Download presentation

Presentation is loading. Please wait.

Published byBryan McCormack Modified over 5 years ago

1
DISAT contribution: Development of a methodology for probabilistic assessments of climate change impacts on typical Mediterranean agric. crops (e.g. durum wheat) R. Ferrise, M. Moriondo and M. Bindi ENSEMBLES WORK PACKAGE 6.2 MEETING HELSINKI, 26-27 APRIL 2007

2
Structure of the presentation 1.What are the main objectives of our study? 2.What have we achieved so far? 3.What are we planning to do in the next 6 months ? 4.What are our main questions requiring discussion in this meeting?

3
1. What are the main objectives of our study? –Select/Test impact models to simulate different Mediterranean ecosystem tasks: Forestry - damage due to forest fireForestry - damage due to forest fire Agriculture - losses due to water and heat stressesAgriculture - losses due to water and heat stresses –Apply the selected models to estimate a range of impacts using the probabilistic representation provided by RT1

4
Previous monthsPrevious months –Impact model selection and calibration for typical Mediterranean crops (e.g. olive, grapevine and durum wheat) and forest fire risk assessments –Collection of data required for the calibration and testing of impact models and as reference data for model input Last 6 monthsLast 6 months 2.1Develop a simple statistical model that emulate process- based crop yield models (e.g. SIRIUS-Quality durum wheat model) and can be used in probabilistic climate change assessments 2.2Create yield response surfaces altering the baseline climate 2.3Define critical thresholds of impacts using yield cumulative distribution from the last 30-years (e.g. yield thresholds) 2.4Obtain preliminary estimates of risk probabilities overlapping response surfaces and joint distribution of T and P changes 2. What have we achieved so far?

5
2.1 Statistical model to emulate process-based crop yield models 9 representative sites over the Mediterranean Basin were selected to perform a scenario sensitivity analysis9 representative sites over the Mediterranean Basin were selected to perform a scenario sensitivity analysis Crop yield model SIRIUS simulations were carried out for each scenario with different soils and N-ratesCrop yield model SIRIUS simulations were carried out for each scenario with different soils and N-rates The outputs of the model were used to train a neural network back-propagation modelThe outputs of the model were used to train a neural network back-propagation model The Artificial Neural Network (ANN) was tested using the One-Leave-Out Cross ValidationThe Artificial Neural Network (ANN) was tested using the One-Leave-Out Cross Validation

6
2.1 Statistical mode to emulate process-based crop yield models Sites: 9 grid points (50 Km side), representatives of the Mediterranean Basin climatic variabilitySites: 9 grid points (50 Km side), representatives of the Mediterranean Basin climatic variability Baseline Climate: 30 years (1975-2005) of daily Temperature (min and max), Rainfall and Global Radiation (from MARS JRC archive)Baseline Climate: 30 years (1975-2005) of daily Temperature (min and max), Rainfall and Global Radiation (from MARS JRC archive) Temperature changes: from 0°C to 8°C with 2°C stepTemperature changes: from 0°C to 8°C with 2°C step Precipitation changes: from -40% to +20% with 20% stepPrecipitation changes: from -40% to +20% with 20% step CO 2 Scenarios: from 350 ppm to 650 ppm with 100 ppm stepCO 2 Scenarios: from 350 ppm to 650 ppm with 100 ppm step Scenario Sensitivity Analysis:

7
Soil types used for SIRIUS simulations LowMediumHigh Nitrogen Availability (Kg ha -1 ) 110170270 Nitrogen levels used for SIRIUS simulations SandySandy-loamLoamy Sand %93725 Clay%3.625.417.5 Soil Water Capacity (mm) 53115215 SIRIUS simulations: –For each of the 9 grid cells SIRIUS was run for the combination of the different climatic scenarios with 3 different soils and 3 levels of Nitrogen fertilization – Sowing Date was set using a climatic criterion: at least 5 consecutive days with mean Temperature < 14°C and Rainfall < 2mm, starting from October 1 st and not later than February 14 th –Nitrogen Fertilization was split in three times: 1/4 at sowing, 1/4 at tillering and 2/4 at jointing 2.1 Statistical model to emulate process-based crop yield models

8
Training the Artificial Neural Network:Training the Artificial Neural Network: –A neural network back-propagation model was trained for emulating the SIRIUS outputs: Network layers: 3 Network layers: 3 Input nodes: 5 (variables: CO 2, SWC, N level, T(AMJ), Prec.(AMJ)) Input nodes: 5 (variables: CO 2, SWC, N level, T(AMJ), Prec.(AMJ)) Hidden layer nodes: 20 Hidden layer nodes: 20 Output: 1 Output: 1 ANN model structure Testing results 2.1 Statistical model to emulate process-based crop yield models

9
Leave-One-Out Cross Validation Test:Leave-One-Out Cross Validation Test: Pearsons correlation coefficients between ANN and SIRIUS estimates of crop yields for each of the 9 grid cells with all climate scenarios, 3 soils and 3 N-rates. Grid Cell Identification Number LatitudeLongitudeAltitude Pearsons correlation Coefficient 3009036.6427.31450.90 3405840.129.296840.94 3408039.2922.113050.95 3704341.120.343120.96 3706341.4212.3020.95 4303142.93-7.405900.92 4304643.931.812100.95 4405644.628.053480.93 4506145.0511.23100.96 2.1 Statistical model to emulate process-based crop yield models

10
2.2 Create yield response surfaces SIRIUS and ANN model yield response surfaces were compared for a study area in France (43.6 N, 5.0 E) :SIRIUS and ANN model yield response surfaces were compared for a study area in France (43.6 N, 5.0 E) : –Yield Response Surfaces were estimated altering the 30-years baseline climate (from MARS-JRC archive): Temperature changes: from 0°C to +8°C Temperature changes: from 0°C to +8°C Precipitation changes: from -40% to +20% Precipitation changes: from -40% to +20% CO 2 concentration scenarios: 350 ppm and 550 ppm CO 2 concentration scenarios: 350 ppm and 550 ppm Soil Water Content: 115 mm Soil Water Content: 115 mm Nitrogen Fertilization: 170 Kg N ha-1 Nitrogen Fertilization: 170 Kg N ha-1

11
2.2 Create yield response surfaces The ANN reproduced quite well the SIRIUS crop yields in the different scenariosThe ANN reproduced quite well the SIRIUS crop yields in the different scenarios SIRIUS and ANN estimated response surfaces for a grid box in Southern France, for two CO 2 scenarios Comparison between ANN and SIRIUS estimates of crop yields used to draw the response surfaces

12
2.3 Define critical thresholds Critical threshold of impact was obtained:Critical threshold of impact was obtained: –Calculating the cumulative probability of selected parameter (in this case yield) –Selecting, as threshold, the values that correspond to the 20% of probability Cumulative distribution of yield in a pilot study area 5.35 Mg ha -1

13
2.4 Estimating risk probability The trained ANN was applied to estimate response surfaces in a study area (Tuscany 50x 50 km grid cells)The trained ANN was applied to estimate response surfaces in a study area (Tuscany 50x 50 km grid cells) The statistical software R was adopted to calculate a polynomial regression model based on response surfacesThe statistical software R was adopted to calculate a polynomial regression model based on response surfaces The regression model was applied to calculate yield using data from perturbed physics experiment of Hadley Centre for future scenariosThe regression model was applied to calculate yield using data from perturbed physics experiment of Hadley Centre for future scenarios The perturbed yields were compared with yield threshold to define risk probabilityThe perturbed yields were compared with yield threshold to define risk probability

14
2.4 Estimating risk probability Study Area: TuscanyStudy Area: Tuscany Grid Cell Id. NumberLatitudeLongitudeAltitudeYield Threshold (Mg ha -1 ) 3906142.411.2144.70 4006142.811.11294.86 4006042.810.5164.81 4106243.211.73506.17 4106143.311.23246.08 4106043.310.51075.88 4206243.711.86006.02 4206143.711.22295.72 4206043.710.5785.79 Sites: 9 grid cells (50 Km side) Sites: 9 grid cells (50 Km side) Baseline climate: from MARS JRC Archive Baseline climate: from MARS JRC Archive CO 2 concentration scenario: a1b CO 2 concentration scenario: a1b Soil properties: from the Eusoils database Soil properties: from the Eusoils database Nitrogen level: 170 Kg ha -1 Nitrogen level: 170 Kg ha -1

15
2.4 Estimating risk probabilities Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2000- 2020)Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2000- 2020) Coor.: 43.7N, 11.2E Coor.: 43.7N, 11.2E CO 2 Scenario: a1b CO 2 Scenario: a1b SWC: 115 mm SWC: 115 mm N level: 170 Kg N ha -1 N level: 170 Kg N ha -1

16
Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2020- 2040)Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2020- 2040) Coor.: 43.7N, 11.2E Coor.: 43.7N, 11.2E CO 2 Scenario: a1b CO 2 Scenario: a1b SWC: 115 mm SWC: 115 mm N level: 170 Kg N ha -1 N level: 170 Kg N ha -1 2.4 Estimating risk probabilities

17
Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2040- 2060)Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2040- 2060) Coor.: 43.7N, 11.2E Coor.: 43.7N, 11.2E CO 2 Scenario: a1b CO 2 Scenario: a1b SWC: 115 mm SWC: 115 mm N level: 170 Kg N ha -1 N level: 170 Kg N ha -1 2.4 Estimating risk probabilities

18
Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2060- 2080)Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2060- 2080) Coor.: 43.7N, 11.2E Coor.: 43.7N, 11.2E CO 2 Scenario: a1b CO 2 Scenario: a1b SWC: 115 mm SWC: 115 mm N level: 170 Kg N ha -1 N level: 170 Kg N ha -1 2.4 Estimating risk probabilities

19
Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2080- 2100)Risk probability overlapping response surfaces and joint distribution of T an P changes for a grid box in Tuscany (2080- 2100) Coor.: 43.7N, 11.2E Coor.: 43.7N, 11.2E CO 2 Scenario: a1b CO 2 Scenario: a1b SWC: 115 mm SWC: 115 mm N level: 170 Kg N ha -1 N level: 170 Kg N ha -1 2.4 Estimating risk probabilities

20
Change in risk probability in the 9 grid cells in the next decadesChange in risk probability in the 9 grid cells in the next decades 2.4 Estimating risk probabilities

21
– Next 6 months (months 33-38) Move on other agric. Crops (grapevine and olive) and forest fire risks to:Move on other agric. Crops (grapevine and olive) and forest fire risks to: –Development simple statistical models that emulates process- based crop yield models –Create preliminary yield response surfaces altering the baseline climate –Define critical thresholds of impacts using yield cumulative distribution from the last 30-years –Obtain preliminary estimates of risk probabilities overlapping response surfaces and joint distribution of T an P changes 3. What are we planning to do in the next 6 months ?

22
4. What are our main questions requiring discussion in this meeting? To get information from: –Chris about overall progress in ENSEMBLES (i.e. new developments, ongoing activities and future plans) –Clare (from RT2B) about the latest status of RT2B on the provision of climate model outputs or their derivatives for use in impact assessment (i.e. by WP 6.2) To discuss with: –Chris, Clare and Glen about: the various methods of applying probabilistic climate projections in impact studies, the format and delivery of climate information for use in impact assessments in ENSEMBLES

Similar presentations

© 2019 SlidePlayer.com Inc.

All rights reserved.

To make this website work, we log user data and share it with processors. To use this website, you must agree to our Privacy Policy, including cookie policy.

Ads by Google