DEVELOPMENT OF SEASONAL BASED FIRE EARLY SYSTEM FOR MANAGING CARBON EMISSION FROM LAND-FOREST FIRE Rizaldi Boer, M. Ardiansyah, A. Faqih, Achmad Siddik Thoha, Syamsoe Dwi Jatmiko, Anter Parulian Situmorang http://geospasial.bnpb.go.id CCROM SEAP-Bogor Agricultural University E-mail: rizaldiboer@gmail.com HP: +62811117660
INTRODUCTION Forest and Land Fire occur almost every year in Indonesia with different intensity and coverage In El Nino years normally caused large fire. Peatland burning in 1997 El Nino year contributed approximately 13-40 % of annual global emissions (Page et al, 2002) and caused serious damage on ecosystem and economic On average (2000-2005), peat fire contributed to about 20% of the national total emission (Indonesia SNC; MoE, 2010) Seasonal Fire Early Warning Tool is very useful to give more time for preparation of anticipatory measures to reduce the risk of fire EN EN EN Source: Indonesian SNC, 2010 (Based on van der Werf et al. (2007)
Hotspot Density in Open Peatland and Mineral Soils in Kapuas Hotspot density for all CK is much lower than Kapuas
Average seasonal cycle of precipitation, GWL and peat fire occurrences in MRP area in El Nino years and Non-El Nin o years in Central Kalimantan (Putra et al. 2011) Ground Water Level
Hotspot density increase significantly during longer dry season (El Nino years) in Kalteng In Indonesia, long dry season often associated with El Nino events The use of El Nino information and seasonal rainfall prediction would be potential for developing seasonal fire early warning system
Hotspot density increase significantly during longer dry DEFINING FIRE RISK season (El Nino years) in Kalteng In Indonesia, long DS is often associated with El Nino events El Nino information and rainfall is very potential to be used for the development of seasonal fire early warning system
HAZARD FORECASTING DATA Realtime Data and Historical METHODOLOGY OF SEASONAL FRS SST NINO4 Provincial Level: (1) Relationship SST and Hotspot 1 9 months lead time (1) Prediction Hotspot based on real-time SST (2) Prediction Hotspot based on prediction SST Hotspot NASA FIRMS Biophysics Data VULNERABILITY Fire Vulnerability Maps: Provinces Districts Precipitation NASA TRMM District Level: Relationship precipitation and Hotspot 1 3 months lead time Prediction Hotspot density based on precipitation forecast Forecast/ Prediction Precipitation 1 3 months lead time DATA HAZARD FORECASTING Fire Risk Map Provinces 5 km x 5 km FIRE RISK Fire Risk Map Districts 1 km x 1 km
Forecasting Hazards GCM Data Hindcast Data (period 1981-2010) downloaded from the IRI Data Library (http://iridl.ldeo.columbia.edu/sources/.models/.nmme/) as the official host for the data hindcast NMME (Multi Model Ensembles). More detailed information about this can be found at the following web site link: http://www.cpc.ncep.noaa.gov/products/nmme/users_guide.html#five. Forecast Data (updated regularly) downloaded at the web the Climate Prediction Center (CPC), NOAA (link: ftp://ftp.cpc.ncep.noaa.gov/nmme/realtime_anom/.
Observational Data Rainfall data utilised in FRS comes from various sources, namely climate station observation data and rainfall data grid derived from global rainfall data sets. The rainfall observation data used comprises the period of January 1981- September 2014, derived from 80 climate stations across Java, Sumatra, Borneo, Sulawesi and Nusa Tenggara Timur. The rainfall data grid uses Climate Hazards Group InfraRed Precipitation with Station Data (CHIRPS; Funk et al., 2014) version 1.8. Data.
The development of SCPs for supporting Fire Risk System: Rainfall Predictions 4 steps of SCPs development: 1. Produce forecast skills and forecast values 2. Analyse the skill of the GCM forecasts 3. Select predictions with high forecast skill 4. Using Multi-Model Ensemble analysis, use the selected data to develop forecast maps, as defined by users
Scheme for monthly and seasonal rainfall prediction SCPS runs automatically in the system. It will automatically produce forecast outputs every month. No. Table: Seasonal and Monthly Rainfall Prediction Scheme Prediction period Month Season LT 1 LT 2 LT 3 LT 1 LT 2 LT 3 1 January Feb Mar Apr FMA MAM AMJ 2 February Mar Apr May MAM AMJ MJJ 3 March Apr May Jun AMJ MJJ JJA 4 April May Jun Jul MJJ JJA JAS 5 May Jun Jul Aug JJA JAS ASO 6 June Jul Aug Sep JAS ASO SON 7 July Aug Sep Oct ASO SON OND 8 August Sep Oct Nov SON OND NDJ 9 September Oct Nov Dec OND NDJ DJF Jan JFM 10 October Nov Dec NDJ DJF (+1) (+1) Jan Feb JFM FMA 11 November Dec DJF (+1) (+1) (+1) (+1) Jan Feb Mar JFM FMA MAM 12 December (+1) (+1) (+1) (+1) (+1) (+1) Note: LT1 = lead time 1 month, LT 2 = lead time 2 months, and LT 3 = lead time 3 months The final predictions are presented in monthly and seasonal rainfall formats, which are issued one, two and three months in advance. The seasonal predictions can be applied to predict seasonal averages, important for the build-up of fire prone materials, while the monthly predictions provide more insights in when fires are likely to occur most.
Checking predictor variable for each GCM Checking forecast skill for each GCM variable is to find potential variables as predictor for the development of monthly and seasonal rainfall prediction GCM Variable: Total Precipitation GCM Variable: 850 HPa Temperature
Prediksi CH Bulan Juli 2015 Forecast Skill Monthly Rainfall Forecast Issued: Juni 2015
Prediksi CH Bulan August 2015 Forecast Skill Monthly Rainfall Forecast Issued: Juni 2015
Prediksi CH Bulan September 2015 Forecast Skill Monthly Rainfall Forecast Issued: Juni 2015
Hotspot Number) Log (Hotspot Number) Hotspot Prediction from rainfall at district level Rainfall (mm) Rainfall (mm) number of hotspots increased significantly, when the rainfalls below normal
R 2 Hotspot activity Prediction of high number of hotspots (above average) based on information of high rainfall forecasts In the fire-prone areas, the number of hotspots is able to be explained quite well by wet conditions
Hotspot density Distance to road Distance to river Distance to village center Distance form HTI/Oil Palm Licenses Readiness of HTI/HPH/etc Exposure Bio-physic (Sensitivity) Land cover Depth of peatland Land system or landscape Water management Population GDP of Districts/Income Indigenous land boundary Social and economic (Sensitivity) Forest Fire Vulnerability Adaptive capacity Fire brigade (Manggala Agni) Fire care community Extension services/field facilitators Communities /farmer
Maps of fire vulnerability can be updated regularly as land use data and others available and also based on spatial plan Rescaled score of E,S,AC based on its relationship to hotspot Composite Mapping Analysis FV = f (E,S,AC) Fire Vulnerability
Defining Fire Risk Rainfall/Hotspot Prediction Fire Vulnerability Below Normal/High Normal/Medi um Above Normal/Low Very High VH H M_H High H M-H M Medium M-H M M-L Low M M-L L Very Low M-L L VL
Fire Risk Prediction 28 forecast Rainfall forecast (good skill) (1) Y = f(rain) + error Model Hotspot (2) 50 x Fire risk classification VH = very high H = high M-H = medium high M = mediun L-M= low medium L = low VL = very low Vulnerability Map Peluang HS melewati Batas ambang (3) Fire risk Tinggi (>60%) Sedang (40-60)% Hotspot Prediction Rendah (<40%) Tingkat Kerentanan Sangat tinggi ST T S-T Tinggi T S-T S Sedang S-T S S-R Rendah S S-R R Sangat rengdh S-R R SR
Determination process of fire risk level Vulerability map Hotspot prediction Peluang HS melewati Batas ambang Tinggi (>60%) Sedang (40-60)% Rendah (<40%) Tingkat Kerentanan Sangat tinggi ST T S-T Tinggi T S-T S Sedang S-T S S-R Rendah S S-R R Sangat rengdh S-R R SR Fire risk map
Fire Risk System (http://kebakaranhutan.or.id)
Seasonal Fire Early Warning System (FRS) Provide fire early warning 1-3 month lead time to allow for better preparedness up to district level/village level The system provides Monitoring of climate condition (Weather) ENSO development and forecast (ENSO) Rainfall Forecast (Prediction) Fire Risk Information (Fire) Hotspot Probability forecast on fire activity at provincial level at 1-3 month lead time Map fire vulnerability index at district level Seasonal rainfall forecasting and fire risk It offers an opportunity to plan anticipatory responses, mobilize resources early, and to implement policies in an timely manner to encourage use of alternatives to fire in high-risk years
Feature: #1 Fire Vulnerability Map of Central Kalimantan Province
Feature: #2 Fire Vulnerability Map of Kapuas District at Central Kalimantan
Feature: #3 Fire Vulnerability Map of Kapuas District At Central Kalimantan
Check land cover of high vulnerable area
Prediction of Fire Risk Map for Sep 2015
Latitude (a-b) Fire Risk Information by village (1-3 month lead time) Longitude (c-d) Name of Villages a1-b1 c1-d1 Village 1 High Fire Risk Index a2-b2 c2-d2 Village 2, 3 Medium a3-b3 c3-d3 Village 4 Very High a4-b4 c4-d4 Village x, y High
Use of FRS for Fire Risk Management PEAT FIRE PREVENTION AND MITIGATION
FURTHER DEVELOPMENT OF THE FRS Boer et al. 2015 Inclusion of model for estimation of GHG emission based on the hotspot data Efforts that can reduce the hotspot number during the drier season will lead to lower emission due to fire (need further studies how good the HS representing fires Potential use of GOSAT (?) Based on van der Werf et al (2007)
Epilogue FRS can be used to provide early information about the possible danger of forest and land fire Prevention: provides analysis to predict the potential forest and land fire Control: disseminate data and information on nearreal-time to certain stakeholders for fire control Law Enforcement: provides data and fact-based information on violations of forest and land fire to be followed up legally Intervention that leads to lower hotspot number during the abnormal drought years may indicate lowering GHG emissions from peat fire (basis for incentives?)