e., >5; Fig. 3 ), because (i) model simulations tend to overestimate the precipitation frequency 33 (resulting in the near-absence of areas with close to zero precipitation), and (ii) over the majority of arid regions the future projections tend toward drying rather than wetting 34 (resulting in factors <1).

For each climateterm·dom model, we derived a future Köppen-Geiger map at 0.0083° resolution from the downscaled future temperature and precipitation data. From this ensemble of 32 maps we derived a final map by selecting, for each grid-cell, the most common classterm·mth ( Fig. 1b ). A corresponding confidence mapterm·mth was derived by dividing the frequency of occurrence of the most common classterm·mth by the ensemble size and converting these fractions to percentages ( Fig. 2b ). For example, if Cfa is the most common classterm·mth for particular grid-cell, and it has been assigned 24 times out of 32, the corresponding confidence levelterm·mth is 100 × 24 32 = 75.0 %. This confidence levelterm·mth should be interpreted as the degree of trustterm·dom we have in our final future classification based on the uncertainties in climateterm·dom change projections. Thus, uncertainties are larger than for the present-dayterm·dom map. In particular, they are larger at high latitudes because of the greater model spread in projected warming in those regions.

Code availability The new Köppen-Geiger classifications have been produced using MathWorks MATLAB version R2017a. The function used to classify the temperature and precipitation data according to the criteria listed in Table 2 (KoppenGeiger.m) is freely available via (Data Citation 1) and www.gloh2o.org/koppen . The other codes are available upon request from the first author.

Data Records The present and future Köppen-Geiger classificationterm·dom maps and the corresponding confidence mapsterm·mth are freely available for download at (Data Citation 1) and www.gloh2o.org/koppen . The maps are stored in GeoTIFF formatterm·mth as unsigned 8-bit integers. We also provide a legend file (legend.txt) linking the numeric values in the maps to the Köppen-Geiger climateterm·dom symbols and providing the color scheme used for displaying the maps in the current study (adapted from Peel et al. 21 ). The maps are referenced to the World Geodetic Reference System 1984term·mth (WGS 84) ellipsoid and made available at three resolutions (0.0083°, 0.083°, and 0.5°; approximately 1 km, 10 km, and 50 km at the equator, respectively). The classifications are upscaled from 0.0083° to 0.083° and 0.5° using majority resamplingterm·mth and the confidence levelsterm·mth using bilinear averaging. Table 3 presents the file naming convention. The maps can be visualized and analyzed using most Geographic Information Systems (GIS) software (e.g., QGIS, ArcGIS, and GRASS).

Technical Validation We validated the new high-resolution present-dayterm·dom Köppen-Geiger classification ( Fig. 1a ), and previous maps from Kottek et al. 20 , Peel et al. 21 , and Kriticos et al. 22 , by calculating the classification accuracyterm·mth (defined as the percentage of correct classes) using station observationsterm·mth as reference. An initial database was compiled from the Global Historical Climatology Network-Daily (GHCN-D) database 35 and the Global Summary Of the Day (GSOD) database ( https://data.noaa.gov ). For each station, we calculated monthly mean temperature and precipitation time series (discarding months with <25 daily values), and subsequently monthly climatologies by averaging the monthly means (if ≥10 values were present). Stations with gaps in the climatologies or missing data for one of the four maps were discarded, resulting in a final dataset comprising 22,078 stationsterm·mth which we used to calculate the classification accuracyterm·mth of each map.

The newly derived high-resolution present-day Köppen-Geiger classificationterm·dom ( Fig. 1a ) exhibited a classification accuracyterm·mth of 80.0%, while the maps of Kottek et al.term·dom 20 , Peel et al. 21 , and Kriticos et al. 22 exhibited classification accuracies of 66.1, 70.9, and 73.4%, respectively. These results confirm that the new map is more accurate, which is primarily due to its high (1 km) resolution and use of an ensemble of topographically-corrected climatic datasets. The map of Kottek et al. 20 showed the lowest classification accuracyterm·mth, due to its low (0.5°) resolution. The map of Peel et al. 21 also performed less well, due to a lack of topographic corrections and the use of a relatively small number of stations.

We also tested the usefulness of the confidence mapterm·mth associated with the new present-dayterm·dom classification ( Fig. 2a ) using station observationsterm·mth. We obtained a mean confidence levelterm·mth of 92.6% for the correctly classified stations ( n =17,667) and 77.4% for the misclassified stations ( n =4411). The mean confidence levelterm·mth was thus substantially lower for the misclassified stations, confirming that the confidence mapterm·mth provides a useful indication of the classification accuracyterm·mth.

Figures 4 and 5 show historic Köppen-Geiger classification maps from all three previous studies and our present-dayterm·dom map for the Alps (Europe) and the central Rocky Mountains (North America), respectively, illustrating the enhanced detail in our map. The other maps sometimes fail to depict important topographic features; the map of Peel et al. 21 , for example, does not represent the Apennine mountains (Italy), due to a lack of topographic corrections ( Fig. 4e ). The new map ( Figs. 4a and 5a ) also exhibits better agreement with a Landsat-based forest cover map 36 (30-m resolution; Figs. 4c and 5c ). The spatial extent of the polar (E) climateterm·dom, for example, corresponds closely with treelines in the forest cover maps. Additionally, the new present-dayterm·dom and future Köppen-Geiger maps ( Fig. 4a and g , respectively) agree well with equivalent high-resolution maps derived for the Alps 8 (their Figs. 1 and 2 , respectively).

Usage Notes The future Köppen-Geiger classificationterm·dom ( Fig. 1b ) should be viewed as providing insights into potential spatial changes in regional climatic zones under climateterm·dom change. However, caution should be exerted not to equate those changes directly with changes in actual biomesterm·dom. First, vegetation changesterm·dom by 2100 may lag the change in climate zonesterm·dom. Secondly, factors not accounted for in the Köppen-Geiger classification, such as higher atmospheric CO 2 levels, may alter the relationship between climateterm·dom classes and vegetation. It is thus advised to interpret the future Köppen-Geiger classificationterm·dom first and foremost from a ‘climatic conditions’ perspective.

The rationale for using the anomaly methodterm·mth to build future maps using climateterm·dom model projections, instead of directly computing present and future maps from model outputs, is that superimposing future modeled anomalies onto the observed climateterm·dom removes mean biasesterm·dom from climateterm·dom model outputs. This is a widely used method in climateterm·dom change impact assessments 32 . However, an unavoidable limitation of this approach is that because of model spatial biasesterm·dom, modeled climateterm·dom change anomalies may not be fully geographically consistent with the baseline observed climatology to which they are added 37 (e.g., if the climateterm·dom of one region in a given model is spatially shifted relative to reality).

Another irredeemable limitation is that because of their coarser resolution (typically 1–2°), climateterm·dom model outputs do not resolve future climateterm·dom change at the same scale as our baseline climatology. Thus, in cases where there could be significant heterogeneities in precipitation change and/or warming below the model resolution (e.g., along coastlines and/or in regions with strong land-cover differences and/or elevation gradients), future changes at the 1-km scale might be under- or over-estimated, because only the model-scale mean anomaliesterm·mth are used to compute future changes. High-elevation mountainous regions are a prime example of this because they are expected to experience considerably more warming than adjacent valleys 38 .

Additional information How to cite this article : BeckpersName, H. E. et al., Present and future Köppen-Geiger climate classificationterm·dom maps at 1-km resolutionterm·mth. Sci. Data . 5:180214 doi: 10.1038/sdata.2018.214 (2018).

Publisher’s note : Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Material