Currently, no extensive, near real time, global soil moisture observation network exists. Therefore, the Met Office global soil moisture analysis scheme has instead used observations of screen ...temperature and humidity. A number of new space-borne remote sensing systems, operating at microwave frequencies, have been developed that provide a more direct retrieval of surface soil moisture. These systems are attractive since they provide global data coverage and the horizontal resolution is similar to weather forecasting models. Several studies show that measurements of normalised backscatter (surface soil wetness) from the Advanced Scatterometer (ASCAT) on the meteorological operational (MetOp) satellite contain good quality information about surface soil moisture. This study describes methods to convert ASCAT surface soil wetness measurements to volumetric surface soil moisture together with bias correction and quality control. A computationally efficient nudging scheme is used to assimilate the ASCAT volumetric surface soil moisture data into the Met Office global soil moisture analysis. This ASCAT nudging scheme works alongside a soil moisture nudging scheme that uses observations of screen temperature and humidity. Trials, using the Met Office global Unified Model, of the ASCAT nudging scheme show a positive impact on forecasts of screen temperature and humidity for the tropics, North America and Australia. A comparison with in-situ soil moisture measurements from the US also indicates that assimilation of ASCAT surface soil wetness improves the soil moisture analysis. Assimilation of ASCAT surface soil wetness measurements became operational during July 2010.
The McArthur Forest Fire Danger Index used in Australia for operational fire warnings has a component representing fuel availability called the Drought Factor (DF). The DF is partly based on soil ...moisture deficit, calculated as either the Keetch‐Byram Drought Index (KBDI) or Mount's Soil Dryness Index (MSDI). The KBDI and MSDI are simplified water balance models driven by observation based daily rainfall and temperature. In this work, gridded KBDI and MSDI analyses are computed at a horizontal resolution of 5 km and are verified against in‐situ soil moisture observations. Also verified is another simple model called the Antecedent Precipitation Index (API). Soil moisture analyses from the Australian Community Climate and Earth System Simulator (ACCESS) global Numerical Weather Prediction (NWP) system as well as remotely sensed soil wetness retrievals from the Advanced Scatterometer (ASCAT) are also verified. The verification shows that the NWP soil wetness analyses have greater skill and smaller biases than the KBDI, MSDI and API analyses. This is despite the NWP system having a coarse horizontal resolution and not using observed precipitation. The average temporal correlations (root mean square difference) between cosmic ray soil moisture monitoring facility observations and modeled or remotely sensed soil wetness are 0.82 (0.15 ±0.02), 0.66 (0.33 ±0.07), 0.77 (0.20 ±0.03), 0.74 (0.22 ±0.03) and 0.83 (0.18 ±0.04) for NWP, KBDI, MSDI, API and ASCAT. The results from this study suggests that analyses of soil moisture can be greatly improved by using physically based land surface models, remote sensing measurements and data assimilation.
Key Points
Simple water balance models have less skill than weather prediction system soil moisture analyses
Weather prediction system soil moisture analyses are unbiased and capture the seasonal variations
The remotely sensed ASCAT soil wetness product is of good quality
We describe the HadGEM2 family of climate configurations of the Met Office Unified Model, MetUM. The concept of a model "family" comprises a range of specific model configurations incorporating ...different levels of complexity but with a common physical framework. The HadGEM2 family of configurations includes atmosphere and ocean components, with and without a vertical extension to include a well-resolved stratosphere, and an Earth-System (ES) component which includes dynamic vegetation, ocean biology and atmospheric chemistry. The HadGEM2 physical model includes improvements designed to address specific systematic errors encountered in the previous climate configuration, HadGEM1, namely Northern Hemisphere continental temperature biases and tropical sea surface temperature biases and poor variability. Targeting these biases was crucial in order that the ES configuration could represent important biogeochemical climate feedbacks. Detailed descriptions and evaluations of particular HadGEM2 family members are included in a number of other publications, and the discussion here is limited to a summary of the overall performance using a set of model metrics which compare the way in which the various configurations simulate present-day climate and its variability.
This study compared surface soil moisture from 11 separate remote sensing and modelled products across Australia in a common framework. The comparison was based on a correlation analysis between soil ...moisture products and in situ data collated from three separate ground-based networks: OzFlux, OzNet and CosmOz. The correlation analysis was performed using both original data sets and temporal anomalies, and was supported by examination of the time series plots. The interrelationships between the products were also explored using cluster analyses. The products considered in this study include: Soil Moisture Ocean Salinity (SMOS; both Land Parameter Retrieval Model (LPRM) and L-band Microwave Emission of the Biosphere (LMEB) algorithms), Advanced Microwave Scanning Radiometer 2 (AMSR2; both LPRM and Japan Aerospace Exploration Agency (JAXA) algorithms) and Advanced Scatterometer (ASCAT) satellite-based products, and WaterDyn, Australian Water Resource Assessment Landscape (AWRA-L), Antecedent Precipitation Index (API), Keetch-Byram Drought Index (KBDI), Mount's Soil Dryness Index (MSDI) and CABLE/BIOS2 model-based products. The comparison of the satellite and model data sets showed variation in their ability to reflect in situ soil moisture conditions across Australia owing to individual product characteristics. The comparison showed the satellite products yielded similar ranges of correlation coefficients, with the possible exception of AMSR2_JAXA. SMOS (both algorithms) achieved slightly better agreement with in situ measurements than the alternative satellite products overall. Among the models, WaterDyn yielded the highest correlation most consistently across the different locations and climate zones considered. All products displayed a weaker performance in estimating soil moisture anomalies than the original data sets (i.e. the absolute values), showing all products to be more effective in detecting interannual and seasonal soil moisture dynamics rather than individual events. Using cluster analysis we found satellite products generally grouped together, whereas models were more similar to other models. SMOS (based on LMEB algorithm and ascending overpass) and ASCAT (descending overpass) were found to be very similar to each other in terms of their temporal soil moisture dynamics, whereas AMSR2 (based on LPRM algorithm and descending overpass) and AMSR2 (based on JAXA algorithm and ascending overpass) were dissimilar. Of the model products, WaterDyn and CABLE were similar to each other, as were the API/AWRA-L and KBDI/MSDI pairs. The clustering suggests systematic commonalities in error structure and duplication of information may exist between products. This evaluation has highlighted relative strengths, weaknesses, and complementarities between products, so the drawbacks of each may be minimised through a more informed assessment of fitness for purpose by end users.
•Ground-based evaluation of SMOS, AMSR2, ASCAT remotely sensed soil moisture•Modelled soil moisture from WaterDyn, AWRA-L, CABLE, API, KBDI and MSDI•Products displayed varied ability to reflect annual cycles of soil moisture.•Cluster analyses suggest that products may share commonalities in error structure.•Evaluation highlights relative strengths and weaknesses across climate zones.
Modern land surface model simulations capture soil profile water movement through the use of soil hydraulics sub-models, but good hydraulic parameterisations are often lacking, especially in the ...tropics. We present much-improved gridded data sets of hydraulic parameters for surface soil for the critical area of tropical South America, describing soil profile water movement across the region to 30 cm depth. Optimal hydraulic parameter values are given for the Brooks and Corey, Campbell, van Genuchten-Mualem and van Genuchten-Burdine soil hydraulic models, which are widely used hydraulic sub-models in land surface models. This has been possible through interpolating soil measurements from several sources through the SOTERLAC soil and terrain data base and using the most recent pedotransfer functions (PTFs) derived for South American soils. All soil parameter data layers are provided at 15 arcsec resolution and available for download, this being 20x higher resolution than the best comparable parameter maps available to date. Specific examples are given of the use of PTFs and the importance highlighted of using PTFs that have been locally parameterised and that are not just based on soil texture. We discuss current developments in soil hydraulic modelling and how high-resolution parameter maps such as these can improve the simulation of vegetation development and productivity in land surface models.
The Bureau of Meteorology Atmospheric high-resolution
Regional Reanalysis for Australia (BARRA) is the first atmospheric regional
reanalysis over a large region covering Australia, New Zealand, and ...Southeast
Asia. The production of the reanalysis with approximately 12 km horizontal
resolution – BARRA-R – is well underway with completion expected in 2019.
This paper describes the numerical weather forecast model, the data
assimilation methods, the forcing and observational data used to produce
BARRA-R, and analyses results from the 2003–2016 reanalysis. BARRA-R
provides a realistic depiction of the meteorology at and near the surface
over land as diagnosed by temperature, wind speed, surface pressure, and
precipitation. Comparing against the global reanalyses ERA-Interim and MERRA-2,
BARRA-R scores lower root mean square errors when evaluated against
(point-scale) 2 m temperature, 10 m wind speed, and surface pressure
observations. It also shows reduced biases in daily 2 m temperature maximum
and minimum at 5 km resolution and a higher frequency of very heavy
precipitation days at 5 and 25 km resolution when compared to gridded
satellite and gauge analyses. Some issues with BARRA-R are also identified:
biases in 10 m wind, lower precipitation than observed over the tropical
oceans, and higher precipitation over regions with higher elevations in south
Asia and New Zealand. Some of these issues could be improved through
dynamical downscaling of BARRA-R fields using convective-scale (<2 km) models.
Abstract
The McArthur Forest Fire Danger Index used in Australia for operational fire warnings has a component representing fuel availability called the Drought Factor (DF). The DF is partly based on ...soil moisture deficit, calculated as either the Keetch‐Byram Drought Index (KBDI) or Mount's Soil Dryness Index (MSDI). The KBDI and MSDI are simplified water balance models driven by observation based daily rainfall and temperature. In this work, gridded KBDI and MSDI analyses are computed at a horizontal resolution of 5 km and are verified against in‐situ soil moisture observations. Also verified is another simple model called the Antecedent Precipitation Index (API). Soil moisture analyses from the Australian Community Climate and Earth System Simulator (ACCESS) global Numerical Weather Prediction (NWP) system as well as remotely sensed soil wetness retrievals from the Advanced Scatterometer (ASCAT) are also verified. The verification shows that the NWP soil wetness analyses have greater skill and smaller biases than the KBDI, MSDI and API analyses. This is despite the NWP system having a coarse horizontal resolution and not using observed precipitation. The average temporal correlations (root mean square difference) between cosmic ray soil moisture monitoring facility observations and modeled or remotely sensed soil wetness are 0.82 (0.15 ±0.02), 0.66 (0.33 ±0.07), 0.77 (0.20 ±0.03), 0.74 (0.22 ±0.03) and 0.83 (0.18 ±0.04) for NWP, KBDI, MSDI, API and ASCAT. The results from this study suggests that analyses of soil moisture can be greatly improved by using physically based land surface models, remote sensing measurements and data assimilation.
Key Points
Simple water balance models have less skill than weather prediction system soil moisture analyses
Weather prediction system soil moisture analyses are unbiased and capture the seasonal variations
The remotely sensed ASCAT soil wetness product is of good quality
•Utilization of soil moisture from a land surface model for wildfire applications.•Robust performance by the new product against ground observations.•Calibration of the soil moisture product for use ...in operational practices.•Improvements to existing drought indices used in operations.
Soil moisture deficit is a key variable used in operational fire prediction and management applications. In Australia, operational fire management practices use simple, empirical water balances models to estimate soil moisture deficit. The Bureau of Meteorology has recently developed a prototype, high-resolution, land surface modelling based, state-of-the-art soil moisture analyses for Australia. The present study examines this new product for use in operational fire prediction and management practices in Australia. The approach used is twofold. First, the new soil moisture product is evaluated against observations from ground based networks. Among the results, the mean Pearson’s correlation for surface soil moisture across the three in-situ networks is found to be between 0.78 and 0.85. Secondly, the study evaluate a few different calibration methods to facilitate the ready utilization of the new soil moisture product in the current operational fire prediction framework. The calibration approaches investigated here are: minimum-maximum matching, mean-variance matching and, cumulative distribution function matching. Validation of the calibrated products using extended triple collocation technique shows that the minimum-maximum method has the highest skill. Evaluation of the calibrated products against MODIS fire radiative power data highlights that large fires correspond to a drier soil in minimum-maximum outputs compared to other calibration results and the current operational method.
In assessing the severity of age-related macular degeneration (AMD), the Age-Related Eye Disease Study (AREDS) Simplified Severity Scale predicts the risk of progression to late AMD. However, its ...manual use requires the time-consuming participation of expert practitioners. Although several automated deep learning systems have been developed for classifying color fundus photographs (CFP) of individual eyes by AREDS severity score, none to date has used a patient-based scoring system that uses images from both eyes to assign a severity score.
DeepSeeNet, a deep learning model, was developed to classify patients automatically by the AREDS Simplified Severity Scale (score 0-5) using bilateral CFP.
DeepSeeNet was trained on 58 402 and tested on 900 images from the longitudinal follow-up of 4549 participants from AREDS. Gold standard labels were obtained using reading center grades.
DeepSeeNet simulates the human grading process by first detecting individual AMD risk factors (drusen size, pigmentary abnormalities) for each eye and then calculating a patient-based AMD severity score using the AREDS Simplified Severity Scale.
Overall accuracy, specificity, sensitivity, Cohen's kappa, and area under the curve (AUC). The performance of DeepSeeNet was compared with that of retinal specialists.
DeepSeeNet performed better on patient-based classification (accuracy = 0.671; kappa = 0.558) than retinal specialists (accuracy = 0.599; kappa = 0.467) with high AUC in the detection of large drusen (0.94), pigmentary abnormalities (0.93), and late AMD (0.97). DeepSeeNet also outperformed retinal specialists in the detection of large drusen (accuracy 0.742 vs. 0.696; kappa 0.601 vs. 0.517) and pigmentary abnormalities (accuracy 0.890 vs. 0.813; kappa 0.723 vs. 0.535) but showed lower performance in the detection of late AMD (accuracy 0.967 vs. 0.973; kappa 0.663 vs. 0.754).
By simulating the human grading process, DeepSeeNet demonstrated high accuracy with increased transparency in the automated assignment of individual patients to AMD risk categories based on the AREDS Simplified Severity Scale. These results highlight the potential of deep learning to assist and enhance clinical decision-making in patients with AMD, such as early AMD detection and risk prediction for developing late AMD. DeepSeeNet is publicly available on https://github.com/ncbi-nlp/DeepSeeNet.