Does the gap-filling method influence long-term (1950–2019) temperature and precipitation trend analyses?

Incomplete climatic series require gap-filling approaches so they can be used in homogeneous long-term spatiotemporal trend analyses. Monthly mean Temperature (MT) and Precipitation (PR) databases from the meteorological stations of the Iberian Peninsula have a high percentage of data gaps: 80.21% and 73.25% for the period 1950–1979 (P1), and 61.82% and 58.03% for the period 1980–2019 (P2). The different gap-filling methods of the Emmentalsoftware were tested to determine their performance and whether the gap-filling method influences these trend analyses. The nonparametric Theil-Sen approach and the Mann-Kendall test were used to assess the trend magnitude and its significance. The results showed (i) similar patterns between the evaluated methods, but with (ii) spatial differences, especially during P1. (iii) The comparison between standardized gap-filled and unfilled series did not show significant differences for MT and PR, although a reduction in the trend variability occurred in the first case (filled). (iv) Summer mean temperatures showed the largest warming trend (0.27 °C/decade), while autumn showed the smallest (0.21°C/decade) (median data for P1 and P2). Overall, an increase of 1.45 °C occurred in the entire period (annual median). (v) PR did not show any clear trend in any month in the entire period. This research has shown how climate trends can be affected by a reduction in data variability due to the application of gap filling methods. Although accounting for variability is of crucial importance for climate analysis, ignoring discontinuities in derived climatic surfaces causes greater spatiotemporal inconsistencies in derived climate products.