Barry Goodison

Interannual Variability in Reconstructed Canadian Snow Cover, 1915–1992

Abstract Seasonal snow cover information over southern Canada was reconstructed from daily snowfall and maximum temperature data back to 1915 using a simple mass balance approach with snowmelt estimated via a calibrated temperature index method. The reconstruction method was able to account for 70%–80% of the variance in annual snow cover duration (SCD) over most of southern Canada for the 1955–1992 calibration period. The data were used to construct regional SCD anomaly series in four regions spanning the continent. The regional SCD series were characterized by high interannual variability, with most of the variance concentrated at periods less than 5 years. Spring (MAM) snow cover variability was characterized by a prominent spectral peak with a period of approximately 4 years, which appeared to be linked to tropical Pacific sea surface temperature variability. There was no evidence of statistically significant long-term trends in snow cover in any of the regions, but the data suggested that winter (DJF...

Adjustment of daily precipitation data at 10 climate stations in Alaska: Application of World Meteorological Organization intercomparison results

A methodology for adjusting the daily precipitation measured by the U.S. National Weather Service (NWS) 8-inch standard gauge for wind-induced undercatch, wetting loss, and trace amount of precipitation is provided. The application of the proposed adjustment procedures was made at 10 NWS climate stations in Alaska for 1982 and 1983. The results show the following: (1) Daily adjustment for wind-induced undercatch, wetting loss, and trace amount of precipitation increases the gauge-measured annual precipitation by 65–800 mm for the 2 years (about 10-140% of the gauge-measured yearly total) at the 10 stations in Alaska; (2) compared to wetting loss and trace amount of precipitation, wind-induced undercatch is the source of greatest error, although wetting loss and trace amount of precipitation are also significant systematic errors in the northern Alaska regions of low precipitation; (3) in the similar climate condition, the NWS 8-inch standard gauges with an Alter wind shield have a much lower adjustment for wind-induced undercatch than the unshielded gauges at nearby stations, and the unshielded gauges placed on the roof of the weather office building have a higher adjustment for wind-induced errors than those gauges mounted on the ground; (4) monthly adjustment factors (adjusted/measured precipitation) differ by station, and at an individual station by type of precipitation; (5) considerable intra-annual variation of the magnitude of the adjustments has been found in Alaska owing to the fluctuation of wind speed, air temperature, and frequency of snowfall. Using the constant correction factors (derived at a single intercomparison site) to the archived monthly precipitation records produces significant undercorrection of the wind-induced errors at high wind sites and overcorrecting of the errors at low wind sites. To avoid the undercorrection or overcorrection of the wind-induced errors, a constant correction factor should not be applied to gauge-measured snow data. Daily adjustments for systematic errors need to be applied to the archived precipitation data. It is expected that the adjustments will have an impact on climate monitoring.

Bias correction of daily precipitation measurements for Greenland

A methodology for correcting the Hellmann gauge-measured daily precipitation for wind-induced undercatch, wetting loss, and trace amount of precipitation was presented and applied at 12 climate stations in Greenland for years 1994–1997. The results show the following: (1) daily corrections of the biases increase the gauge-measured annual precipitation by 35–280 mm for the 4 years (about 25–90% of the gauge-measured yearly total) in Greenland; (2) of the biases, wind-induced undercatch is the source of greatest error; wetting loss and trace amount of precipitation are also significant biases in the northern Greenland regions of low precipitation; (3) monthly correction factors (corrected/measured precipitation) differ by station and, at an individual station, by type of precipitation; (4) considerable intra-annual variation of the yearly corrections has been found in Greenland due to the fluctuation of wind speed, air temperature, and frequency of snowfall. These results suggest that annual precipitation in Greenland is much higher than previously reported, particularly in the southern regions of high precipitation; and the latitudinal precipitation gradient may also be greater over Greenland. These results will have a significant impact to water budget and glacier mass balance studies in Greenland.

Quantification of precipitation measurement discontinuity induced by wind shields on national gauges

Various combinations of wind shields and national precipitation gauges commonly used in countries of the northern hemisphere have been studied in this paper, using the combined intercomparison data collected at 14 sites during the World Meteorological Organizations (WMO) Solid Precipitation Measurement Intercomparison Project. The results show that wind shields improve gauge catch of precipitation, particularly for snow. Shielded gauges, on average, measure 20–70% more snow than unshielded gauges. Without a doubt, the use of wind shields on precipitation gauges has introduced a significant discontinuity into precipitation records, particularly in cold and windy regions. This discontinuity is not constant and it varies with wind speed, temperature, and precipitation type. Adjustment for this discontinuity is necessary to obtain homogenous precipitation data for climate change and hydrological studies. The relation of the relative catch ratio (RCR, ratio of measurements of shielded gauge to unshielded gauge) versus wind speed and temperature has been developed for Alter and Tretyakov wind shields. Strong linear relations between measurements of shielded gauge and unshielded gauge have also been found for different precipitation types. The linear relation does not fully take into account the varying effect of wind and temperature on gauge catch. Overadjustment by the linear relation may occur at those sites with lower wind speeds, and underadjustment may occur at those stations with higher wind speeds. The RCR technique is anticipated to be more applicable in a wide range of climate conditions. The RCR technique and the linear relation have been tested at selected WMO intercomparison stations, and reasonable agreement between the adjusted amounts and the shielded gauge measurements was obtained at most of the sites. Test application of the developed methodologies to a regional or national network is therefore recommended to further evaluate their applicability in different climate conditions. Significant increase of precipitation is expected due to the adjustment particularly in high latitudes and other cold regions. This will have a meaningful impact on climate variation and change analyses.

Compatibility evaluation of national precipitation gage measurements

Compatibility of precipitation measurements of various national gages commonly used in the Northern Hemisphere countries has been evaluated, based on the gage intercomparison data collected at 10 stations during the World Meteorological Organization (WMO) Solid Precipitation Measurement Intercomparison Project. Little difference (less than 5%) is found between national rainfall data, but a significant discrepancy (up to 110%) exists between national snowfall records. This difference is not constant and it varies with wind speed and temperature. It is certain that use of different precipitation gages in neighboring countries has introduced a significant discontinuity into precipitation records, particularly in cold and windy regions. Strong linear relations among daily national gage measurements have been defined for several national gages commonly used in the Northern Hemisphere. These linear relations provide a useful technique to adjust gage records when wind speed and temperature data are not available. The linear relations have been tested at selected WMO intercomparison stations, and good agreements of the adjusted amounts to other gage measurements are obtained at most of the test sites, indicating that the linear relations perform reasonably well at the selected WMO sites. Use of the proposed adjustment procedure will reduce inconsistency between precipitation measurements of national gages.

An evaluation of the Wyoming Gauge System for snowfall measurement

The Wyoming snow fence (shield) has been widely used with precipitation gauges for snowfall measurement at more than 25 locations in Alaska since the late 1970s. This gauges measurements have been taken as the reference for correcting wind-induced gauge undercatch of snowfall in Alaska. Recently, this fence (shield) was tested in the World Meteorological Organization Solid Precipitation Measurement Intercomparison Project at four locations in the United States of America and Canada for six winter seasons. At the Intercomparison sites an octagonal vertical Double Fence with a Russian Tretyakov gauge or a Universal Belfort recording gauge was installed and used as the Intercomparison Reference (DFIR) to provide true snowfall amounts for this Intercomparison experiment. The Intercomparison data collected were compiled at the four sites that represent a variety of climate, terrain, and exposure. On the basis of these data sets the performance of the Wyoming gauge system for snowfall observations was carefully evaluated against the DFIR and snow cover data. The results show that (1) the mean snow catch efficiency of the Wyoming gauge compared with the DFIR is about 80–90%, (2) there exists a close linear relation between the measurements of the two gauge systems and this relation may serve as a transfer function to adjust the Wyoming gauge records to obtain an estimate of the true snowfall amount, (3) catch efficiency of the Wyoming gauge does not change with wind speed and temperature, and (4) Wyoming gauge measurements are generally compatible to the snowpack water equivalent at selected locations in northern Alaska. These results are important to our effort of determining true snowfall amounts in the high latitudes, and they are also useful for regional hydrologic and climatic analyses.

Temporal and spatial variability of North American prairie snow cover (1988–1995) inferred from passive microwave- derived snow water equivalent imagery

Estimates of regional snow water equivalent (SWE) are essential for hydrological prediction, climatological analysis, and meteorological forecasting. Passive microwave-derived estimates of snow cover have unique benefits such as all-weather imaging, rapid scene revisit capabilities, and the ability to provide these quantitative SWE data. For this study the available time series of special sensor microwave/imager (SSM/)) brightness temperatures in the equal area SSM/I Earth grid projection were processed with the Canadian Atmospheric Environment Service dual-channel SWE algorithm for a ground-validated North American prairie region. Seven winter seasons (December, January, and February) of SWE imagery spanning 1988–1995 and averaged for 5 day intervals were subjected to a rotated principal components analysis (PCA) performed individually for each season. A final PCA considering all 7 winter seasons was performed in order to investigate the degree to which snow cover patterns reappear from one season to the next. Results indicate that modes of snow cover in the North American prairies are most persistent during the late winter (February) and exhibit a greater degree of variability during December than the other winter months. Two snow cover regimes are identified for the study region, with the winters of 1988/1989–1991/1992 characterized in a manner that is unique in both temporal and spatial aspects from the winters of 1992/1993–1994/1995.

Influence of Sensor Overpass Time on Passive Microwave-Derived Snow Cover Parameters

Abstract Passive microwave-derived retrieval of terrestrial snow water equivalent (SWE) is strongly influenced by snowpack wetness. The presence of water in the crystal matrix increases the microwave emissivity, destroying the between-channel brightness temperature gradient used to make quantitative estimates of SWE. To obtain the most accurate SWE imagery, overpass times from orbiting sensors such as the Special Sensor Microwave/Imager (SSM/I) can be chosen so the diurnally coldest and driest snowpack is being monitored. In this study we seek to evaluate the role of sensor overpass time when mapping SWE by comparing two data sets of 5-day averaged, winter season (December, January, and February) SWE imagery for a ground-validated prairie study area. The first data set is derived from SSM/I morning overpass times, the second from afternoon overpass times. Correlation analysis and modified mean bias error calculations are used to quantify the association between the two data sets. In addition, a series of principal components analysis (PCA) tests have been utilized to quantify the spatial and temporal association between the two time series. Results indicate a strong agreement in SWE retrievals between the two data sets, with little systematic bias that can be attributed to sensor overpass time. Differences in snow-covered areas are more marked, with the morning overpass data consistently estimating greater snow extent. The PCA indicates that the results of time series analysis can be influenced by the choice of sensor overpass time, although the dominant characteristics of the seasonal snow cover evolution are captured by both data sets. Surface temperature data are utilized to illustrate the dependence of algorithm performance on the physical state of the snowpack.

Time-series analysis of passive-microwave-derived central North American snow water equivalent imagery

Abstract The Meteorological Service of Canada has developed a series of operational snow water equivalent (SWE) retrieval algorithms for central Canada, based on the vertically polarized difference index for the 19 and 37 GHz channels of the Special Sensor Microwave/Imager (SSM/I). Separate algorithms derive SWE for open environments, deciduous, coniferous and sparse forest cover. A final SWE value represents the area-weighted average based on the proportional land cover within each pixel. In this study, 5 day averaged (pentad) passive-microwave-derived SWE imagery for the winter season (December–February) of 1994/95 is compared to in situ data from central Canada in order to assess algorithm performance. Investigation of regions with varying proportional land cover within the four algorithm classes shows that retrieved SWE remains within ±10–20mm of surface observations, independent of fractional within-pixel land cover. Following algorithm evaluation, ten winter seasons (1988/89 through 1997/98) of pentad central North American SWE imagery are subjected to a rotated principal-component analysis (PCA). Although there are no trends in total study-area SWE, the PCA results identify the interseasonal variability in the SWE accumulation and ablation centers of action through the SSM/I time series.

Associations between spatially autocorrelated patterns of SSM/I-derived prairie snow cover and atmospheric circulation

Passive-microwave derived observations of snow cover show potential to provide synoptically sensitive, and hydrologically and climatologically significant, information because of all-weather imaging capabilities, rapid scene revisit time and the ability to derive quantitative estimates of snow water equivalent (SWE). In this study, we seek to identify the dominant patterns of clustering in SWE imagery using the Getis statistic, a local indicator of spatial association. The SWE data were derived from five day-averaged Special Sensor Microwave/ Imager (SSM/I) brightness temperatures using the Canadian Atmospheric Environment Service dual channel algorithm. The analysed data span one winter season (December-February 1988-1989) and are limited to a ground-validated prairie scene. National Center for Environmental Prediction (NCEP) gridded atmospheric data (500 mb geopotential height; 700 mb temperature) were incorporated into the study to investigate whether the spatial orientation of the Getis statistic clusters provides information on interaction between snow cover and the atmosphere. Results show that the direction of atmospheric airflow as expressed by the 500 mb geopotential height field corresponds strongly to the orientation of surface snow cover clusters with no time lag. The 700 mb temperature field is also a controlling influence on the snow cover clusters both through modifying cluster orientation and reinforcing cluster magnitude.