Significant anthropogenic-induced changes of climate classes since 1950

Abstract

Anthropogenic forcings have contributed to global and regional warming in the last few decades and likely affected terrestrial precipitation. Here we examine changes in major Köppen climate classes from gridded observed data and their uncertainties due to internal climate variability using control simulations from Coupled Model Intercomparison Project 5 (CMIP5). About 5.7% of the global total land area has shifted toward warmer and drier climate types from 1950-2010, and significant changes include expansion of arid and high-latitude continental climate zones, shrinkage in polar and midlatitude continental climates, poleward shifts in temperate, continental and polar climates, and increasing average elevation of tropical and polar climates. Using CMIP5 multi-model averaged historical simulations forced by observed anthropogenic and natural, or natural only, forcing components, we find that these changes of climate types since 1950 cannot be explained as natural variations but are driven by anthropogenic factors.

Figure 1

(a) Percentage of world land area with a climate type change in each year compared to 1950 from UD (black solid line) and CRU (black dashed line) observations, and HIST-ALL (yellow), HIST-GHG (red), and HIST-NAT (blue) runs. Dark shading shows the 95^th percentile of changed climate types relative to the starting year based on 54-year samples of PI-CTL simulations as illustrated in (b), and light shading shows the estimated 95^th percentile if the variance is doubled. (b) Distribution of 54-year changed area percentage due to internal climate variability based on 225 PI-CTL samples (gray bars) with their mean value of 3.1% indicated by the black dashed line. The vertical solid lines are the observed percentages of global land area with a changed climate class in the UD dataset (about 5.7%, black line) and the CRU dataset (about 5.6%, gray line). (c) Same as (a) but for the GISS dataset. (d) Same as (b) but for the GISS dataset. Panel (d) uses the same PI-CTL runs as in (c) but grids are averaged at the GISS resolution and calculations exclude GISS “missing” grid boxes.

Figure 2

(a) Linear trends in areas of 5 major climate types for 1950–2003 using the UD dataset; asterisks denote significant trends at the 5% level. A positive trend of high-latitude (north of 55°N) D climate and a negative midlatitude (south of 55°N) D climate are over-plotted in blue with the net negative trend of D climate in dark blue. (b) and (c) are the same as (a), but for trends in average absolute latitude (positive indicates poleward) and elevation, respectively. (d) Map showing grid boxes with a major climate type in 1950 that “disappeared” (changed to another type) by 2003. Grid boxes are 1° × 1°. Colors are the same as in upper panels. (e) is the same as (d), but for “emerging” climate types (by 2003) in the same grid boxes. We generate five sub-panels (a–e) using Matlab software, and integrate them into this figure using Adobe Illustrator.

Figure 3

Significant observed trends (black bars) marked with * in Fig. 2 and the corresponding simulated trends of indices for 1950–1998; yellow, red and blue bars denote HIST-ALL, HIST-GHG and HIST-NAT runs, respectively. Each error bar at the left of an observed trend is the standard deviation (σ) of such trend estimated from 225 samples of 54-yr CMIP5 control runs and represents the natural variability of the observed or modeled trend. Simulated trends significantly different from the observation at the 5% level are marked with diamonds. The units are 2 × 10⁵m² decade⁻¹ for area, 10 km decade⁻¹ for latitude and 5 m decade⁻¹ for elevation indices.