Search

Atmospheric humidity plays an important role in climate analyses. Here we describe the production and key characteristics of a new quasi-global marine humidity product intended for climate monitoring, HadISDH.marine. It is an in situ multivariable marine humidity product, gridded monthly at a 5∘×5∘ spatial resolution from January 1973 to December 2018 with annual updates planned. Currently, only reanalyses provide up-to-date estimates of marine surface humidity, but there are concerns over their long-term stability. As a result, this new product makes a valuable addition to the climate record and will help address some of the uncertainties around recent changes (e.g. contrasting land and sea trends, relative-humidity drying). Efforts have been made to quality-control the data, ensure spatial and temporal homogeneity as far as possible, adjust for known biases in non-aspirated instruments and ship heights, and also estimate uncertainty in the data. Uncertainty estimates for whole-number reporting and for other measurement errors have not been quantified before for marine humidity. This is a companion product to HadISDH.land, which, when combined, will provide methodologically consistent land and marine estimates of surface humidity. The spatial coverage of HadISDH.marine is good over the Northern Hemisphere outside of the high latitudes but poor over the Southern Hemisphere, especially south of 20∘ S. The trends and variability shown are in line with overall signals of increasing moisture and warmth over oceans from theoretical expectations and other products. Uncertainty in the global average is larger over periods where digital ship metadata are fewer or unavailable but not large enough to cast doubt over trends in specific humidity or air temperature. Hence, we conclude that HadISDH.marine is a useful contribution to our understanding of climate change. However, we note that our ability to monitor surface humidity with any degree of confidence depends on the continued availability of ship data and provision of digitized metadata. HadISDH.marine data, derived diagnostics, and plots are available at http://www.metoffice.gov.uk/hadobs/hadisdh (last access: June 2019) and https://doi.org/10.5285/463b2fcd6a264a39b1e3249dab16c177 (Willett et al., 2020).

With the ongoing warming of the globe, it is important to quantify changes in the recent behaviour of extreme events given their impacts on human health, infrastructure and the natural environment. We use the sub-daily, multivariate, station-based HadISD dataset to study the changes in the statistical distributions of temperature, dew point temperature and wind speeds. Firstly, we use zonally averaged quantities to show that the lowest temperatures during both day and night are changing more rapidly than the highest, with the effect more pronounced in the northern high latitudes. Along with increases in the zonally averaged mean temperature, the standard deviation has decreased and the skew increased (increasing positive tail, decreasing negative tail) over the last 45 years, again with a stronger, more robust signal at higher latitudes. Changes in the distribution of dew point temperature are similar to those of temperature. However, changes in the distribution of wind speeds indicate a more rapid change at higher speeds than at lower. Secondly, to assess in more detail the spatial distribution of changes as well as changes across seasons and hours of the day we study each station individually. For stations which show clear indications of change in the statistical moments, the higher the statistical moment, generally the more spatially heterogenous the patterns of change. The standard deviations of temperatures are increasing in a band stretching from Europe through China but are decreasing across North America and in the high northern latitudes, indicating broadening and narrowing of the distributions, respectively. Large seasonal differences are found in the change of standard deviations of temperatures over North America and eastern China. Temperatures in eastern Asia also have increasing skew in the winter in contrast to the remainder of the year. The dew point temperatures show smaller variation in all of the moments but similar patterns to the temperatures. For wind speeds, apart from the USA, standard deviations are decreasing across the world, indicating a decrease in variability. Finally, we use quantile regression to show changes in the percentiles of distributions over time. These show an increase in high quantiles of temperature in eastern Europe during the summer and also in northern Europe for low quantiles in the winter, also indicating broadening and narrowing of the distributions, respectively. In North America, the largest changes are at the lower quantiles in northern latitudes for autumn and winter. Quantiles of dew point temperature are changing most in the autumn and winter, especially in the northern parts of Europe.

We compare the latest observational land surface humidity dataset, HadISDH, with the latest generation of climate models extracted from the CMIP5 archive and the ERA-Interim reanalysis over the period 1973 to present. The globally averaged behaviour of HadISDH and ERA-Interim are very similar in both humidity measures and air temperature, on decadal and interannual timescales. The global average relative humidity shows a gradual increase from 1973 to 2000, followed by a steep decline in recent years. The observed specific humidity shows a steady increase in the global average during the early period but in the later period it remains approximately constant. None of the CMIP5 models or experiments capture the observed behaviour of the relative or specific humidity over the entire study period. When using an atmosphere-only model, driven by observed sea surface temperatures and radiative forcing changes, the behaviour of regional average temperature and specific humidity are better captured, but there is little improvement in the relative humidity. Comparing the observed climatologies with those from historical model runs shows that the models are generally cooler everywhere, are drier and less saturated in the tropics and extra-tropics, and have comparable moisture levels but are more saturated in the high latitudes. The spatial pattern of linear trends is relatively similar between the models and HadISDH for temperature and specific humidity, but there are large differences for relative humidity, with less moistening shown in the models over the tropics and very little at high latitudes. The observed drying in mid-latitudes is present at a much lower magnitude in the CMIP5 models. Relationships between temperature and humidity anomalies (T–q and T–rh) show good agreement for specific humidity between models and observations, and between the models themselves, but much poorer for relative humidity. The T–q correlation from the models is more steeply positive than the observations in all regions, and this over-correlation may be due to missing processes in the models. The observed temporal behaviour appears to be a robust climate feature rather than observational error. It has been previously documented and is theoretically consistent with faster warming rates over land compared to oceans. Thus, the poor replication in the models, especially in the atmosphere-only model, leads to questions over future projections of impacts related to changes in surface relative humidity. It also precludes any formal detection and attribution assessment.

HadISD is a sub-daily, station-based, quality-controlled dataset designed to study past extremes of temperature, pressure and humidity and allow comparisons to future projections. Herein we describe the first major update to the HadISD dataset. The temporal coverage of the dataset has been extended to 1931 to present, doubling the time range over which data are provided. Improvements made to the station selection and merging procedures result in 7677 stations being provided in version 2.0.0.2015p of this dataset. The selection of stations to merge together making composites has also been improved and made more robust. The underlying structure of the quality control procedure is the same as for HadISD.1.0.x, but a number of improvements have been implemented in individual tests. Also, more detailed quality control tests for wind speed and direction have been added. The data will be made available as NetCDF files at http://www.metoffice.gov.uk/hadobs/hadisd and updated annually.

HadISDH.2.0.0 is the first gridded, multi-variable humidity and temperature in situ observations-only climate-data product that is homogenised and annually updated. It provides physically consistent estimates for specific humidity, vapour pressure, relative humidity, dew point temperature, wet bulb temperature, dew point depression and temperature. It is a monthly mean gridded (5° by 5°) product with uncertainty estimates that account for spatio-temporal sampling, climatology calculation, homogenisation and irreducible random measurement effects. It provides a tool for the long-term monitoring of a variety of humidity-related variables which have different impacts and implications for society. It is also useful for climate model evaluation and reanalyses validation. HadISDH.2.0.0 is shown to be in good agreement both with other estimates and with theoretical understanding. The data set is available from 1973 to the present. The theme common to all variables is of a warming world with more water vapour present in the atmosphere. The largest increases in water vapour are found over the tropics and the Mediterranean. Over the tropics and high northern latitudes the surface air over land is becoming more saturated. However, despite increasing water vapour over the mid-latitudes and Mediterranean, the surface air over land is becoming less saturated. These observed features may be due to atmospheric circulation changes, land–sea warming disparities and reduced water availability or changed land surface properties.

We report on preliminary steps in the homogenisation of HadISD, a sub-daily, station-based data set covering 1973–2013. Using temperature, dew point temperature, sea-level pressure and wind speeds, change points are detected using the Pairwise Homogenisation Algorithm from Menne and Williams Jr. (2009). Monthly-mean values and monthly-mean diurnal ranges (temperature and dew point temperature) or monthly-maximum values (wind speeds) are processed using the full network of 6103 stations in HadISD. Where multiple change points are detected within 1 year, they are combined and the average date is used. Under the assumption that the underlying true population of inhomogeneity magnitudes is Gaussian, inhomogeneity magnitudes as small as around 0.5 °C, 0.5 hPa or 0.5 m s−1 have been successfully detected. The change point dates and inhomogeneity magnitudes for each of the calculation methods will be provided alongside the data set to allow users to select stations which have different levels of homogeneity. We give an example application of this change point information in calculating global temperature values from HadISD and comparing these to CRUTEM4. Removing the most inhomogeneous stations results in a better match between HadISD and CRUTEM4 when matched to the same coverage. However, further removals of stations with smaller and fewer inhomogeneities worsen the match.

HadISDH is a near-global land surface specific humidity monitoring product providing monthly means from 1973 onwards over large-scale grids. Presented herein to 2012, annual updates are anticipated. HadISDH is an update to the land component of HadCRUH, utilising the global high-resolution land surface station product HadISD as a basis. HadISD, in turn, uses an updated version of NOAA's Integrated Surface Database. Intensive automated quality control has been undertaken at the individual observation level, as part of HadISD processing. The data have been subsequently run through the pairwise homogenisation algorithm developed for NCDC's US Historical Climatology Network monthly temperature product. For the first time, uncertainty estimates are provided at the grid-box spatial scale and monthly timescale. HadISDH is in good agreement with existing land surface humidity products in periods of overlap, and with both land air and sea surface temperature estimates. Widespread moistening is shown over the 1973–2012 period. The largest moistening signals are over the tropics with drying over the subtropics, supporting other evidence of an intensified hydrological cycle over recent years. Moistening is detectable with high (95%) confidence over large-scale averages for the globe, Northern Hemisphere and tropics, with trends of 0.089 (0.080 to 0.098) g kg−1 per decade, 0.086 (0.075 to 0.097) g kg−1 per decade and 0.133 (0.119 to 0.148) g kg−1 per decade, respectively. These changes are outside the uncertainty range for the large-scale average which is dominated by the spatial coverage component; station and grid-box sampling uncertainty is essentially negligible on large scales. A very small moistening (0.013 (−0.005 to 0.031) g kg−1 per decade) is found in the Southern Hemisphere, but it is not significantly different from zero and uncertainty is large. When globally averaged, 1998 is the moistest year since monitoring began in 1973, closely followed by 2010, two strong El Niño years. The period in between is relatively flat, concurring with previous findings of decreasing relative humidity over land.

This paper describes the creation of HadISD: an automatically quality-controlled synoptic resolution dataset of temperature, dewpoint temperature, sea-level pressure, wind speed, wind direction and cloud cover from global weather stations for 1973–2011. The full dataset consists of over 6000 stations, with 3427 long-term stations deemed to have sufficient sampling and quality for climate applications requiring sub-daily resolution. As with other surface datasets, coverage is heavily skewed towards Northern Hemisphere mid-latitudes. The dataset is constructed from a large pre-existing ASCII flatfile data bank that represents over a decade of substantial effort at data retrieval, reformatting and provision. These raw data have had varying levels of quality control applied to them by individual data providers. The work proceeded in several steps: merging stations with multiple reporting identifiers; reformatting to netCDF; quality control; and then filtering to form a final dataset. Particular attention has been paid to maintaining true extreme values where possible within an automated, objective process. Detailed validation has been performed on a subset of global stations and also on UK data using known extreme events to help finalise the QC tests. Further validation was performed on a selection of extreme events world-wide (Hurricane Katrina in 2005, the cold snap in Alaska in 1989 and heat waves in SE Australia in 2009). Some very initial analyses are performed to illustrate some of the types of problems to which the final data could be applied. Although the filtering has removed the poorest station records, no attempt has been made to homogenise the data thus far, due to the complexity of retaining the true distribution of high-resolution data when applying adjustments. Hence non-climatic, time-varying errors may still exist in many of the individual station records and care is needed in inferring long-term trends from these data. This dataset will allow the study of high frequency variations of temperature, pressure and humidity on a global basis over the last four decades. Both individual extremes and the overall population of extreme events could be investigated in detail to allow for comparison with past and projected climate. A version-control system has been constructed for this dataset to allow for the clear documentation of any updates and corrections in the future.