Surface meteorology and tropospheric cloud near Ross Island in Antarctica.

This thesis presents a series of studies into cloud and surface weather conditions present near Ross Island in Antarctica to investigate local-scale meteorology in the area and explore connections with larger scale atmospheric processes. Technical work on the development of specialist, low-cost, portable weather stations (SNOWWEB) is described with results and corresponding analyses from two successful field seasons presented. Covering the Austral summers of 2013/14 and 2014/15, both deployments utilized 15 to 20 weather stations over areas in the order of hundreds of square kilometers. A third-party classification product derived from surface-level winds in ERAInterim is used to provide synoptic context for these deployments and link results to an analysis of the combined radar (CloudSat) and lidar (CALIPSO) cloud product over the Ross Ice Shelf and southern Ross Sea. Located at the north-western corner of the Ross Ice Shelf - due south of New Zealand - the topography around Ross Island is complex and substantial. This creates associated complex interactions with air flow in the region, particularly near the surface, as winds flowing north over the large and featureless ice shelf encounter the terrain. A large-scale network of automated weather stations (AWS) exists over the greater Ross Ice Shelf area with good coverage for mesoscale studies, however logistical constraints limit the number that can be deployed and maintained with a paucity of observations at the local scale. SNOWWEB is a system of low-cost weather stations easy to transport and very quick to deploy designed to augment existing AWS observations. Substantial technical development of SNOWWEB occurred during the course of this thesis, with improvements to physical design and wireless networking capabilities presented. SNOWWEB observations were found to match well with those from nearby existing AWS during two summer season deployments near Ross Island, with results from the network as a whole showing coherent spatial structure in wind, temperature, and pressure fields. One SNOWWEB deployment covered the northern and western edges of White Island immediately south-east of Ross Island. Observations showed the interaction of a Ross Ice Shelf airstream (RAS) southerly storm event with the complex terrain of the deployment area, including a resulting small but intense gap wind. There was also a substantial dampening effect on the diurnal temperature cycle over the SNOWWEB network during the RAS that was not observed on the ice shelf. These observations were used for a case study validation of the Antarctic Mesoscale PredicOne SNOWWEB deployment covered the northern and western edges of White Island immediately south-east of Ross Island. Observations showed the interaction of a Ross Ice Shelf airstream (RAS) southerly storm event with the complex terrain of the deployment area, including a resulting small but intense gap wind. There was also a substantial dampening effect on the diurnal temperature cycle over the SNOWWEB network during the RAS that was not observed on the ice shelf. These observations were used for a case study validation of the Antarctic Mesoscale Prediction System (AMPS). While AMPS forecast the larger scale winds and temperature well, it did not predict the gap wind or the suppression of the diurnal cycle. A subsequent SNOWWEB deployment to the east of Ross and White islands over the Ross Ice Shelf for a longer duration allowed a more in-depth validation of AMPS conducted using selforganizing maps (SOMs). A combined SNOWWEB/AMPS dataset was created to train a single SOM which then classified each dataset independently, allowing a direct comparison between the classification time-series. AMPS was found to perform well during high wind periods, however problems arose during low wind periods when synoptic forcing was weak. AMPS was able to forecast the periods themselves well, but the actual wind speeds correlated very poorly at the local scale near complex terrain. Model grid length and initialization data were likely contributors given the scale and complexity of the area, though model grid length probably played a role as well. Known problems with cloud modeling and associated effects on the radiation budget would also have had an increased effect. The spatial density of SNOWWEB stations was extremely helpful when validating high resolution output from AMPS. Finally, observations from the CloudSat and CALIPSO satellites were used to quantify cloud incidence over the Ross Ice Shelf and Ross Sea using a series of existing synoptic weather regimes (Coggins regimes) also used during earlier SNOWWEB analyses. Cloud appeared to be sensitive to moisture transport, with higher incidence during summer and autumn when sea ice extent is lower and open ocean closer to the study area. The western Ross Ice Shelf had the lowest cloud incidence, though a persistent cloud signature was found along the Transantarctic Mountains. Weather regimes associated with high surface wind speeds and intense synoptic forcing produced more cloud over the ice shelf with a link to the RAS, however periods of minimal forcing still resulted in substantial amounts of cloud at low altitudes. A link was also found between the RAS and low-level cloud over the Ross Sea during winter, likely a result of interactions with the Ross Ice Shelf Polynya