PUBLICATIONS
Published works
Mitigating the effects of severe fire, floods and heatwaves through the improvements of land dryness measures and forecasts - final project report
Title | Mitigating the effects of severe fire, floods and heatwaves through the improvements of land dryness measures and forecasts - final project report |
Publication Type | Report |
Year of Publication | 2021 |
Authors | Kumar, V, Dharssi, I, Fox-Hughes, P |
Document Number | 646 |
Date Published | 02/2021 |
Institution | Bushfire and Natural Hazards CRC |
City | Melbourne |
Report Number | 646 |
Keywords | Floods, forecasts, heatwaves, land dryness, measures, mitigation, severe fire |
Abstract | This Bushfire and Natural Hazards CRC project, titled Mitigating the effects of severe fires, floods and heatwaves through the improvements of land dryness measures and forecasts, was a partnership with the Bureau of Meteorology, and examined the use of detailed land surface models, satellite measurements and ground-based observations for the monitoring and prediction of landscape dryness. This project addresses a fundamental limitation in our ability to prepare for fires, floods and heatwaves and is directly linked to pre-event planning as well as forecasting of events. The research conducted in the present project solely focuses on the application of soil and land dryness/moisture in the context of fire danger and fire management practices. The lack of focus on flood and heatwave is circumstantial. The research priorities were set and driven by the requirements of the project end-users, all of them from various fire management agencies across Australia. Hence, the end-use interest was solely on the application of the research in fire management. Nevertheless, it is worth pointing that there is a substantial amount of research literature which establishes the importance of soil moisture in flood and heatwave prediction and applications. Currently, landscape dryness for fire management is estimated in Australia using simple empirical models developed in the 1960s. The most prominent of those used in Australia are the Keetch-Byram Drought Index (KBDI) and the Soil Dryness Index (SDI). An initial study performed as part of this project suggested that analyses of soil moisture can be improved by using physics-based land surface models, remote sensing measurements and data assimilation. JASMIN prototype To address this, the present project developed a standalone prototype land surface modelling system, called Joint UK Land Environment Simulator based Australian Soil Moisture Information (JASMIN) to produce daily soil moisture analyses at 5km resolution and 4 soil layers. Verification against ground-based soil moisture observations shows that this prototype system is significantly more skilful than both KBDI and SDI. Though JASMIN can supplement many applications that require accurate soil moisture estimates, the biggest beneficiary of this new system will be the fire agencies. The soil moisture estimate from the new system provides a robust alternative to the methods currently used in fire prediction. This is evident from the verifications performed against in situ measurements. KBDI and SDI show large errors over regions where they are used operationally. KBDI, for example, has a large wet bias over southern regions that could undermine fire danger ratings. The JASMIN system can produce reliable soil moisture information over a wide range of land-use types, which potentially extends its applicability to other fields as well. Also, JASMIN is shown to have good skill for both surface and deep soil horizons. JASMIN calibration To promote an effective adoption of JASMIN in current operational practices, calibration methods were applied to the native JASMIN soil moisture datasets. The key aim of these methods was to calibrate JASMIN outputs in units of moisture excess to moisture deficit values that range from 0–200, as required by McArthur's Forest Fire Danger Index (FFDI; McArthur, 1967). The calibration offers a simple, faster and cost-effective way to make significant upgrades to the existing operational systems used by fire and other environmental agencies. The calibration methods applied were minimum-maximum matching, mean-variance matching, and cumulative distribution function matching. The selection of these calibration methods was based on the potential end-user requirement, whether that is to simply replace the legacy systems with a new product with high skill (e.g., minimum-maximum method), or to replace the existing system that captures the temporal variations better while preserving the climatology of the older system (e.g., mean-variance and cumulative distribution function matching). The latter could be useful if existing operating systems are already tuned to offset the bias in the current soil moisture deficit methods. Improving high spatial resolution mapping This project also aimed to improve applications such as fire danger mapping that may require soil moisture information at higher spatial resolution due to the large spatial variability of soil moisture in the landscape. A common practice to overcome such a problem is to employ downscaling methods to increase the spatial scale of the product. Recent advances in optical remote sensing have allowed researchers to use different remote sensing products that reflect soil moisture variability as ancillary information. A method based on a “universal triangle” concept is used in several previous studies, which establishes a relationship between soil moisture, vegetation index, and surface radiant temperature from optical remote sensing. This project applied three downscaling methodologies: two based on regression and one based on a physics-based approach. Results from the downscaling methodologies indicate that it is feasible to improve the spatial resolution of JASMIN using all three disaggregating algorithms and preserve the general large-scale spatial structure seen in JASMIN soil moisture estimates. However, the seasonal means obtained at 1 km show that each product displays characteristic soil moisture spatial variability at fine scales. Results from the comparison with ground-based soil moisture measurements indicate that there is no significant degradation of the bias in the three methods when moving to higher spatial resolution. Predicting live fuel moisture content Prediction of the moisture status in live fuels is an important gap in current fire management practices which, if filled, can potentially be useful for spatial and temporal assessment of landscape dryness. The final objective of the project was thus to explore the relationship between soil moisture and live fuel moisture content (LFMC) using the datasets from JASMIN and Australian Flammability Monitoring System (AFMS), respectively. The analysis carried out indicates that soil moisture is a leading indicator of LFMC. This project developed a simple yet skilful model to predict live fuel moisture content for the whole of Australia. The key variable is the 0-350 mm layer soil moisture derived from the JASMIN system. The modelling strategy pursued consists of a linear combination of two sub-models: one to capture the annual cycle and one to capture the daily variations. A time function represents the LFMC annual cycle model. The daily deviations in LFMC are captured by using a linear regression model with 14-day lagged daily deviations in soil moisture as the input. The daily changes in soil moisture are computed by deviations from its annual cycle. When evaluated over 60 sites, the approach returned an average R2 of 0.64 with normalised root mean square error values of <25% at all sites. As researchers were employing a gridded soil moisture product, this strategy facilitates the reconstruction of past events, as well as data gap filling. The lag of 14 days implies a lead time of 14 days for predicting the LFMC. This has significant operational implications, as daily variations in LFMC can be predicted using soil moisture information from JASMIN on a national scale. JASMIN is currently run as a prototype research system, with soil moisture analysis done only near-real-time. However, JASMIN can be extended to produce both real-time analysis and forecasts. The prognostic mode can provide soil moisture forecasts for up to 10 days. This means a maximum lead time of 24 days can be achieved by utilising soil moisture forecasts. JASMIN utilisation A key focus of the project from its inception was to create pathways for easier utilisation of the project deliverables. This is reflected in both the scientific and technical approaches adopted in this project. For example, the calibration of JASMIN to KBDI and SDI was done to facilitate the ready utilisation of JASMIN in the existing operational system. A total of 8 calibrated JASMIN soil dryness products were developed and made available through the Bureau of Meteorology's THREDDS server. The JASMIN soil moisture in volumetric units at 4 layers are also provided via the THREDDS server for interested parties to evaluate. The volumetric soil moisture fields from the top two JASMIN layers (0-100 mm and 100-350 mm) are available via AFMS as well. The datasets on both THREDDS and AFMS are updated near-real-time. There is a continuing interest in the end-user community in utilising JASMIN for various fire management applications. In that respect, JASMIN has been assessed in the Western Australian Department of Biodiversity, Conservation and Attractions study on tall wet forest fuel availability. Tasmania Parks and Wildlife has also been using JASMIN as a decision-support tool to restrict the use of open fires in national parks. Also, JASMIN data were updated specifically to assist with Tasmanian decision-making for 2018-19 seasonal bushfire assessment workshop and preseason consultative committee on fire weather. The JASMIN system can produce reliable soil moisture estimates over a wide range of land-use types and can support many applications that require accurate soil moisture information. However, there is still scope for improvements to the JASMIN system, whether it be the skill or the scale. An immediate focus could be the use of data assimilation techniques to improve the skill of JASMIN. Data assimilation allows uncertainties in land surface model soil moisture to be offset to some extent by routinely updating the hydrological conditions using the information provided by observations on state variables used by land surface models. The assimilation of satellite observations is shown to improve the model soil moisture state. In that respect, the use of NASA's land information system (LIS) is being evaluated at the Bureau. The LIS is a complex framework that uses extensible interfaces to allow the incorporation of new domains, land surface models, land surface parameters, meteorological inputs, data assimilation and optimisation algorithms. The extensible nature of these interfaces, and the component style specifications of the system, allow rapid prototyping and development of new applications. The JASMIN system can be incorporated within LIS to facilitate the assimilation of various observation types. Further, it can be leveraged to run JASMIN with an enhanced spatial resolution, desirably at 1 km. |
Refereed Designation | Refereed |